PATH REDUCTION FACTOR FOR MICROWAVE TERRESTRIAL LINKS DERIVED FROM THE MALAYSIAN METEOROLOGICAL RADAR DATA Nor Hisham bin Haji Khamis Supervisors Associate Professor Dr Jafri bin Din Professor Dr Tharek bin Abdul Rahman Faculty of Electrical Engineering Universiti Teknologi Malaysia BAHAGIAN A - Pengesahan Kerjasama* Adalah disahkan bahawa projek penyelidikan tesis ini telah dilaksanakan melalui kerjasama antara dengan Disahkan oleh: Tandatangan : Tarikh : : Jawatan : (Cop rasmi) * Jika penyediaan tesis/projek melibatkan kerjasama. BAHAGIAN B - Untuk Kegunaan Pejabat Sekolah Pengajian Siswazah Tesis ini telah diperiksa dan diakui oleh: Nama dan Alamat Pemeriksa Luar : Prof. Dr. Syed Idris bin Syed Hassan School Of Electrical & Electronic Eng., Universiti Sains Malaysia, 14300 Nibong Tebal, Penang Nama dan Alamat Pemeriksa Dalam I : Prof. Madya Dr. Norazan bin Mohd Kasim Fakulti Kejuruteraan Elektrik UTM, Skudai Pemeriksa Dalam II : Prof. Dr. Abu Bakar Mohamad Fakulti Kejuruteraan Elektrik UTM, Skudai Nama Penyelia lain : (j ika ada) Disahkan oleh Penolong Pendaftar di Sekolah Pengajian Siswazah: Tandatangan : Nama : Tarikh : GANESAN A/L ANDIMUTHU Nama PATH REDUCTION FACTOR FOR MICROWAVE TERRESTRIAL LINKS DERIVED FROM THE MALAYSIAN METEOROLOGICAL RADAR DATA NOR HISHAM BIN HAJI KHAMIS A thesis submitted in fulfilment of the requirements for the award of the degree of Doctor of Philosophy Faculty of Electrical Engineering Universiti Teknologi Malaysia JUNE 2005 DEDICATIONS To Almighty ALLAH, Most Gracious, Most Merciful To my parents Haji Khamis bin Haji Othman Hajjah Esah binti Haji Md Sab To my dearest wife Hajjah Aisyah binti Che Mat To my children Muhammad Khalid Nurul Iman Luqman Hamdani Izzatul Huda Nasrul Azizi Akmal Husaini To all my teachers To my brothers and sisters For all the joy, love, understanding, and sacrifices. Religion without science is lame, Science without religion is blind ACKNOWLEDGMENTS I would like to express my deepest gratitude and appreciation to both my supervisors, Assoc. Professor Dr. Jafri bin Din and Professor Dr. Tharek Abd. Rahman. For their suggestions, assistance, and support during the entire project. No doubt, without their constant help and encouragement, this project would not be completed. A special thanks to Mr Tan Boon Eng of Perkhidmatan Kajicuaca Malaysia for his assistance in obtaining the radar data and during my practical training at Jabatan Kajicuaca Malaysia, Petaling Jaya. My deepest thanks to friends and colleagues, who helped me directly or indirectly. For their continuing encouragement and support. Finally, an acknowledgement to UTM, for providing me with the opportunity and facilities to do and complete this project. ABSTRACT Attenuation due to rain is a major concern in transmission of microwave signals. The effect of rain attenuation is more pronounce when signals are being transmitted at higher frequencies. For tropical countries like Malaysia, rain occurs almost yearound and in most instances, much heavier than temperate region. Even rain itself, does not distribute evenly in a region experiencing precipitation. This gives rise to the need of a correction or reduction factor when calculating attenuation due to rain. This topic has been the focus of many researches. However, many of these researches were done in temperate regions, making it necessary for a study using local data. This study utilized the local weather radar data obtained from the Meteorological Department of Malaysia, and data from two rain gauge networks installed in UTM, Skudai campus. From the analysis of the radar data, a reduction factor is deduced. The reduction factor obtained in this study follows the same pattern as other models but has a lower value as the path link increases. This shows that attenuation due to rain is lower than as predicted using other models. Rain rate distribution and rain cell size distribution is also formulated from radar data. R0.01 of 120.907 mm/hr agrees very well with the ITU-R and the Meteorological Department of Malaysia values. Using data from the rain gauge networks, the profile and the size of rain cells at different rain rates are determined. This study finds that most rain cells in Malaysia are highly convective with an average cell size of 1.2 to 1.5 km. All the information are important for attenuation predictions, link budget estimation, microwave system planning, slant path rain attenuation modeling and remote sensing of the earth’s surface, and have important applications in attenuation mitigation techniques such as space diversity. Further study can be done with more precise, elaborate, and sophisticated measuring systems such as Doppler or polarimetric radar, complemented with microwave links and rain gauge networks. ABSTRAK Rosotan yang disebabkan oleh hujan adalah satu masalah yang besar di dalam penghantaran isyarat gelombangmikro. Kesan rosotan hujan adalah lebih ketara pada frekuensi yang lebih tinggi. Untuk negara-negara tropikal seperti Malaysia, hujan berlaku hampir sepanjang tahun dan adalah lebih lebat dari kawasan berhawa sederhana. Taburan hujan juga adalah tidak sekata di dalam kawasan hujan. Keadaan ini menjadikan keperluan faktor pengurangan dalam pengiraan rosotan hujan. Topik ini telah menjadi fokus untuk banyak penyelidikan. Walaubagaimanapun, kebanyakan penyelidikan adalah dalam kawasan berhawa sederhana, menjadikan satu kajian yang menggunakan data tempatan satu kepentingan. Kajian ini menggunakan data radar cuaca yang diperolehi daripada Jabatan Kajicuaca Malaysia dan rangkaian tolok hujan yang dipasang di UTM, Skudai. Penganalisaan data radar telah menghasilkan faktor pengurangan, taburan lebat hujan dan taburan saiz sel hujan. Faktor pengurangan daripada kajian ini mempunyai bentuk yang serupa dengan lainlain model tetapi mempunyai nilai yang lebih rendah apabila jarak bertambah. Ini menunjukkan bahawa ramalan rosotan daripada model lain adalah lebih tinggi dari nilai sebenar. Nilai R0.01 sebanyak 120.907 mm/jam adalah setara dengan ITU-R dan Jabatan Kajicuaca Malaysia. Daripada data tolok hujan, profil dan saiz sel hujan pada kelebatan hujan yang berlainan telah ditentukan. Kajian juga menunjukkan bahawa hujan di Malaysia adalah sangat konvektif dengan purata saiz sel hujan antara 1.2 hingga 1.5 km. Hasil kajian adalah sangat penting untuk jangkaan pelemahan, anggaran bajet rangkaian, dan perangkaan sistem gelombangmikro. Hasil kajian juga adalah berguna untuk teknik pengurangan seperti kepelbagaian tapak. Kajian selanjutnya boleh dilakukan dengan menggunakan teknik pengukuran yang lebih jitu dan canggih seperti penggunaan radar Doppler atau polarimetrik, dilengkapi dengan laluan jalurmikro dan rangkain tolok hujan. TABLE OF CONTENTS CHAPTER TITLE PAGE TITLE PAGE i DECLARATION ii DEDICATIONS iii ACKNOWLEDGEMENTS iv ABSTRACT v ABSTRAK vi TABLE OF CONTENTS vii LIST OF FIGURES xi LIST OF TABLES xiii LIST OF SYMBOLS xv LIST OF ABBREVIATIONS xvii LIST OF APPENDICES xviii INTRODUCTION 1 1.1 Overview 1 1.2 Problem Statement 2 1.3 Objectives 4 1.4 Scope of Study 4 1 viii 1.5 Outline of Thesis 5 Methodology 7 2.1 Introduction 7 2.2 Radar Data 7 2.3 Disadvantages of Radar Measurement 9 2.4 Radar Measurement Principle 10 2.5 Z-R Relationship 10 2.6 Selection of radar Data 12 2.7 Rain Gauge Data 15 2.8 Attenuation Measurements Using Microwave 16 2 Links 3 RAIN ATTENUATION, RAIN MODELS, 17 AND REDUCTION FACTORS 3.1 Background 17 3.2 Attenuation by Rain 18 3.3 Rainfall Rate, Rain Cell Size, and Rain Height 22 3.4 Rainfall Rate for 0.01 % of the time or R0.01 26 3.5 Rain Prediction Models 27 3.6 3.5.1 Crane’s Global Model 28 3.5.2 ITU-R Model 30 Reduction Factor 33 3.6.1 Lin Model 33 3.6.2 Moupfouma Model 34 3.6.3 CETUC Model 35 3.6.4 Improved CETUC Model 35 ix 3.6.5 Goddard and Thurai Model 36 3.6.6 ITU-R Model 37 3.6.7 Singapore Model 38 3.6.8 DAH Model 40 3.6.9 Comments on reduction Factor Models 40 3.7 Rain in Malaysia 41 3.8 Determination of Rain Cell Size 42 RAIN GAUGE AND RADAR DATA 44 4.1 Introduction 44 4.2 Rain Gauge and Rain Gauge Networks 45 4 4.2.1 Casella Rain Gauge 4.3 Radar Data Collection 4.3.1 Kluang Radar Data Format 48 50 58 DATA ANALYSIS AND RESULTS 61 5.0 Introduction 61 5.1 Rain Gauge Data Analysis 61 5.1.1 Preliminary Data Analysis 62 5.1.2 Selection of Rain Gauge Network 62 5 (RGN-UTM 1) Data 5.2 5.1.3 RGN-UTM 1 Data Analysis 62 5.1.4 Cell Size and Intensity Distribution 72 5.1.5 Rain Distribution Inside a Rain Cell 73 Radar Data Analysis 79 5.2.1 Rejection of Permanent Echo 80 5.2.2 Preliminary Results 80 x 5.2.3 Distribution of Rain Rate from Radar Data 82 5.2.4 Determination of Rain Cell Size from Radar 85 Data 5.2.5 Rain Attenuation Measurements in UTM 5.3 Deducing the Reduction Factor from Radar 88 89 Data CONCLUSION AND FUTURE STUDIES 110 6.1 Conclusion 110 6.2 Future Studies 114 REFERENCES 116 APPENDICES 128 6 LIST OF FIGURES TITLE FIGURE PAGE 2.1 Examples of virtual links 13 3.1 Rain formation through cold and warm air fronts 23 3.2 Rain cell diameter versus rainfall rate 23 3.3 Height of melting layer 25 3.4 Rain height 26 3.5 The global rain rate regions 28 3.6 Rain rate distribution curves for various regions 29 3.7 Rainfall climatic zones 30 3.8 Rainfall rate contours for 0.01% of the time for Asia 31 3.9 Revised rainfall rate contours for 0.01% of the time 32 4.1 RGN-UTM 1 Rain gauge network stations 46 4.2 RGN-UTM 2 Rain gauge network stations 47 4.3 Casella Tipping Bucket Rain Gauge with Integral 49 Logger 4.4 Tipping buckets of a Casella Rain Gauge 50 4.5 Merged PPI scan 51 4.6 A Kluang radar station PPI scan 53 4.7 A Kluang radar station RHI scan 54 4.8 A MATLAB radar plot 55 4.9 Top view of typical composite image 56 4.10 A cross-section of a typical composite image 56 xii 5.1 Graph of Rainfall Rate Recordings at all Rain Gauge 63 Stations 5.2 Rain Rate Distribution for 120 mm/hr at Civil Station 66 5.3 Rain Rate Distribution for 90 mm/hr at Civil Station 67 5.4 Rain Rate Distribution for 60 mm/hr at Civil Station 68 5.5 Rain Rate Distribution for 30 mm/hr at Civil Station 68 5.6 Averaged Rain Rate Distributions for RGN-UTM 1 69 54.7 Averaged Rain Rate Distribution for all stations 72 5.8 Rain rate distributions for RGN-UTM 2 74 5.9 Rain distributions assuming that TV Station is the 74 center of the rain cell 5.10 Rain distribution assuming the IVAT Station to be the 75 center of the rain cell 5.11 Averaged rain cell size 77 5.12 RGN-UTM 2 Rainfall rate distribution 78 5.13 Rain rate distribution 82 5.14 The plots of original data and the curve-fit line 84 5.15 Rain Cell Size Distribution 87 5.16 The best-fit curve for 1-km path links attenuation 92 distribution 5.17 Attenuation for 0.01% of the time at 7, 10, and 15 94 GHz 5.18 Reduction factor (r) plots for 1 to 10-km path lengths 95 at frequencies of 7, 10, and 15 GHz for 0.01% of the time 5.19 Best fit line for coefficient a 97 5.20 Best fit line for coefficient b 97 5.21 Plots of all the reduction factor models at 7 GHz 100 5.22 Plots of all the reduction factor models at 10 GHz 101 5.23 Plots of all the reduction factor models at 15 GHz 102 6.1 Proposed reduction factor r 111 6.2 Predicted attenuation due to rain for 0.01 % of the 111 time LIST OF TABLES TABLE 3.1 TITLE Point Rain Rate (RP) Distribution Values (mm/h) PAGE 29 Versus Percent of Year Rain Rate is exceeded 3.2 Rainfall rates in the climatic zones 31 4.1 dBZ-R values for Kluang radar 52 4.2 Radar Scan Elevation Angles 57 4.3 Number of matrices in each data folders 59 4.4 The corresponding actual rain rate values in mm/hr to 60 the rain rate level 5.1 Rain events recorded at all rain gauges locations 62 5.2 Correlations for no. of readings at all stations 63 5.3 Equal Rainfall Rate At All Three Stations 65 5.4 The Rainfall Rates and their durations for all Rain 70 Gauge Stations RGN-UTM 1 5.5 The Rainfall Rate Percentages and the Averaged 71 Rainfall Rate Percentage for all Rain Gauge Stations 5.6 Simultaneous rain rate readings at all stations when 76 IVAT and TV Studio stations register 120 mm/hr rain rate 5.7 Distribution of rainfall rate of RGN-UTM 2 78 5.8 R0.01 Values for RGN-UTM 2 Stations 79 5.9 Rainfall Intensity for averaged 1-km path 81 xiv 5.10 Rain rate distribution for range-bin size of 1-km from 83 radar data 5.11 Rain Cell Size Distributions 86 5.12 Specifications for Binariang system links 88 5.13 Specifications for Digi system links 89 5.14 Distribution of attenuation for 1-km links operating at 7 91 GHz 5.15 Attenuation (dB) for 0.01% of the time; at 7, 10, and 15 93 GHz for path lengths of 1 to 10-km 5.16 Reduction factor (r) values for 1 to 10-km path lengths 95 at frequencies of 7, 10, and 15 GHz for 0.01% of the time 5.17 a & b values for best fit lines of reduction factors at 7, 96 10, and 15 GHz 5.18 a and b values at 7, 10, and 15 GHz 98 5.19 The reduction factor (r) for the proposed Malaysia 99 model 5.20 Comparison of r from various models at 7 GHz 100 5.21 Comparison of r from various models at 10 GHz 101 5.22 Comparison of r from various models at 15 GHz 102 5.23 Comparison of predicted attenuations (dB), with 105 measurements LIST OF SYMBOLS α regression exponential for specific attenuation β exponential in Moupfouma’s model. γs specific attenuation due to rain (dB/km) A attenuation of radiowave propagating through free-space (dB) a excess attenuation due to water vapor Ar attenuation of radiowave due to rain (dB) b excess attenuation due to mist and fog; regression exponential for radar reflectivity c excess attenuation due to oxygen D diameter of rain cell (km) d absorption losses due to other gasses dB decibels e excess attenuation due to rainfall dBZ radar reflectivity factor in dB F Frequency (GHz) k regression coefficient for specific attenuation L path length (km) m exponential in Moupfouma’s model. n(r) number of raindrops per unit volume per radius interval (m-3mm-1) p time percentage, percentage Qt(r) total extinction cross-section (cm2) r radius (cm), reduction factor R rainfall rate (mm/hr) R0.01 rainfall rate (mm/hr) for 0.01 % of the time xvi Rt threshold values of a specific rain rate Z radar reflectivity factor (dBZ or mm6/m3) xvii LIST OF ABBREVIATIONS 3D-RAPIC 3 dimensional software to process and handle weather radar data CETUC Center for Telecommunication Studies of the Catholic University of Rio de Janeiro CMOS Complementary metal-oxide semiconductor DAH Dissanayake, Allnutt, Haidara DSD, dsd drop size distribution EPROM Erasable programmable read-only memory ITU International Telecommunication Union ITU-R International Telecommunication Union-Radio communication Section LOS Line-of-sight PPI Pulse-Position Indicator PPI Pulse-position-indicator r Reduction factor RGN-UTM 1 Rain Gauge Network in UTM no. 1 RGN-UTM 2 Rain Gauge Network in UTM no. 2 RHI Rain Height Indicator RHI Rain height indicator STC Sensitivity Time Control UTC Universal Time Constant UTM Universiti Teknologi Malaysia xviii LIST OF APPENDICES APPENDIX TITLE PAGE 1.1 Earth’s Climate and Raindrops 128 3.1 Specifications of the Casella Rain Gauge 137 3.2 Data Sample from RGN-UTM 1 138 3.3 Radar Measurement Theory 139 3.4 Radar Calibration Checklist 144 3.5 Kluang Radar Station Specifications 148 3.6 An Example of a Radar Data 149 3.7 Radar Data Encoding Format 153 4.1 MATLAB programs 157 4.2 Radar Rain Rate Distribution 162 4.3 173 4.4 Rain Cell Diameter from Radar Data 1 to 10-km path attenuations at 7, 10, and 15 GHz 174 4.5 Rec. ITU-R P.838-1 222 CHAPTER 1 INTRODUCTION 1.1 Overview Radio link is very important in communication systems. Large number of parameters must be considered and optimized to build an efficient radio link. That is, radio signals must suffer minimum degradation possible. For reliable communication link, selection of operating frequency and possible atmospheric attenuation must be studied and understood. Since the governing body usually determines the operating frequency, attenuation, especially due to rain, has become the subject of numerous studies and researches. For good engineering and economic practice, it is always desirable to reduce cost of a system and avoid interference to other radio systems. Some factors which introduce attenuation to radio propagation are gaseous absorption, absorption and scattering due to clouds, fogs, precipitation, atmospheric turbulence and ionospheric effects. Since Malaysia is in the tropical region, studies of attenuation due to hydrometers are very important. Rain, hail, ice, cloud and snow are all types of hydrometers but raindrops cause the most attenuation by absorption and scattering of radio waves. Eventhough the International Telecommunication Union (ITU) recommends a technique to estimate rain attenuation, studies (Ajayi et al, 1988; Juy et al, 1990; Yagasena et al, 1995) have shown that results using this techniques does not agree with actual measurements done locally. This is 2 understandable as the recommendation is meant to be applicable to wide area as possible and most studies were done using temperate region data. Thus, local studies are needed and based on these studies; a model for rain attenuation in Malaysia can be developed. 1.2 Problem Statement Power budget and fade margin are important factors to be considered in designing microwave transmission systems. The world of telecommunication is very competitive such that when providing for a system, careful infrastructure planning is needed to avoid unnecessary costs. Microwave links are designed to meet specific reliability factor. Reliability, or sometimes is known as availability of a system, is usually expressed as a percentage. It represents the percentage of the time the link is expected to operate without an outage caused by propagation conditions. It has been widely accepted that a good communication system must provide at least 99.99% reliability (IEEE, 2004; ITU-R SA 1414). In other words, the system can only be down for 0.01% of the time, which is usually referenced to a year. This means that the system can be unavailable for no less than 52.6 minutes per year [365.25*24*60*0.01/100]. (For emergencies, call 999; for no emergency, it is 99.99%!) A very important factor that affects path reliability is rain attenuation. It also contributes in power budget and fade margin considerations. Thus, it is very important to properly quantify rain attenuation. Due to the nature of rain events, a reduction factor is needed in order to calculate rain attenuation. A rain event occurring in an area is not constant. Rain does not distribute evenly in a region experiencing precipitation. Even though specific attenuation due 3 to rain for a specific distance or per kilometer can be formulated; there arises a need to find a reduction factor to account for the non-uniformity of rain for larger distances. This is especially crucial in tropical regions as rain has been found to be more convective in nature rather than widespread. Tropical region also suffers heavier rainfall rates as compared to temperate regions. To formulate the reduction factor, the experimental procedure would requires several links with different path lengths to be set up in close proximity. However, this would be very difficult to be constructed. An alternative approach is to use radar data to obtain attenuation statistics for simulated links of various lengths. In addition, even after a proposed microwave link has been evaluated with regard to reliability, the calculations may show that it will not meet the required standards. Or a designer may want to improve the reliability of the telecommunication system. In this situation, mitigation techniques such as diversity may be employed. One such technique is space diversity where an additional receiver may be constructed (Nor Hisham Khamis et al, 2000). By switching and/or combining the signals received by the two receivers, the reliability of the communication link is greatly increased. An important parameter to consider is site separation or the distance between the two receivers. When spacing is adequate between the two receivers, there should be little correlation between the two paths. Site separation or distance is used to determine the diversity improvement factor and diversity gain when employing diversity (ITU-R P.618-5, 1997). Knowing rain cell size distribution will help to determine site separation. Thus, the aim of this thesis is to find the reduction factor to be used in rain attenuation calculation using radar data, and also to determine the rain cell size distribution needed for site diversity application using rain gauge networks. 4 1.3 Objectives The objectives ensure that the aim of this study is achieved. These objectives are contributions accomplished during this study. The objective of this study are as follows. To formulate a suitable reduction factor (r), to be used in the calculation of attenuation due to rain in the local Malaysian environment from the local weather radar data obtained from the Meteorological Department of Malaysia. To estimate the attenuation due to rain using the formulated r and other models proposed by other researchers, and compare the results with measured values. To find the profile of rain rate distribution inside a rain cell and to determine the rain cell size distribution of local rain. 1.4 Scope of Study The scope of study indicates the basic guidelines and techniques that this study examined in achieving the objectives. It also ensures that the work done stays within the intended study. There are two types of data that are utilized in this study, radar data and rain gauge network data. 5 Radar data gives the averaged rainfall rate for a range-bin size of 1-km each. This rainfall data is used to calculate the rain attenuation for ‘virtual’ microwave links of 1 to 10-km path lengths. Attenuation due to rain is then calculated using the rainfall rate from the radar data. Reduction factor is then deduced from the rain attenuation calculation. The radar data is obtained from the Kluang Radar Station of the Meteorological Department of Malaysia. Rain cell size distribution is also obtained from radar data. However, rain cell size distribution from radar data is limited to 1-km integration size. This is due to the fact that the radar uses a range bin size of 1 km. To determine the profile of rain rate distribution inside a cell, rain gauge networks were utilized. These rain gauge networks consisted of several rain gauges that were fixed in almost a single line. Rain gauge gives point rainfall rate values. Rain intensity profile inside a cell, and rain cell size are obtained from rain gauge networks. Two rain gauge networks have been set up in UTM, Skudai campus. 1.5 Outline of Thesis The outline of this thesis indicates the organization of this thesis. This thesis is separated into 6 chapters. Chapter One gives the introduction to this study, the problem statement, objectives, and the scope of this study. 6 Chapter Two focuses on the methodology that was used in this study. It discusses the feasibility of a study using radar data, some issues concerning radar, rain gauge networks and rain cell size. Chapter Three goes through the theory on attenuation due to rain, and explained some parameters that important in propagation study. It also reviews a couple of rain models and some reduction factors models. Chapter Four explains the rain gauge networks that were set up in this study. It also explains the data that were obtained from the Kluang radar, and how the data were retrieved. Chapter Five is an important part of this project. It analyses both data that were utilized in this study. Results in this study are given in this chapter. The main contribution of this study which the formulation of a reduction factor, is also given in this chapter. Comparison with other models and actual measurement are also done. Chapter Six concludes this study. It discusses the results of this study and shows that the objectives of this study are achieved. Future studies are also briefly given. CHAPTER 2 METHODOLOGY 2.1 Introduction The methodology will determined that the objectives are properly developed and gives the steps and procedures involved in achieving them. The methodology used in this study follows some works that have been published (Goddard, 1991; Goddard et al, 1997; Goldhirsh, 1992; Sauvageot et al, 1999). In this study, two types of measuring techniques are used. Radar data is used to deduce the reduction factor for calculating rain attenuation, to formulate rainfall rate distribution and rain cell size distribution, while rain gauge networks are used to obtain rain cell size and the profile of rainfall rate inside a rain cell. 2.2 Radar Data The ability of a radar to scan a wide area around the radar site and not just a particular path made it a very attractive for many types of investigations. Radar can be used to measure the rainfall rate indirectly. This is achieved by knowing the radar 8 reflectivity and then converting them into rainfall rate. The S-band frequency of the meteorological radar ensures that propagation effects such as attenuation are negligible (Goddard 1991). Many studies have utilized meteorological radar data (Battan, 1973; Puhakka, 1974;.Goldhirsh, 1979; Wilson et al, 1979). A radar will actively probes a specific region. The ability to scan a large area continuously makes radar measurement technique very attractive. This enables a large amount of data to be collected in a short period of time. A large database can be used to provide statistical information by simulating particular systems. Compared to a rain gauge network, radar observes larger variability of precipitation characteristics over a short period of time, and at a faster rate (Olsen, 1982). Radar provides valuable information that is relevant in modelling rain-induced propagation effects (Goddard, 1991). Radar also provides spatially and temporally continuous measurements that are immediately available at one location. Through technology and computer software advancement, radar can scan 3-D space, seeks out region of rain, and acquire a quasi-photograph of the precipitation structure. Goldhirsh (1979) has noted that researchers have demonstrated that a summer’s database of radar reflectivity enabled the prediction of rain rate distribution, which agreed in shape to the distribution acquired using 10 years of continuous rain gauge data. Lahaie et al (1993) suggested the use of 1000 virtual links to an attenuation model. (This study uses more than 5 millions virtual links and that is just for the 1-km path lengths). Seed et al (1990), utilized a month of radar data for his study. Wilson (Wilson, 1964) used radar data that covered a period of 19 days, while Jatila et al (1973) used radar data taken during summer of 1969. Thus, by using radar, adequate rain attenuation statistics can be obtained in a shorter or limited period of time. Techniques that utilize Doppler radar and polarimetric radar have also been employed (Hornbostel et al, 1979; Meneghini et al, 1997; Zhang et al, 2001). Using these kinds of radars, rain rate can be deduced, and hence attenuation can be 9 calculated. Also, drop size distribution (dsd) and rain height can be estimated. However, these data are not readily available in Malaysia. 2.3 Disadvantages of Radar Measurement In order to get reliable data; some disadvantages of radar measurement need to be discussed. Steps must be taken to minimize these effects. Radar reading tends to underestimate attenuation due to rain attenuation on the radar signal (Wilson et al, 1979). Other factors that affect radar measurements are radar domain cannot be sampled at consistent elevations, with consistent bin volumes, or for precipitation with similar stage of development or phase; beam elevation increases with distance resulting in low-level precipitation not represented well; bin size increases with distance, thus incomplete beam filling, smearing of small-scale structure; ground clutter and terrain features may block low level information. Some correction factor or technique has to be included in determining the received or backscattered power from precipitation. This is known as the STC Sensitivity Time Control, which takes into account the time and distance of precipitation. STC provides range normalization that eliminates the effect of range attenuation. Thus target of equal reflective area regardless of their range will give equal amplitude returns. The radar system also has to be calibrated regularly to ensure reliable data collection. The technology of radar for precipitation measurement is very stable and the results of observations are highly accurate (Japan Meteorology Agency, 1979; Goddard, 1991). Meteorological radar is also said to be a better tool to build a rain attenuation prediction model for terrestrial microwave radiocommunication (Lahaie et al, 1993). 10 2.4 Radar Measurement Principle Radar will emit a pulse towards a target. If the pulse hit the target, some energy will be scattered back to the radar. The backscattered or average power received by the radar depends on the radar parameters, on the shape, size, number and dielectric constant of raindrops. If these parameters are known, the radar reflectivity factor Z can be calculated. An empirical relationship can be established between Z and the rainfall rate R. By knowing R, specific attenuation can be determined. A brief radar measurement theory is given in Appendix 3.3. 2.5 Z-R Relationship Power received or returned from a scatterer depends on what is called the radar reflectivity factor Z. The widely accepted empirical relationship of the radar reflectivity factor Z, and rainfall rate R is given by eq. A22, in Appendix 3.3. A method of obtaining this relationship is by using disdrometer measurement. Disdrometer measures the drop size distribution of a falling rain. From time integration of the drops sampled by the disdrometer, the rain rate can be determined. This work has been carried out in UTM by Din (1997). From knowing the drop size distribution (dsd), the radar reflectivity that would have been can be calculated. Z can be plotted against R, and it is found that they have a “power law” relationship. This is given by equation A22, and written again here, Z = aRb mm6/m3, (R in mm/hr) (2.1) 11 Thus, if Z can be measured and thus known, then R can be found. However, a and b varies for different types of rain, as given by eq. A23, in Appendix 3.3. The constant a and b are related to the intercept and slope of the best-fit line through a plot of R versus Z on a log-log plot. Puhakka (1974) has given that if b is fixed at 1.6, then for convective rain, a has an average of 360, 196 for continuous rain, and 56 in drizzle. Reported values of a varies between 100 to 600, while b varies between 1.3 to 1.8. Meanwhile, Battan (1973) reported that Cantaneo (1969) proposed that the appropriate values of a and b can be predicted by using a = 1.372 (TD) - 4.702 (RH) + 571 (2.2) b = -0.00444 (RH) + 1.776 (2.3) where TD is the mean annual percentage of rain days which are thunderstorm days, and RH is the mean annual relative humidity at a level 500 m above the ground. However, Battan and other researchers (Battan, 1973) have noted that these parameters vary depending on geographic locality and types of rain. In addition, error in measuring the average backscattered power also contributed to the uncertainty. Notwithstanding, Wexler et al (1963) concludes that Z-R relationship is fairly constant at low frequency (less than 3 GHz), but deviates considerably at high frequency (more that 9 GHz). Hunter (1996) noted that the choice of Z-R relationship has small outcome in determining R. Z-R relationship is mitigated by averaging rain rate from radar data over large time and space scales. The most common values for a and b are 200 and 1.6, respectively. These are also the values used by the radar operator of the Malaysian Meteorological Department of Malaysia. Din (1997) proposed values of 320 for a, and 1.4 for b. 12 Falling raindrops are found to be mostly oblate. However, in this study, the radar uses vertical polarization. Thus, it is assumed that the raindrops are spherical, and small enough compared to wavelength so that Rayleigh scattering is applicable. This assumption is valid for detecting precipitation particles by meteorological data (Japan Meteorology Agency, 1979). The wavelength of the meteorological radar is 107 mm while study in Malaysia (Din, 1997) showed that for rain rates of 226.33 mm/hr and 105.02 mm/hr, most raindrops have diameters of between 1 – 2 mm. 2.6 Selection of Radar Data As it was mentioned before, the radar data is obtained from the Meteorological Department of Malaysia. Thus, one of the earlier steps in this study is the retrieval of radar data from the Meteorological Department. This is done through visitation, personal communications, and practical training at the Meteorological Department of Malaysia, Petaling Jaya during the course of this study. One of the first data processing activities is the rejection of permanent echo or permanent echo check. This is done by taking samples from the radar data and check whether there are any instances where there is always a reading at a particular point. This will indicates permanent echo. From this analysis, there is no permanent echo in the radar data. The operator of the weather radar of the Meteorological Department of Malaysia explained that they have included echo rejection in their data processing routines. Radial lines from radar data that are utilized in this study are selected. The “virtual links” for the deduction of reduction factor and the formulation of rain cell size distribution will be on these radial lines. This virtual links concept is as used by 13 Goddard et al (1997) and Jafri (1997). Since the rain rate along each radial lines are known, then the attenuation due to rain can be calculated as if there is a link or a microwave signal transmission along the path link. Figure 2.1 shows the virtual links as used by Goddard. Figure 2.1 Examples of virtual links (Goddard et al, 1997). In their study, Goddard et al (1997) used both links along the radial lines and orthogonal to the scanning radar. In this study, since the amount of data is very large, it is decided to use virtual links along the radial lines only. Also, not all the radial lines will be utilized. The range bin size for the Kluang radar is 1-km. 14 The total data that is available for this study is very large. It is unnecessary to put a virtual link in every radial line of the radar data. Thus, several locations or azimuthal angles were chosen in a quasi-random manner. Radial lines with the angles of 3, 6, 12, 15, 238, 241, 244, 247, 250, 294, 297, 300, 303, 306, 334, 337, 340, 343 degrees were chosen in this study. These angles for the “virtual paths” links were chosen by adopting similar technique employed by ITU-R and other researches (ITU-R Rec. 581-2; Chebil, 1997; Yagasena et al, 2000) in which the concept of “worst month” is considered and the highest exceedence probability were chosen. This is done so that “worst case scenario” is taken into consideration. In determining the suitable radial lines for the virtual links, results from the permanent echo test were used. From the permanent echo test, several areas where rain is likely to occur are determined. Bin ranges of 32 to 51 were chosen for analysis. This means that the data chosen were rain events occurring 36 km away from the Kluang Radar Station (readings for the Kluang radar station starts only after 4 km away from the station) and the virtual link has a maximum path length of 20 km. This range in adequate for deduction of reduction factor. It is also sufficient to cover rain cell size of interest. The 20 km range would well cover the individual rain cells that are likely to occur. This will also ensure that the problem of beam elevation is minimized. In addition, this exercise reduces the raw data considered in this study. The elevation angle chosen in this study is 0.5o, which is the lowest elevation angle of the radar system. Readings from this elevation angle is much nearer to the ground and thus giving the precipitation that occurs near the ground. Rain cell size distribution of 1-km integration is obtained from radar data. From this distribution and using curve fitting technique, the rain rate for 0.01% of the 15 time or R0.01 is determined. This value is used in the calculation of rain attenuation along the virtual paths. For reduction factor deduction, virtual links of 1-km to 10-km are constructed in each radial line. Since, bin range of 32 to 51 are used, this means that there will be two 10-km links in a radial line. Using the rainfall rate data in each bin, attenuation due to rain is calculated using parameter values obtained from ITU-R. Attenuation for 0.01 % of the time for 1 to 10-km path lengths are calculated at frequencies of 7, 10 and 15 GHz. They were chosen as they represent the spectrum that is usually used for terrestrial and satellite links. Moreover, frequencies of 7 and 15 GHz will enable comparison to be made with measurements done here in UTM. Reduction factor is then deduced from these calculations. 2.7 Rain Gauge Data Eventhough radar data is able to give the rain rate; it is averaged over 1-km range bin. It is interesting to know the profile of rain rate inside a rain cell. Thus, data from rain gauge networks are used to complement the radar data. However, there has been no work published in using rain gauge networks specifically for the determination of rain rate profile inside the rain cell. As a result, this will be the first attempt of such work. Most works has been using radar data, beacon or satellite signals. These are discussed in the next section. Since high rainfall intensity is usually of short duration in nature, the integration time for the recorded rain rate is a critical parameter. Studies (Ruthroff, 16 1970; Bodtmann et al, 1974; Watson et al, 1981) have shown that a 1-minute integration time should be adequate for rain rate measurements using a rain gauge. Two rain gauge networks were set up in this study. Details of these rain gauge networks will be given in the next chapter. All the rain gauges used in this study uses 1-minute integration time. Rain intensity and cell size for a single cell are obtained from these rain gauge networks. 2.8 Attenuation Measurements Using Microwave Links Some measurements on rain attenuation have been in UTM (Chebil, 1997; Md Rafiqul Islam, 2000; Karim, 2000). They concluded that available reduction factor models did not represent the attenuation in Malaysia. Md Rafiqul proposed modifications to a certain model to fit the measured attenuation data. This will be discussed further in the proceeding chapters. There was no work that has been done in UTM to obtain the reduction factor through radar data or rain gauge network. CHAPTER 3 RAIN ATTENUATION, RAIN MODELS, AND REDUCTION FACTORS 3.1 Background Signal attenuation becomes more severe as the rain rate increases. When the budget link is exceeded, communication ceased to exist. In a system design, the highest rain rate and its duration are considered and a suitable mathematical model is used. The rain responsible for producing a rain attenuation event can occur anywhere along the propagation path at heights where liquid precipitation is possible. The specific attenuation at any point along the path is related to the intensity of the rain at that point. However, rain rate along the propagation path is not constant. In order to account for the spatial variability of a rain event, a reduction factor r is applied to the specific attenuation to get the overall path attenuation due to rain. The purpose of this study is to analyze data obtained from weather radar, and to 18 formulate a suitable reduction factor that can be used to predict rain attenuation according to the Malaysian local conditions. 3.2 Attenuation by Rain Attenuation of radio propagation through space is given by (Freeman, 1987) A (dB) = 92.45 + 20 log F(GHz) + 20 log L(km) + a + b + c + d + e (3.1) where F = operating frequency in GHz, L = path length or distance in km, a = excess attenuation due to water vapor, b = excess attenuation due to mist and fog, c = excess attenuation due to oxygen, d = absorption losses due to other gases, and e = excess attenuation due to rainfall. Parameters b and d can often be neglected in calculation for propagation loss with respect to excess attenuation due to rainfall. Parameters a and c are known as atmospheric attenuation. For example, at 22 GHz for a 10-km line-of-sight (LOS) link, attenuation due to atmospheric attenuation is 1.6 dB, where as the free-space loss is 139.3 dB. This means that excess attenuation due to rainfall is more severe compared to other factors (Freeman, 1987). 19 Attenuation of radio wave due to rain is caused by absorption and scattering mechanism (Allnutt, 1989). Absorption occurs when the incident radiowave energy heats the scatterer until its temperature is above the surroundings and the scatterer reradiates the absorbed energy isotropically. Scattering occurs when energy is reradiated in all directions. The scattering cross-section can be calculated for rain attenuation studies. Frequency and temperature are included to find the scattering cross-section. Other important parameters for rain attenuation studies are drop shape, drop size distribution and rain rate (Freeman, 1987: Allnutt, 1989; Aydin et al, 2002). Attenuation of radio waves through rain (Ar), over path length (L), is given by (Rogers, 1976) L Ar = ∫ γs(l) dl dB (3.2) 0 where γs(l) is the specific attenuation of rain (dB/km). Equation 3.2 can be modified to use constant specific attenuation value, As, and include a reduction factor r, to account for the nonuniformity of rain (ITU-R P.530-8; ITU-R P.618-5), giving Ar = γs r L dB where γs = specific attenuation of rain (dB/km). r = reduction factor, and L = path length (km). (3.3) 20 Now, specific attenuation γs, can be found directly by theoretical calculation in which either measured drop size distribution or available drop size distribution models are used, or by indirect method of measuring drop size distribution, rain rate, radar reflectivity, or propagation of microwave through rain. The apparatus used for measurements are disdrometer, rain gauge, radar, and radiometer, respectively. Theoretical calculation of rain induced attenuation is a rather complex and tedious process. Specific attenuation can be calculated using the equation (Rogers, 1976) ∞ γs = 0.4343 ∫ Qt(r) n(r) dr dB/km (3.4) 0 where Qt(r) is the total attenuation cross-section (cm2), and n(r) dr is the number of drops per unit volume of space with radius in the interval between r and r + dr (m-3) or drop size distribution (dsd). Equation 3.4 takes into account attenuation due to absorption and scattering. Even then, some kind of statistics are needed to formulate the reduction factor r. Attempts are being made to find solution so that attenuation due to rain can be predicted and thus, reliable communication link can be designed. Reliable estimates of rain attenuation can be obtained through measurements taken over several years at frequency and elevation angle (for satellite link) of interest, or by using indirect measurements using disdrometer, rain gauge, radar, and radiometer. However, it has been shown that radar can provides data at a much shorter time period (Wilson, 1964; Jatila et al, 1973; Olsen, 1982; Seed et al, 1990; Lahaie et al, 1993). 21 Although rain attenuation can be determined by measuring drop size distribution, it is found out that specific attenuation due to rain is closely related to rainfall rate (Rogers, 1976; Olsen et al, 1978). Thus, specific attenuation can also be estimated by finding the rain rate percentage. And it is useful to relate rain induced attenuation to the rainfall rate since the rainfall rate R data is available throughout the world. An empirical equation that is much easier than the theoretical equation, has been shown and widely used, to relate specific attenuation to the rainfall rate is given by (Rogers, 1976; Olsen et al, 1978) γs = k Rα (3.5) where k and α are parameters that depend on drop size distribution, temperature and polarization. To produce k and α, a regression analysis was done by Olsen et al, 1978. Thus, drop size distribution is very important for rain attenuation prediction. It can be use to predict excess rain attenuation, and relate rainfall rate to excess rain attenuation. Drop size distribution is also used to relate radar reflectivity with rainfall rate. Some parameters that were usually used in propagation studies will be discussed. Overview of several models for estimating rainfall rates will also be done in the next chapter. A suitable rainfall rate to be used in this study is also determined. In addition to all these, some models for reduction factor is reviewed and then compared with the one formulated through this study. 22 3.3 Rainfall Rate, Rain Cell Size, and Rain Height A brief discussion on atmospheric conditions, terminal velocity, shape and drop size distribution of a raindrop is given in Appendix 1.1. Figure 3.1 shows how rain can occur. When a cold front of air meets a warm front, condensation of water vapor can occur. If the water vapor is heavy enough, precipitation, as manifested by rain, will occur. The intensity of rain is known as the rainfall rate, R, and measured in mm/hr. It has also been noted that hot and cold fronts developed at high heights experience significant microwave signal fade (Barbara et al, 1993). The heavier the rain rate, the higher will be the attenuation (Ajayi et al, 1988; Allnutt, 1989; Juy et al, 1990; Yagasena et al, 1995). Statistically, the diameter of rain cell decreases as the intensity of rain increases (Allnutt, 1989; Mitnik et al, 1998; Nor Hisham Khamis et al, 1999; Konefal et al, 2002) as shown in Figure 3.2. From these studies, the rain cell size that must be taken into account for measurements using radar or rain gauge network must cover an area of at least 1 km. For system planning, the rainfall structure such as horizontal and vertical dimensions of rain cells, spatial and temporal variability of rain is very important. With these information, rain attenuation and scattering can be determined and rain attenuation mitigation technique such as site diversity can be applied. 23 Figure 3.1 Rain formation through cold and warm air fronts (Allnutt, 1989). Figure 3.2 Rain cell diameter versus rainfall rate (Allnutt, 1989). 24 For slant-path propagation applications, such as satellite communications, the height of the rain structure is important. A region known as the ‘melting layer’ or the ‘bright band’ must be identified. Below this region, rain occurs and microwave propagation suffers attenuation due to rain. Losses in the ice region above the bright band are negligible for the purposes of most calculations (Eastment et al, 1996). Above this region, no rain occurs and the propagation suffers only spatial attenuation. Studies (Pontes et al, 1995; ITU-R P.839-2) have shown that convective rain in tropical region often has bright band above 10 km and also suffers from severe updraft. However, the bright band region usually exists at the height of 4.5 – 5 km. The ITU recommended that rain height for tropical region is 5 km (ITU-R P.839-2). A more recent study (Ong et al, 2000) has shown that the variability of rain height according to seasonal period and types of rain. Identification of the melting layer can be done through radar reflectivity measurement. Figure 3.3 shows the height of the melting layer. However, rain can sometimes occurs below the melting layer. This is shown in Figure 3.4. This is especially true for heavy rainfall situations. The uncertainty of the difference in height between rain event and the melting layer need to be investigated so that a reliable prediction of attenuation can be made. 25 Figure 3.3 Height of melting layer (Allnutt, 1989). 26 Figure 3.4 Rain height (Allnutt, 1989). 3.4 Rainfall Rate for 0.01 % of the time or R0.01 An important characteristic of rainfall is known as the R0.01 rain rate. This is the rainfall rate that will occur for 0.01% of the time. For a period of one year (364.25 days) which is equivalent to 524,5200 minutes, 0.01% of the time will be for 52.45 minutes. The determination of the rain rate for 0.01 percent of the time or R0.01 comes from the fact that a good system must provide at least 99.99% reliability. Design and system engineers use this value to construct communications system such 27 that the link is available for 99.99% of the time. I.e. the link can only be down for only for a total time of 52.45 minutes for the entire year. Systems built with this value ensure reliable microwave link and guarantee customer satisfaction. ITU-R has recommended the use of this parameter for attenuation studies. To determine rain rate for a specific percentage of time, rain prediction models can be utilized. These rain models can be used to determine the rain rate for 0.01 % of the tome or R0.01. 3.5 Rain Prediction Models Prediction models are used to provide the best possible estimates given the available information. Using these models, the rain intensity can be known and thus, the specific attenuation due to rain can be predicted. There are several rain models that are available in the literature of which, two of the often-used models will be discussed. These rain models can be used to calculate the specific rain attenuation, especially rain rate for 0.01% of the time or R0.01 where it usage is recommended by ITU-R. Values of R0.01 obtained from these models will be compared to the values obtained in this study. Two of the widely used models for rain intensity prediction, Crane’s Global Model and the ITU-R Model will be discussed here. 28 3.5.1 Crane’s Global Model Crane’s Global model (Crane, 1980) divides the world into eight regions, as shown in Figure 3.5. It is based on total rain accumulation, and the number of thunderstorm days from maps published by Landsberg (1974). Crane obtained additional guidance from the Koppen world climate classification. Boundaries were adjusted to accommodate variations in terrain, predominant storm type and motion, general atmospheric circulation, and latitude. Satellite and precipitation frequency data were used to extend the climate over the oceans. The measured 1-minute rain rate distributions that were available for each of the regions were pooled to construct the rain rate distributions as shown in Figure 3.6 and Table 3.1. Malaysia falls into region H, and has a rain rate of 145 mm/hr for 0.01 percent of the time. Figure 3.5 The global rain rate regions (Crane, 1980). 29 Figure 3.6 Rain rate distribution curves for various regions (Crane, 1980). Table 3.1 Point Rain Rate (RP) Distribution Values (mm/h) Versus Percent of Year Rain Rate is exceeded (Crane, 1980). Percent of Year 0.001 0.002 0.005 0.01 0.02 0.05 0.1 0.2 0.5 1.0 2.0 Number of Station Years of Data Rain Climate Region A 28 24 19 15 12 8.0 5.5 4.0 2.5 1.7 1.1 B 54 40 26 19 14 9.5 6.8 4.8 2.7 1.8 1.2 C 80 62 41 28 18 11 7.2 4.8 2.8 1.9 1.2 D1 90 72 50 37 27 16 11 7.5 4.0 2.2 1.3 D2 102 86 64 49 35 22 15 9.5 5.2 3.0 1.8 D3 127 107 81 63 48 31 22 14 7.0 4.0 2.0 E 164 144 117 98 77 52 35 21 8.5 4.0 0.4 F 66 51 34 23 14 8.0 5.5 1.2 1.2 0.8 0.4 G 129 109 85 67 51 33 22 14 7.0 3.7 1.6 H 251 220 178 147 115 77 51 31 13 6.4 2.8 0 25 44 15 99 18 12 20 2 11 30 3.5.2 ITU-R Model The ITU-R (Allnutt, 1989) (formerly known as CCIR) divides the world into a number of rainfall climatic zones as shown in Figure 3.7. Table 3.2 gives the rainfall rates for the climatic zones. Figure 3.8 is the rainfall rate contours for 0.01% of the time for Asia. From Table 3.2 and Figure 3.8, R0.01 for Malaysia is 145 mm/hr. Figure 3.7 Rainfall climatic zones (Allnutt, 1989) 31 Table 3.2 Rainfall rates in the climatic zones (Allnutt, 1989) Percentage of time 1 0.3 0.1 0.03 0.01 0.003 0.001 A B C D E F G H J K L M N P Q <0.1 0.8 2 5 8 14 22 0.5 2 3 6 12 21 32 0.7 2.8 5 9 15 26 42 2.1 4.5 8 13 19 29 42 0.6 2.4 6 12 22 41 70 1.7 4.5 8 15 28 54 78 3 7 12 20 30 45 65 2 4 10 18 32 55 81 8 13 20 28 35 45 55 1.5 4.2 12 23 42 70 100 2 7 15 33 60 105 150 4 11 22 40 63 95 120 5 15 35 65 95 140 180 12 34 65 105 145 200 250 24 49 72 96 115 132 180 Figure 3.8 Rainfall rate contours for 0.01% of the time for Asia (Allnutt, 1989) 32 Recent publication of the ITU-R shows a revised rainfall rate for 0.01% of the time for Asia. Figure 3.9 shows the recommended rainfall rate for 0.01% of the time for the Asian region (ITU-R P.837-4, 2003). The new R0.01 for Malaysia is 120 mm/hr. Figure 3.9 Revised rainfall rate contours for 0.01% of the time (ITU-R P.837) The new value for R0.01 of 120 mm/hr may had resulted from the fact that newer and more accurate data are available. Eventhough long-term data are available, most measurements have been carried out for meteorological or hydrological purposes. These data are usually taken for intervals of one hour or longer, whereas it has been proposed that a 1-minute integration time is better. These newer data also enables methods of rain rate conversion to have better accuracy in converting a tminutes rain rate into an equivalent 1-minute integration time rain rate (Chebil, 33 1997). Newer measurement themselves uses a 1-minute integration time, giving more accurate data in terms of rainfall rate. 3.6 Reduction Factor The inhomogenuity of rain cell give rise for the need to compensate for rain variability in rain induced attenuation calculation. One method of doing this is by introducing the reduction factor in the calculation. By doing so, the specific attenuation per distance can be employed and better value of rain attenuation on specific microwave link can be estimated. The following sections will discuss some of the reduction factors available in the literature. Even though they give the reduction to be used in calculating rain attenuation, none of the available literature describe in detail on how the reduction factor was obtained. A reduction factor is function of length, rain rate, frequency, and polarization (Assis, 1990). Some models are discussed here. 3.6.1 Lin Model The Lin Model (Lin, 1977) is rather simple and easy to compute. Lin found that the relationship between rain attenuation and path length is nonlinear. The Lin Model proposed the reduction factor to be 34 r= 1 (3.6) L 1+ L(R) where L(R) ≈ 2636 R P - 62 km. L is a characteristic path length such that the nonlinear factor equals one-half when L = L , and L related to the diameter of the rain cell. 3.6.2 Moupfouma Model Moupfouma (Moupfouma, 1984) based his works from data gathered in Congo, Europe, and the United States. He proposed the reduction factor to be r= 1 (3.7) −β p m 1 + 0.03 L 0.01 where m = 1 + 1.4 x 10-4 f 1.76 loge L 0.45 0.6 for for 0.001 ≤ p ≤ 0.01 0.01 < p ≤ 0.1 for L < 50 km 0.36 0.6 for for 0.001 ≤ p ≤ 0.01 0.01 < p ≤ 0.1 for L ≥ 50 km β = β = 35 3.6.3 CETUC Model Pontes (Pontes et al, 1993), Silva Mello (Silva Mello et al, 1997) and their colleagues did their research in Rio de Janeiro, Brazil, at the Center for Telecommunication Studies of the Catholic University of Rio de Janeiro (CETUC). Thus the climate is similar to Malaysia, which is also in the tropical region and near the equator. From their works, they produced a reduction that is described below, r= 1 1+ (3.8) L Lo where Lo = µ {1 + Rp(ν - ωlogp)}-1 (3.9) µ = 200, ν = 0.425, ω = 0.089 3.6.4 Improved CETUC Model Perez Garcia and da Silva Mello (Perez Garcia et al, 2004) proposed new CETUC model based on new research works, and additional data together with the already available data from ITU-R. This model is developed with the assumption that the nonuniformity of rainfall along the propagation path can be modelled by an equivalent uniform cell with a rainfall rate that is dependant on length of the terrestrial link, or specifically, Ar = γs.r.L (dB) (3.10) 36 where Ar = rain induced attenuation along a terrestrial link (dB), γs = specific attenuation due to rain (dB/km), r = reduction factor, and L = actual path link (km). The reduction factor is given as r = 3.445 L-0.164 Rp(-0.369 + 0.115/L) (3.11) where Rp is the rainfall rate at the percentage time of interest. 3.6.5 Goddard and Thurai Model Goddard and Thurai (Goddard et al, 1997) outlined several steps to calculate rainfall attenuation for a particular microwave link. They are as follows: - First, obtain rainfall rate, Rp that occurs for the time percentage p, of interest. Usually, this would be R0.01. Then compute specific attenuation As, for this rain rate, at the required frequency using ITU-R 838. Compute the path reduction factor r, from 37 s(L) = 2L-0.053 – 2.25 (3.12) where L = path length (km) r = 1.35 + s(L) log Rp (3.13) if r > 1, then r = 1 Compute the total path attenuation At using Equation 3.10. 3.6.6 ITU-R Model The ITU-R model (ITU-R P.530-8) is the model that many researches compared their results with. This is because ITU is the authoritative body governing the world of communications. With its working committees around the world, and results are always updated when new data are available. The ITU-R model can be utilized by following the steps below, Obtain R0.01 which is the rain intensity exceeded for 0.01% of the time (with 1-minute time integration). Compute specific attenuation As, for R0.01 at the required frequency using Recommendation 838. Compute the reduction factor r, using 38 r0.01 = 1 1+ (3.14) L Lo 35 e -0.015 R 0.01 where L 0 = 35 e -1.5 for R 0.01 ≤ 100 mm/hr for R 0.01 > 100 mm/hr Finally, compute A0.01 using Equation 3.10. For other time percentages, ITU-R recommends the use of the following equation A s = 0.12 A 0.01 p -[0.546 + 0.043 log p ] (3.15) 3.6.7 Singapore Model Singapore is located just below the Malay Peninsula. Thus it shares many of rain characteristics with Malaysia. The distance between UTM, Skudai and Singapore is about 50 km. However, Singapore is a small island surrounded by water, which might affect the local weather conditions. The reduction factor proposed by Ong (Ong et al, 1995) has similar characteristics as compared to the ITU-R model. It is described as follows, r = 1/(1 + L0.9/L0) (3.16) 39 where L0 = 35 exp (-0.015R0.01) 3.6.8 DAH Model Asoka Dissanayake is a senior scientist at the COMSAT Laboratories, Maryland, U.S.A. Dr Jeremy E. Allnutt is a professor and the Director of Masters in Telecommunication program at the George Mason University. Fatim Haidara is a senior engineer at Intelsat. They (Dissanayake et al, 1997) proposed a reduction factor to be used in the calculation of rain attenuation for earth-satellite path that can be also applied for terrestrial path. The reduction factor they proposed is 1 r= (3.17) LAs 1 + 0.78 − 0.38[1 − exp(−2 L)] F where As = specific attenuation calculated using R0.01, or = k R0.01 α L = path length (dB/km), (km), and F = frequency in GHz. This model has been adopted by the ITU-R recently (ITU-R P.618-6, 1999). 40 3.6.9 Comments on Reduction Factor Models Lin deduced his model from measurements done in Palmetto, Georgia, U.S.A. Thus, the weather conditions are different from Malaysia. He obtained his data using 5-minutes integration time tipping bucket rain gauges, and an 11 GHz microwave link with a path length of 42.5 km. The current accepted integration time as proposed by ITU-R is 1-minute. The shorter time period would give a better reading of the rainfall rate especially for climatic region such as Malaysia where the rainfall rate is much heavier. Lin also included wet radomes effect in his calculations. Depending on the structure of the radomes, he used the values of 4-8 dB. These values may be different from the microwave systems that are employed in Malaysia. Thus, accurate determination of the wet antenna effect is desirable. The Moupfouma, ITU-R, and DAH models used data that were available from several places. Some data are taken from temperate regions, while others are obtained from tropical regions. These models tried to be as universal as possible. However, studies (Ajayi et al, 1988; Juy et al, 1990; Yagasena et al, 1995) have shown that rain events are rather localized and models need to be formulated using local data. This has been the argument of carrying out this study. The CETUC, Improved CETUC, and Singapore models are works that have been done in tropical region. This means that the weather conditions are rather similar. So, it interesting to compare such works with this study. However, there are several differences in the local terrain conditions. Singapore is a small island. A body of land surrounded by water. The CETUCs models are formulated in Brazil, near Rio de Janeiro. Heavy rain forest and near the ocean. Only the Goddard and Thurai used radar as their measurement instrument to formulate the reduction factor, which the same as this study. However, their works 41 were done in Chibolton, UK. Thus the climate is different. The Singapore model used rain gauge network and beacon data. The CETUCs models used rain gauge, satellite and microwave links. The DAH model is formulated using beacon signal and radiometer measurements. No doubt there are many factors and differences in the formulation process of these models. Investigations on reduction factor models have been done locally by Chebil (1997) and Md Rafiqul Islam (2000). Md Rafiqul Islam concluded that the ITU-R model does not reflect the measured attenuation done at seven sites in Malaysia. The model underestimates rain attenuation. The Moupfouma and Lin models are closer to the measured data, but the Moupfouma model overestimated attenuation in some place where attenuation measurement were done. Md Rafiqul Islam proposed modifications to the Moupfouma model to suit the attenuation measurement data. However, there are some issues regarding the attenuation measurement done. This will be discussed later in this thesis. 3.7 Rain in Malaysia The monthly cumulative distribution of rainfall is influenced by the seasonal monsoons, namely the Northeast monsoon from October to March, and the Southwest monsoon from April to September. The Northeast monsoon is the primary rainy season in Malaysia. Together with cold air rushing from Siberia this monsoon produce heavy rainfall and usually causes floods in the east coast part of the Malay Peninsular and Sarawak. The Southwest monsoon is relatively drier except in Sabah. The rain events consist of convective and widespread rain. Convective rain is characterised by intense rainfall over a short period and covers a limited area. The widespread/stratiform rain is characterised by medium and low intensity rain over a longer duration and covers a wider area (Yagasena et al, 1995). 42 Interesting parameters that are closely related to rain attenuation are rain cell size or diameter (d) and the effective rain column height (he). In this study, rain cell sizes were investigated through rain gauge data and radar data. Data for rain height were available but due to the scope of this study, was not investigated. It will be however, look into in future studies. It is found out that most rainfall in tropical areas is ‘convective’ in nature (Pan et al, 1994; Nor Hisham Khamis et al, 1999, Nor Hisham Khamis et al, 2004). At lower rainfall rates, rain is rather widespread and uniform. The higher rainfall rates exhibited rather random distribution, having multiple cells and also higher rain height. 3.8 Determination of Rain Cell Size Most studies (Kuhn, 1989; Pan et al, 1992) concluded that rain cell depends on rain intensity. Kuhn (1989) found that for rain rate above 60 mm/hr, the mean diameter of rain cell was less than 1.5 km. He did his research in Germany which is of course, has different climate as compared to Malaysia. However, Pa et al (1992) did their research in Papua New Guinea, which is in the tropical region. They concluded that rainfalls are mainly convective with relatively small diameters. These intense convective rains also usually occurred in relatively shorter period of time as compared to widespread rain. Goldhirsh et al (1992) and Sauvageot et al (1999) noted that there is no generally accepted and preferred definition of a rain cell. A number of researchers have employed their own definitions. One definition is the minimum detectable contour values of rain rate (Goldhirsh et al, 1992). In his earlier work, Goldhirsh 43 (1983) defined rain cell diameter as “those distances over which the rain rate continuously exceeds Rt,” where Rt is the threshold values of a specific rain rate. He has selected Rt’s of 2, 5, 10, 20, 50, 100, and 150 mm/hr. This is also the definition used by Sauvageout et al (1999). If Rt is equal to the minimum detectable rain rate, then both definitions of rain cell are in agreement. Goldhirsh later defined rain cells as having “core” values of rain intensities and nested families of rain rate isopleths (Goldhirsh et al, 1992). In this study, rain cell size distribution is of interest, regardless of the core rain rate values. Previous study (Nor Hisham Khamis et al, 1999), has shown that rain cell cores having intensities of 120, 90, 60, and 30 mm/hr have diameter between 0.95 to 1.2 km. The averaged cell size of core rain rate of 75 mm/hr is shown to be 1.2 km. Thus, it is felt that it is better to formulate the rain cell distribution against the percentage of time. As a result, the definition of a rain cell size in this study is the minimum detectable rain rate, which is 0.2 mm/hr from the radar data. CHAPTER 4 RAIN GAUGE AND RADAR DATA 4.1 Introduction An important parameter in calculating attenuation due to rain is the rainfall rate. This is also the main parameter being measured in meteorology. The apparatus widely used to measure rainfall rate is the rain gauge. A rain gauge provides a fairly simple and accurate measure of point rainfall rates. However, the measurement is only at certain points. Rainfall events vary widely in both time and space. This means that an occurring rain event is not constant. A rain event may start heavily but the rain rate may tapers off after a while. Characteristics of rain events may also differ from one place to another. For example, temperate region has significantly larger drop size compared to tropical region when both experienced the same rain type. Even in tropical region, stratiform rains have larger drop size than convective rains (Atlas et al, 1995). Thus, while rain gauge gives good measurement of point rainfall rate; it is not adequate for most needs. 45 However, measurement done at a particular location will give some indications of the characteristics of the rain events. This will enable comparison with data obtained using the same technique. The data gained from a network of rain gauges is accurate for a small area, but not practical for large areas, remote land areas, and the ocean. In order to overcome this problem, other types of measurements such as radar, remote sensing and radiometer are used. Rain gauge data for this study was obtained by setting up two rain gauge networks in the Universiti Teknologi Malaysia, Skudai Campus. While radar data was obtained from the Meteorological Department of Malaysia (Kluang Radar Station). 4.2 Rain Gauge and Rain Gauge Networks A preliminary rain gauge network was set up in the Skudai campus (RGNUTM 1). The network of rain gauges consisted of three rain gauges. One rain gauge was placed in an open area in Kolej Merbau (Merbau), one on top of a Civil Engineering Lab building (Civil), and the other one on top of a Mechanical Engineering Lab building (Mechanical). The positions of the rain gauges were almost on a straight line with distances of 0.41 km from Merbau to Civil, 0.55 km from Civil to Mechanical, and 0.95 km from Merbau to Mechanical. Figure 4.1 shows the locations of all the rain gauges. 46 Figure 4.1 RGN UTM 1 Rain Gauge Stations The rain gauges were 0.5 mm Casella tipping-bucket type, which record the number of tipping in a 1-minute integration time. No data is recorded if the bucket does not tip which means there is no or very little rain (less than 0.5 mm/minute or 30 mm/hr). The specifications of the Casella rain gauge are given in Appendix 3.1. Data were collected from 16 September until 28 November 1996. This amounted to 74 days of data collection. However, rain events occurred only for 43 days, while there was 36 days where rain was recorded at all three rain gauge stations simultaneously, giving valuable insight into the spread of rain events. A sample of the collected data is given in Appendix 3.2. A second network of rain gauge was installed in the UTM, Skudai campus consisted four Casella Tipping Bucket with Integral Logger (RGN-UTM 2). Four 47 locations were chosen, the first one being near the Faculty of Chemical Engineering (Chemical). Except for the Chemical site, which has a 0.2 mm tipping bucket, the rest of the rain gauges have 0.5 mm tipping buckets. Other sites are Institut Voltan dan Arus Tinggi (IVAT), TV Studio, and Kempas. The rain gauges were installed such that the distances between them are about 250 meters. Figure 4.2 shows the locations of all the rain gauges. Figure 4.2 RGN-UTM 2 Rain gauge network stations. 48 This network of rain gauges was set-up around the campus in December 1998. RGN-UTM 2 was set-up for a year but unfortunately some data were lost during transferring from the data logger into the computer notebook. Thus, available data were from December 1998 until July 1999. The actual data collection ran until December 1999. 4.2.1 Casella Rain Gauge The Casella rain gauge is of tipping bucket type, which is fitted with a solidstate logger (Casella, 1996). It employs a lightweight plastic injection moulded tipping bucket and support assembly. The construction of the rain gauge is quite robust and has a very long life term. The Casella rain gauge is supplied readily calibrated to indicate 0.1 mm, 0.2mm or 0.5 mm of rainfall with each bucket tip. In this study, bucket sizes of 0.2 mm and 0.5 mm are used. The number of tips per minute are counted and stored in the integral logger. Figure 4.3 shows the Casella tipping bucket rain gauge with an integral logger. 49 Figure 4.3 Casella Tipping Bucket Rain Gauge with Integral Logger. The integral logger operates on 9 V batteries. It contains a powerful CMOS micro controller with 4K of EPROM program storage, 32K of solid-state data storage, real time clock, battery backup and a serial communication interface. Figure 4.4 shows the functional diagram of the tipping bucket rain gauge. During rain fall, rainwater is funnelled into one of the two buckets, which is rest on a stainless steel pivot pins in such a way that when a 0.2 mm (or 0.5 mm) of rain has been collected in the buckets, it tips and momentarily closes a magnetic reed switch. 50 The tipping action discharges the water and at the same time causes the other bucket to start collection. The number of tips per minute are counted and stored in integral logger. Rainfall data is stored in the logger on a 1-minute time basis. Thus, the integration rainfall rate stored will be 12, 24, 32, 64, 108, …, mm/hr for 1, 2, 3, 4, 5, …, tips per minute of the bucket for the 0.2 mm bucket Figure 4.4 Tipping buckets of a Casella Rain Gauge. 4.3 Radar Data Collection The radar data was obtained from the Kluang Radar Station of The Meteorology Department of Malaysia. The Kluang radar station is located about 61 km from the UTM, Skudai campus. For the Malay Peninsular, the Meteorology Department has five radar stations located in Kluang, Subang, Kuantan, Butterworth, and Kota Bharu. For system management and connectivity, the Meteorological Department of Malaysia uses the 3D-RAPIC software produced by the Australian 51 Bureau of Meteorology, Observations and Engineering Branch, Radar Engineering Section. Calibration of the Kluang radar system is done every 6 months. A partial checklist for calibration is given in Appendix 3.4. Using this 3D-RAPIC system, data from all radar stations were integrated to give a whole radar scan of the Malay Peninsula. Figure 4.5 shows an example of a ‘merged’ PPI scan of a radar scan produced by the 3D-RAPIC system. Figure 4.5 Merged PPI scan. 52 The reflectivity color palette at the bottom of Figure 4.5 indicates the intensity of the occurring rain events. The specifications of the radar system have 16 levels to indicate the intensity of the falling rain. Every level corresponds to a particular reflectivity value or reflectivity factor (Z), which in turns corresponds to a particular rain rate. Table 4.1 gives the dBZ level (value of Z in dB), and the corresponding rain rate R (mm/hr) for the Kluang radar station. Table 4.1 dBZ-R values for Kluang radar. Level dBZ R (mm/hr) 1 11.8 0.2 2 23.0 1.0 3 28.1 2.1 4 31.0 3.2 5 34.0 4.9 6 37.0 7.5 7 40.0 11.5 8 43.0 17.8 9 46.0 27.3 10 49.0 42.1 11 52.0 64.8 12 55.0 99.9 13 58.0 153.8 14 61.0 236.8 15 64.0 364.6 The radar scans can also be viewed individually for each radar station. Figure 4.6 shows a pulse-position-indicator (PPI) radar scan for the Kluang radar station. 53 The PPI gives the raining events as they occur. Figure 4.7 shows the rain height indicator (RHI) scan, which is the product of volumetric scanning. The RHI scan gives the vertical profile of the raining events. It indicates the height at which the rain is falling. Both figures are shown simultaneously on the monitor screen of the 3DRAPIC system. Figure 4.6 A Kluang radar station PPI scan. 54 Figure 4.7 A Kluang radar station RHI scan. In this study, only the data from the Kluang radar station is utilized. The specifications of the Kluang radar stations are as given in Appendix 3.5. The chief controller of these radar stations is located at the Head Office of the Meteorological Department in Petaling Jaya. To operate the radar system, the Meteorology Department uses a Silicon Graphics Indigo UNIX Station. As mentioned earlier, the RAPIC software, developed by a company in Australia was used to process the radar data. After processing and extracting the necessary information from the radar data, these data are stored in DAT tapes. Thus, to obtain these data, a DAT tape system (HP SureStore) was acquired. After copying the data in DAT tapes, these data are then written on CDs for use with PCs. This is done in the campus using our own UNIX 55 station. The data consist of all the data from the five radar stations. Thus, using MATLAB, a program was written to extract the Kluang radar data. Further analysis and processing the data also uses MATLAB. Figure 4.8 shows an example of a radar scan plot using MATLAB. Figure 4.8 A MATLAB radar plot. In its normal operational mode, the Kluang radar will do a composite PPI scan every 10 minutes, and a volumetric scan every 30 minutes. The PPI scan will lasts for a minute where the revolution or rotation of the antenna is 3 rpm. The displayed composite PPI scan is a combination of 3 scan-angles 3.5o (0 – 30 km), 2o (30 – 100 km), and 0.5o (100 – 500 km). The STC for Kluang radar station is up to 56 230 km. Figure 4.9 shows the top view of a typical composite image, while Figure 4.10 shows a cross-section of a typical composite image. Figure 4.9Top view of typical composite image. 1 2 3 4 5 6 Legend antenna beam buildings hills rain below antenna beam cloud formation rain above antenna beam Figure 4.10 A cross-section of a typical composite image. 57 For volumetric scan, the antenna beam will pass or rotates azimuthally 15 times in the duration of 5 minutes. After each pass or rotation, the antenna beam is elevated to a higher elevation angle. These angles are also known as ‘volumetric elevation angles’ as it is done in the volumetric scanning. Since the antenna does 15 rotations, there will be 15 volumetric elevation angles. Table 4.2 gives the volumetric or scanning elevation angles for the Kluang radar station. The range of the Kluang radar station is from 4 km up to 512 km in a radial direction. The range bin is 2 km for the composite PPI scan and 1 km for the volumetric scan. Scanning is done for every angle in the azimuthal plane. The mechanism of the radar system is such that for a 1-km range bin resolution, it will send 8 pulses in every 250-meter range. The reflected power is then averaged. Four averaged readings will again be averaged for four 250-m ranges making a reading for 1-km range bin. Thus, 32 pulses are averaged for a 1-km range bin. Table 4.2 Radar Scan Elevation Angles. Pass/Rotation 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Volumetric Elevation Scanning angles 0.5 deg 1.2 deg 1.9 deg 2.7 deg 3.5 deg 4.6 deg 6.0 deg 7.5 deg 9.2 deg 11.0 deg 13.0 deg 16.0 deg 20.0 deg 25.0 deg 32.0 deg 58 For this study, the 1-km range bin is more useful and thus, the volumetric scan. Operation of Kluang Radar Station was altered so that it will do volumetric scan for every 10 minutes instead of every 30 minutes. This generated enough amount of data that was used in this study. The period of the radar data is from 2 January 1998 at 4:23 (Universal Time Constant - UTC) up to 10 March 1998 at 1:03 (UTC). From 2 January to 8 January, the volumetric scanning was done at 30-minute interval, and from 8 January (1:03 UTC), the volumetric scanning was done in a 10minute interval. Total scans where rain is detected are 7998 scans. 4.3.1 Kluang Radar Data Format An example of the radar data is given in Appendix 3.6. It is stored in the ASCII Asynchronous Format. This data format enables high data transfer rate since it uses run length encoding and empty radials were not recorded. The format and encoding algorithm for the radar data is as explained in Appendix 3.7. In order to utilize the data, the encoded data is decoded. A MATLAB program was written for this purpose. The program produces 360 x 101 matrices of the radar scans in which the first column is the azimuthal scanning angles and the following columns are the level of rain rate for each range bin. This gives a radial of 100 km that is sufficient as it well covers the typical diameter of a rain cell. The data is kept in 17 MATLAB file folders. These folders are called Data1 to Data17. The number of 360 x 101 matrices in each folders is as given in Table 4.3. Since rain rate is the wanted parameter, the level of rain rate in all the matrices is then converted into actual rain rate using Table 4.4. The values in this table are supplied by the Meteorological Department of Malaysia. 59 Table 4.3 Number of 360 x 101 matrices in each data folders. Data folders Number of matrices in each data files Data1 175 Data2 648 Data3. 949 Data4 906 Data5. 721 Data6 1041 Data7 955 Data8 792 Data9 869 Data10 777 Data11 785 Data12. 893 Data13 807 Data14 662 Data15 888 Data16 1049 Data17 996 Total no. of matrices 13913 60 Table 4.4 The corresponding actual rain rate values in mm/hr to the rain rate level. Level Rain rate in mm/hr 0 0 1 0.2 2 1.0 3 2.1 4 3.2 5 4.9 6 7.5 7 11.5 8 17.8 9 27.3 10 42.1 11 64.8 12 99.9 13 153.8 14 236.8 15 364.6 From Table 4.4, the actual number of individual data is 500,868,000. This is a very large number. Not all the data are used in this study. Data are selected in a quasi-random manner as will be explained in the next chapter. CHAPTER 5 DATA ANALYSIS AND RESULTS 5.0 Introduction Data analysis will be divided into two parts. Section 5.1 will be on the analysis and results on the rain gauge data while the following Section 5.2 will be on the radar data. 5.1 Rain Gauge Data Analysis In this section, data from the two networks of rain gauge is discussed. The important information extracted from the analysis are rain cell size and the distribution of rainfall inside a rain cell. 62 5.1.1 Preliminary Data Analysis Preliminary analysis was done on the data obtained from the network of 3 rain gauges set up in the Skudai campus (RGN-UTM 1). Table 5.1 shows the rain events recorded at each station, and plotted as in Figure 5.1. The rain rate is obtained from the 1-minute integration time for the Casella rain gauge. Amount of rain per minute is then converted to per hour. Table 5.1 Rain events recorded at all rain gauges locations. No. of readings Rain Rate (mm/hr) Merbau Civil Mechanical 30 577 533 575 60 120 95 102 90 18 17 15 120 3 10 8 150 0 0 1 Total 718 655 701 Table 5.1 shows the individual readings of rainfall rate occurring at all the rain gauge stations. It shows that most rain events occur at low rainfall rate, namely 30 and 60 mm/hr. High rainfall rates are few with 150 mm/hr occurs only once at the Mechanical station. From the total time of the collection of data, which is 105,400 minutes; the 120 mm/hr rain rate is 0.0028 %, 0.0095 % and 0.0076 % for the Merbau. Civil, and Mechanical station, respectively. From the ITU-R and radar data, these values are very close to the value of 0.01 % for rainfall rate of 120 mm/hr. This means that the data collected were good and reliable. 63 A closer examination of Table 5.1 indicates that rain events that have occurred are rather distinctive. Eventhough the Civil station is located between the other two stations; it recorded the lowest number of rain events. Considering the fact that these rain gauges are rather close to one another, this means that the boundary between raining and nonraining area is quite distinguishable. This may be due to the highly convective nature of rain. 700 No. of readings 600 Merbau 500 Civil 400 Mechanical 300 200 100 0 0 20 40 60 80 100 120 140 Rain rate (mm/hr) Figure 5.1 Graph of Rainfall Rate Recordings at all Rain Gauge Stations. Table 5.2 Correlations for no. of readings at all stations. Merbau Civil Merbau 1 Civil 0.999306 1 Mechanical 0.999379 0.999975 Mechanical 1 160 64 There is a high correlation of rain event at all three stations as can be seen from Table 5.2 and Figure 5.1. The high correlation is for the cumulative collected data after a period of time. Eventhough individual rain events might differ and vary in both time and space; rain events recorded at two different places will produce similar patterns if they experienced the same climatic weather conditions. This again shows that the data collected were stable and sufficient. 5.1.2 Selection of Rain Gauge Network (RGN-UTM 1) Data To get the distribution of rainfall rates inside a rain cell, rain events that occur simultaneously at all rain gauge stations are considered. When this happens, the rainfall rate recorded at all stations can be assumed to be from the same rain cell. Since the Civil station is at the middle of the rain gauge stations arrangement, and to get the profile of rainfall rate distribution inside a rain cell; of all the simultaneous recordings; only when the readings at Civil station are equal or higher than the other two stations are selected. Rain events that were recorded only at the Civil station are also included for there is the possibility that the rain cell is very small that it only covers the Civil station. Thus, of the rain events that occurred simultaneously; the ones in which the Civil station have a rain rate of equal or greater value than the other two stations, plus rain events that occurred only at the Civil station were selected for analysis. This will give the conditional cumulative distribution function for rain events in the Civil station to indicate the rainfall rate distribution inside a rain cell. 65 5.1.3 RGN-UTM 1 Data Analysis There are 360 times or events that is equivalent to 360 minutes of rainfall for which the category of rain events as mentioned in the preceding section, occurred in the selected rain gauge data. Out of the 360 recorded rain events, 205 times, rain was detected only at Civil, and none or less than 0.5 mm/min at the other two stations. This shows that 57 % of raining events considered, (205/360 * 100%) or a little more than half of rain cells, are less than 1 km in diameter (assuming circular rain cells). There are 113 times of rainfall rates that are equal at all stations (see Table 5.3). This phenomenon can be attributed to the widespread rain events. This is about 31.4 % (113/360 * 100%) of the total raining events being considered. Of these equal rain intensity recorded at all three stations, none was detected at 120 mm/hr, 1 time occurred at 90 mm/hr, 15 times occurred at 60 mm/hr, and 97 times occurred at 30 mm/hr. This showed that heavy, widespread rain events are low occurrence phenomena and most rain cells are small and convective. These results are summarized in Table 5.3. Table 5.3 Equal Rainfall Rate At All Three Stations. Rain Rate (mm/hr) No of readings 120 0 90 1 60 15 30 97 Total 113 66 Further analysis of the data being considered shows that when a rainfall rate of 120 mm/hr was recorded at the Civil Station, average rainfall rate at Merbau and Mechanical Stations are 18 and 72 mm/hr, respectively. Total time for this event was 5 minutes. This is shown in Figure 5.2. From the curvefit line, the rain cell size is about 1.2 km in diameter. The curvefit line also shows that the rain distribution inside the rain cell is convective. Rain Gauge Stations 140 rain rate (mm/hr) 120 120 Curvefit Line 100 Civil R2 = 1 80 72 60 Merbau 40 Mechanical 20 18 0 -0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 distance (km) Figure 5.2 Rain Rate Distribution for 120 mm/hr at Civil Station. For 90 mm/hr at Civil Station, 39 mm/hr was recorded at Merbau Station, and 42 mm/hr was recorded at Mechanical Station. This occurred for 10 minutes. For 60 mm/hr at Civil Station, 25.6 mm/hr ware recorded at both Merbau and Mechanical stations, for 27 minutes. When 30 mm/hr was recorded at the Civil Station, no rain event (or very light rain less than 30 mm/hr) was recorded at both Merbau and 67 Mechanical stations. This kind of rain event occurred for 205 minutes. These results are given in Figure 5.3 to Figure 5.5. These figures also give the interpolation of rainfall rate versus distance for the rain gauge network. This is shown by the curve fit lines in the figures. Figure 5.6 gives the averaged rain cell size distribution and averaged rain intensity distribution inside a rain cell. 100 Rain Gauge Stations 90 rain rate (mm/hr) 80 Curvefit Line 60 Civil 40 42 39 R2 = 1 20 Merbau Mechanical 0 -0.2 0 0.2 0.4 0.6 0.8 1 -20 distance (km) Figure 5.3 Rain Rate Distribution for 90 mm/hr at Civil Station. 1.2 1.4 68 Rain Gauge Stations Curvefit Line 70 60 rain rate (mm/hr) 60 50 40 Civil R2 = 1 Mechanical 30 25.6 25.6 20 Merbau 10 0 -0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 distance (km) FIGURE 5.4 Rain Rate Distribution for 60 mm/hr at Civil Station. 35 Rain Gauge Stations 30 30 Curvefit Line rain rate (mm/hr) 25 Civil 20 R2 = 1 15 Mechanical 10 Merbau 5 0 -0.2 0 0.2 0.4 0.6 0.8 1 1.2 distance (km) FIGURE 5.5 Rain Rate Distribution for 30 mm/hr at Civil Station. 1.4 69 90 75 80 averaged rain rate (mm/hr) Rain Gauge Stations Curvefit Line 70 60 50 R2 = 1 Civil 40 34.9 30 Mechanical 20.65 20 Merbau 10 0 -0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 distance (km) FIGURE 5.6 Averaged Rain Rate Distributions for RGN-UTM 1. Figure 5.2 to Figure 5.5 showed that most rain cells are convective. Except for rain rate of 30 mm/hr at the center of a rain cell, which has cell size of 0.95 km or maybe less; all other rain rates showed rain cell size of 1.2 km. From Figure 4.6, it can be seen that the averaged rain rate cell size is also 1.2 km. The total rain rates occurring at all rain gauge stations are given in Table 5.4. The duration of the measurement taken is 74 days. The exact duration is 105,400 minutes. 70 Table 5.4 The Rainfall Rates and their durations for all Rain Gauge Stations RGNUTM 1. Duration of events in minutes Rain Rate Merbau Civil Mechanical 0 104682 104745 104699 30 577 533 575 60 120 95 102 90 18 17 15 120 3 10 8 150 0 0 1 105400 105400 105400 (mm/hr) Total time duration (min) The percentages of rain rates occurring at all rain gauge stations are given in Table 5.5. This table also gives the averaged time percentage of all stations. The time percentages are taken for the duration of a particular rain rate over the total duration of the measurement taken. 71 Table 5.5 The Rainfall Rate Percentages and the Averaged Rainfall Rate Percentage for all Rain Gauge Stations. Rain Rate (mm/hr) Merbau Civil Mechanical Averaged Rain Rate (Percentage of time) 0 99.31879 99.37856 99.3349146 99.34409 30 0.547438 0.505693 0.5455408 0.532891 60 0.113852 0.090133 0.09677419 0.100253 90 0.017078 0.016129 0.0142315 0.015813 120 0.002846 0.009488 0.00759013 0.006641 0.00094877 0.000316 150 0 0 The values of the averaged rain rate are plotted in Figure 5.7. Best-fit equation for averaged rain rate from Figure 5.7 gives R0.01, which is the percentage of the rain rate for 0.01 % of the time as 119.031 mm/hr. This value is very close to the one suggested by ITU-R (120 mm/hr), and the value that is extracted from the radar data (120.907 mm/hr), as will be shown later in this thesis. 72 10 averaged rain rate (mm/hr) 10 10 10 50 Average R vs. percentage curvefit line 0 -50 f(x) = 129.2*exp(-8.205*x) R2 = 0.8845 -100 0.01% ==> 119.028 10 10 10 10 -150 -200 -250 -300 0 20 40 60 percentage of time 80 100 Figure 5.7 Averaged Rain Rate Distribution for all stations. 5.1.4 Cell Size and Intensity Distribution From the analysis of the data, it can be said that the rain cell size is closely related to the type of rain, either widespread or scattered. For widespread rain events, the cell size is large (more than 1 km). The majority of this type of rain occurred at 30 mm/hr for about 97 minutes of the raining time being considered. The heaviest for this type of rain is 90 mm/hr, which occurred for 1 minute. These results can be used for rain attenuation considerations in microwave link design. For scattered rain, the average rain cell size is 1.2 km. This is also true for higher rain rate, namely 120, 90, and 60 mm/hr. There are indications that several cells may be present in a wider 73 scale of raining events. Not withstanding, the results given can be used for microwave link design and site diversity applications. The limitation of the equipment used for data recording enables only for the rain rate to be recorded in 0.5 mm/minute integrations. This means that the recorded rain rates are multiples of 0.5 mm/minute or 30 mm/hr. Interpolation has been used to smooth out the data. Preliminary results from the first rain gauge network set-up (RGN-UTM 1) in the Skudai campus showed that Malaysia has a different rain cell distribution from temperate region where available attenuation models are formulated. This preliminary investigation has given an indication of rain cell size and distribution of rain intensity inside a cell. For lower rain rate, the rain cell size tends to be widespread and uniform. The larger rain rates, have narrower the rain cell size, as can be deducted from the collected data. (Nor Hisham Khamis et al, 1999). 5.1.5 Rain Distribution Inside a Rain Cell Figure 5.8 shows the rain rate distribution for RGN-UTM 2. The Chemical Station uses 2 mm buckets and is able to measure lower rainfall rate. However, readings at higher rainfall rate percentage are almost similar. Figure 5.9 shows the rain distribution assuming the TV Station to be the center of the rain cell while Figure 5.10 shows rain distribution assuming the IVAT Station to be the center of the rain cell. From these figures, it can be seen that a rain cell with a center rainfall rate of 120 mm/hr will have a diameter of about 1.2-1.5 km. Trend of high rainfall rates have a smaller diameter when compared to low rainfall rates is also exhibited by data from RGN-UTM 2. 74 2500 no of occurences Kempas 2000 TV Lab 1500 IVAT 1000 Chemical 500 0 0 50 100 150 200 rain rate (mm/hr) Figure 5.8 Rain rate distributions for RGN-UTM 2. TV 120 140 rain rate (mm/hr) TV 90 120 120 TV 60 100 100 80 78 64 60 70 54 57.6 51.7 40 20 Curvefit Line TV 120 90 R2 = 0.9295 60.0 37.7 31.1 Chem IVAT 0.25 0.5 TV Studio Kempas 0 0 0.75 1 1.25 distance (km) Figure 5.9 Rain distributions assuming that TV Station is the center of the rain cell. 1.5 75 IVAT 150 160 150 IVAT 120 140 120 IVAT 90 rain rate (mm/hr) 120 105 IVAT 60 100 80 77 60 72 56.6 40 77.5 67.5 Curvefit Line IVAT 120 60 57.9 R2 = 0.5782 90.0 60.0 49.3 41.6 20 32.4 Chem IVAT 0.25 0.5 30.0 TV Studio Kempas 0 -0.25 0 0.75 1 1.25 distance (km) Figure 5.10 Rain distribution assuming the IVAT Station to be the center of the rain cell. Simultaneous readings at all stations in which the IVAT Station and the TV Studio Station recorded readings of rain rate of 120 mm/hr are shown in Table 5.6. Using these selected data, the assumption is that the center of the rain cell will be either at the IVAT Station or TV Studio Station. All the readings are averaged and these values are given in Table 5.6. The data are plotted in Figure 5.11. 1.5 76 Table 5.6. Simultaneous rain rate readings at all stations when IVAT and TV Studio stations register 120 mm/hr rain rate. Averaged Rain Rate (mm/hr) for each Rain Gauge Stations Chemical IVAT TV Studio Kempas 77 120 77.5 67.5 64 100 120 70 70.5 110 98.75 68.75 120 mm/hr at IVAT 120 mm/hr at TV Studio Average Table 5.6 shows the simultaneous recordings at all stations, assuming the center of the rain events or rain cells to be at either IVAT or TV Studio. Whenever IVAT station registers a rainfall rate of 120 mm/hr the simultaneously readings at Chemical, TV Studio, and Kempas stations are 77, 77.5, and 67.5 mm/hr, respectively. Similarly, whenever TV Studio station registers 120 mm/hr, the rainfall rate readings at the respective stations are 64, 100, and 70 mm/hr. From the curve fit lines in Figure 5.11, it can be seen that the rain events were highly convective. The center of the rain cell experienced a heavy rainfall while the rain rate tapers rather rapidly toward the edge. The averaged rain cell is about 1.25 km. It is determined here that the rain cell size is between 1.2 km to 1.5 km. (Nor Hisham Khamis et al, 2004). 77 180 IVAT vs. distance Curve fit IVAT TV vs. distance Curve fit TV LAB AVERAGE vs. distance Curve fit AVERAGE 160 140 120 100 80 60 40 20 0 -400 -200 0 200 400 600 800 1000 1200 distance (meters) Figure 5.11. Averaged rain cell size. Table 5.7 gives the percentages of the distribution of the rainfall rates at all the rain gauge stations. Figure 5.12 shows the plots of the all rainfall rate distributions and their respective curve fit lines. The equations for these curve fit lines and values of R0.01 for all rain gauge stations are also shown in Figure 5.12. 78 Table 5.7 Distribution of rainfall rate of RGN UTM 2 Kempas Rain rate Readings % (mm/hr) 0 12 24 30 36 48 60 72 84 90 96 108 120 132 144 150 180 Total 151494 0 0 967 0 0 131 0 0 37 0 0 7 0 0 4 0 152640 TV Studio Readings 99.249 0 0 0.6335 0 0 0.0858 0 0 0.0242 0 0 0.0046 0 0 0.0026 0 100.00 273175 0 0 1527 0 0 223 0 0 82 0 0 27 0 0 5 1 275040 IVAT % Readings 99.321 0 0 0.5552 0 0 0.0811 0 0 0.0298 0 0 0.0098 0 0 0.0018 0.0004 100.00 234468 0 0 1388 0 0 203 0 0 76 0 0 21 0 0 4 0 236160 Chemical % 99.283 0 0 0.5877 0 0 0.0860 0 0 0.0322 0 0 0.0089 0 0 0.0017 0 100.00 Readings 236058 2103 403 0 167 98 78 54 29 0 26 11 9 3 1 0 0 239040 % 98.752 0.8798 0.1686 0 0.0699 0.0410 0.0326 0.0226 0.0121 0 0.0109 0.0046 0.0038 0.0013 0.0004 0 0 100.00 1 0 20 40 60 80 100 120 140 160 Kempas y = 1.7849e-0.0464x R0.01 = 111.7 mm/hr 0.1 Kempas TV IVAT percentage 2 0.01 TV y = 1.7453e-0.0452x R0.01 = 114.2 mm/hr R = 0.9722 2 R = 0.9872 0.001 Chemical IVAT -0.0477x y = 0.6745e y = 1.9841e-0.0466x R0.01 = 88.3 mm/hr R0.01 = 113.5 mm/hr R2 = 0.9563 R2 = 0.9901 0.0001 rain rate (mm/hr) Figure 5.12 RGN UTM 2 Rainfall rate distributions Chemical Chemical Fit Line Kempas Fit Line TV Fit Line IVAT Fit Line 79 Table 5.8 shows the R0.01 values for all the gauge stations in RGN-UTM 2. Also given in Table 5.8 are R0.01 values from the Meteorological Department in Johor (S.K.A. Rahim, 2001) and ITU-R (ITU-R P.837-4, 2003). The lower value of R0.01 for all stations may have resulted from the fact that data were taken during the drier period of the season. Table 5.8 R0.01 Values for RGN-UTM 2 Stations. 5.2 Location Rain rate Kempas Station 111.7 mm/hr TV Lab Station 114.2 mm/hr IVAT Station 113.5 mm/hr Chemical Station 88.3 mm/hr Johor Bahru 120 mm/hr ITU-R 120 mm/hr Radar Data Analysis The analysis of radar data will follow steps to extract information from the radar data. This will also include procedures to process the data to gather the needed information. Most of the processing’s are done using programs in MATLAB and EXCELL. Some MATLAB programs used in this study are enclosed in Appendix 4.1. Important information extracted from radar data are rain rate distribution, rain cell size distribution, and the reduction factor. 80 5.2.1 Rejection of Permanent Echo The first analysis done on the radar data is to check for permanent echo. This is to ensure none of the data collected by the radar contained permanent echo, which will affect the data. The elevation angles for volumetric scanning are 0.5, 1.2, 1.9, 2.7, 3.5, 4.6, 6.0, 7.5, 9.2, 11.0, 13.0, 16.0, 20.0, 25.0, and 32.0 degrees. For permanent echo check, 150 scan of the lower elevation angles, namely 0.5, 1.2, 1.9, 2.7, and 3.5 degrees were checked. Permanent echo is checked by looking for any spot in the radar scanning area that will return scattered energy at each scan, which will indicate permanent echo. It is found that none of the selected elevation angles contained permanent echo. 5.2.2 Preliminary Results From the volumetric scanning of the radar, the data for the 1.20 and 3.50 elevation angles were chosen and the range selected was 65 km. Folders 1 to 10 were selected for this exercise. The radar data gives the rain intensity for every kilometer where there is rain. The rain intensity range is as given in Table 5.9 and is plotted in Figure 5.13. From the data, it can be seen that for a 1-km path lengths, about 41.7% for 1.20 elevation angle and almost 59.85% for 3.50 elevation angle, rain will be occurring at 0.2 mm/hr rate. Very high intensity rainfall rate, which is above 364.6 mm/hr, occurs only 0.0022 % and 0.0002% of the 1-km path lengths for both elevation angles respectively. For 0.01% of rain occurrences, the rain rate is about 192.778 mm/hr, by taking into account data for 1.20 elevation angle. For 3.50 elevation angle, the rain rate is about 78.183 mm/hr. 81 The result also showed that lower elevation angle was able to detect higher rain rate reading. The higher elevation angle was able to detect very low rain rate, which was not detected by the lower elevation angle eventhough these readings were taken almost simultaneously. This is explained by the “updraft” phenomenon during raining events where there is wind blowing upwards and causes small raindrops to evaporate. It is decided that a lower elevation angle that is available (0.50) would be utilized in this study because it will give better reading at high rainfall rate and nearer to the ground giving ground level precipitation, comparable to rain gauge measurement. Table 5.9 Rainfall Intensity for averaged 1-km path. Rain rate (mm/hr) 1.2 degrees 3.5 degrees No. of readings Percentage 0.2 5266864 41.7092 1051604 59.8502 1.0 2464238 19.5147 308820 17.5760 2.1 1687092 13.3604 139243 7.9248 3.2 1566119 12.4024 111385 6.3393 4.9 672429 5.3251 61789 3.5166 7.5 429690 3.4028 42213 2.4025 11.5 276478 2.1895 26017 1.4807 17.8 162252 1.2849 11977 0.6817 27.3 77966 0.6174 3032 0.1726 42.1 14143 0.1120 588 0.0335 64.8 5254 0.0416 227 0.0129 99.9 2419 0.0192 95 0.0054 153.8 1749 0.0139 53 0.0030 236.8 612 0.0048 14 0.0008 364.6 279 0.0022 3 0.0002 Total 12627584 100 No. of readings 1757060 Percentage 100 82 1.2 degrees 3.5 degrees Power (1.2 degrees) rain rate (mm/hr) 1000.0 100.0 10.0 1.0 Power (3.5 degrees) Best-fit line for 1.2 degrees y = 10.492x-0.6321 R2 = 0.9265 Best-fit line for 3.5 degrees y = 7.0261x-0.5232 R2 = 0.9341 0.1 0.00 0.00 0.01 0.10 percentage 1.00 10.00 100.00 Figure 5.13 Rain rate distribution. 5.2.3 Distribution of Rain Rate from Radar Data The first information extracted from radar data is the rain rate distribution. The rain rate in the radar data is the averaged rain rate over a 1-km range-bin size. The rain rate distribution for the 1-km range-bin size is as given in Table 5.9. 83 Table 5.10 Rain rate distribution for range-bin size of 1-km from radar data. 0.5 degrees elevation angle Rain rate (mm/hr) No. of readings Percentage (%) 0 4743844 85.24121 0.2 236698 4.25318 1 124172 2.23122 2.1 114667 2.06043 3.2 125677 2.25827 4.9 65680 1.18019 7.5 54020 0.97068 11.5 44683 0.80290 17.8 33062 0.59409 27.3 17019 0.30581 42.1 2818 0.05064 64.8 1427 0.02564 99.9 727 0.01306 153.8 378 0.00679 236.8 203 0.00365 364.6 125 0.00225 Total 5565200 100 The percentage is calculated by, dividing the number of events for a particular rain rate by the total number of events and then multiplying them by 100%. Using the Curve Fitting Toolbox available in Matlab, the best-fit equation for the data in Table 5.10 is obtained. This window is invoked using the “cftool” command in Matlab’s Command Window. The procedure followed for curve fitting 84 in this study is given in Appendix 4.2. Figure 5.14 shows the plots data in Table 5.9 and the curvefit line for the rain rate distribution. R vs. P powerfit line rain rate (mm/hr) 10 10 10 2 1 0 powerfit1(x) = a*xb Coefficients (with 95% confidence bounds): a= 4.382 (2.959, 5.806) b = -0.7204 (-0.777, -0.6637) R2 = 0.9950 10 -2 10 -1 10 percentage 0 10 1 Figure 5.14 The plots of original data and the curve-fit line. From the analysis of the curvefit line, the rain rate for 0.01 % of the time is 120.907 mm/hr (as given in Appendix 4.2). This value will be used in the mathematical model that is used to obtain the reduction factor. The rain rate distribution Rp from radar data is given by (see Appendix 4.2) 85 Rp = 4.382 p –0.7204 (5.1) where p is the percentage of time of interest. This rain rate distribution can be used as a rain rate prediction model to predict the rain rate at the percentage of interest. 5.2.4 Determination of Rain Cell Size from Radar Data For the determination of rain cell size, only rain cells that occurred inside the virtual links were considered. If a cell has a value in the 32nd range bin or in the 51st range bin, it will not be considered. Thus, the maximum rain cell size considered is 18 km. The result of this analysis is shown in Table 5.11 and Figure 5.15. The complete tabulation for rain cell size according to data files is given in Appendix 4.3. 86 Table 5.11 Rain Cell Size Distributions. Rain cell diameter (km) Total # of rain cells Percentage 1 146661 70.507 2 30604 14.713 3 14811 7.120 4 6745 3.243 5 4276 2.056 6 1585 0.762 7 1209 0.581 8 738 0.355 9 393 0.189 10 213 0.102 11 157 0.075 12 178 0.086 13 118 0.057 14 116 0.056 15 82 0.039 16 57 0.027 17 41 0.020 18 24 0.012 87 D vs. P curve-fit line 1 diameter (km) 10 b powerfit(x) = a*x Coefficients (with 95% confidence bounds): a= 5.803 (5.316, 6.291) b = -0.2731 (-0.2987, -0.2474) 2 R = 0.9830 0 10 -1 10 0 10 1 10 percentage Figure 5.15 Rain Cell Size Distribution. From Table 5.10 and Figure 5.15, it can be seen that about 70% of rain cells are of 1-km in diameter. This value is useful when mitigation technique such as site diversity is to be employed. In Figure 5.15, the formula for the rain cell size distribution is D = 5.803 p – 0.2731 (5.2) 88 where D is the diameter of the rain cell in km, and p is the percentage of interest. For example, the diameter of rain cell size for 99.99 % of the time would be 1.65 km. This means that eventhough the rain cell for R0.01 of 120 mm/hr (from rain gauge data) is about 1.2 km; the rain cell size for 99.99 % of the time is 1.65 km (from radar data). Thus, the rain cell size information from both types of data agreed very well. This information is useful for some communication system that may have the requirement of knowing rain cell size distribution. 5.2.5 Rain Attenuation Measurements in UTM Two systems of microwave links operating at two different frequencies were utilized for attenuation measurements in UTM (Karim, 2001). Binariang provided a system operating at 7 GHz, and the other was provided by Digi Communications operating at 15 GHz. The Binariang system utilizes 4 ft radomeless parabolic reflectors while the Digi system employs parabolic antenna of 0.6 meters in diameter with radomes. The Binariang system consisted of three links and measurement period was from 15 January, 1999 up to 14 January, 2000. The Digi system has six links and measurement period was from 1 December 1998 to 30 November 1999. Specifications for both systems are given in Table 5.12 and Table 5.13. Table 5.12 Specifications for Binariang system links. Hop Frequency Hop length Polarization Site A Site B (GHz) (km) UTM Ulu Choh 7.807 11.87 Vertical UTM Tmn Perling 7.491 10.26 Vertical UTM Senai 7.491 5.57 Vertical 89 Table 5.13 Specifications for Digi system links. Location Frequency (GHz) Hop length (km) Polarization Butterworth 14.81775 11.33 Vertical Johor Bahru 14.83875 5.83 Vertical Temerloh 14.83175 5.36 Vertical Jitra 15.32525 4.85 Vertical Kuala Lumpur 14.8195 3.96 Vertical Taiping 14.83175 3.48 Vertical 5.3 Deducing the Reduction Factor from Radar Data Formulation of the reduction factor from radar data followed the basis adopted by Goddard and Thurai (Goddard et al, 1997). Attenuation of radio waves through rain is given by Equation 3.3 in section 3.2. Rearranging for the reduction factor, r (t) = A d (t) γs L (5.3) where r(t) is the reduction factor for a given time percentage, Ad(t)is the radar derived total attenuation at corresponding time percentage (dB), γs is the specific attenuation at same time percentage from available rain gauge data (dB), and L is total path length (km). 90 The next step is to find Ad(t). In this study, attenuation at three different frequencies was considered. These frequencies are 7, 10, and 15 GHz. Using the values recommended by ITU-R (ITU-R P.838-1) as in Appendix 4.5 for the k and α coefficients at 7, 10, and 15 GHz and Equation 3.5; values for the specific attenuation at the respective frequencies are calculated. For 7 GHz, γs = 0.00265 (120.907)1.312 = 1.4303 dB (5.4a) γs = 0.00887 (120.907)1.264 = 3.8031 dB (5.4b) For 10 GHz, And for 15 GHz, γs = 3.347e-2 (120.907)1.128 = 7.4758 dB (5.4c) Thus, the reduction factor will be derived using Equation 5.3 for all the frequencies chosen. The path lengths chosen are 1, 2, 3, 4, 5, 6, 7, 8, 9, and 10 km. The same radial lines as for the cell size distribution analysis are used. For a 1-km path length, it is assumed that there will be 20 virtual path links within the range bins. For a 2-km path link, the first one will be from range bin 31 to 32. The next link will be from 32 to 33, and so on. Thus, there will be 19 virtual links within the range chosen. This is repeated for the rest of virtual links where there will be two 10-km path links for each radial line. This will produce a reduction factor for path length of10 km and 91 using curve-fitting tool available in MATLAB and EXCELL, can be interpolated and extended for other path lengths. The distribution of the attenuation for 1-km path length for links operating at 7 GHz is given in Table 5.14. Table 5.14 Distribution of attenuation for 1-km links operating at 7 GHz. Attenuation (dB) 0 0.082323 Total Number of readings for each attenuation value Percent 4743844 85.241 818496 14.707 0.631 1427 0.025641 1.1135 727 0.013063 1.9612 378 0.0067922 3.4549 203 0.0036477 6.0862 125 0.0022461 5565200 100 The values obtained in Table 5.14 uses the same data as in Table 5.9. Attenuation values are calculated using Equation 3.5. Attenuation values are very close to one another, so they were grouped in classes of 0.5 dB each. Values that fall within each class are averaged and the number of occurrences or readings are taken. Thus, the number of columns in Table 5.13 is less than Table 4.9 due to the grouping of attenuation values into classes. The total data is still the same. The best-fit curve for the data in Table 5.14 with attenuation values as the yaxis and percentage as the x-axis is as shown in Figure 5.16. 92 Attenuation vs. Percentage Attenuation Distribution Fit Line distribution of attenuation (dB) 10 10 10 10 0 -1 -2 -3 fittedmodel1(x) = a*xb Coefficients (with 95% confidence bounds): a = 0.02367 (-0.01339, 0.06072) b = -0.8873 (-1.133, -0.6411) 0.01 ==> 1.40835 R2 = 0.9815 10 -2 10 -1 10 percentage 0 10 1 Figure 5.16 The best-fit curve for 1-km path links attenuation distribution. The important information or parameter extracted from the best-fit curve of Figure 5.16 is the attenuation value for 0.01% of the time for the 1-km path links operating at 7 GHz. From the analysis, it is found out that the attenuation for 0.01% of the time for a 1-km path link operating at 7 GHz is 1.40835 dB. As a comparison, the specific attenuation calculated using ITU-R parameters is 1.4163 dB (with R0.01 = 120 mm/hr). % difference = [(1.4163 – 1.40835)/1.4163] * 100% = 0.55% 93 Thus the value deduced from radar data is very close to the value calculated using ITU-R parameters. The attenuation distribution tables and the best-fit curves for the attenuation distributions of all 1 to 10-km path links operating at 7, 10, and 15 GHz are given in Appendix 4.4. As can be seen from Appendix 4.4, this is a tedious and lengthy procedure, which requires careful attention and determination when carrying out the task. This is true not only when calculating the values and running the curve fitting exercise, but also when compiling the results. The attenuation values for 0.01% of the time for each path links of 1 to 10 km lengths; operating at 7, 10, and 15 GHz are extracted from the best-fit lines. These values are summarized in Table 5.15 and plotted as shown in Figure 5.17. Table 5.15 Attenuation (dB) for 0.01% of the time; at 7, 10, and 15 GHz for path lengths of 1 to 10-km. Frequency Path length 7 GHz 10 GHz 15 GHz 1-km 1.40835 3.7641 7.44543 2-km 2.62741 4.22063 6.4967 3-km 2.91245 5.13745 8.80735 4-km 3.17858 5.57111 9.04752 5-km 3.33466 5.84443 9.297 6-km 3.49907 5.95246 9.492 7-km 3.66776 6.18554 9.91478 8-km 3.84256 6.4482 10.3971 9-km 3.99616 7.23961 10.7589 10-km 4.14 6.74703 11.2355 94 16 14 attenuation (dB) 12 Attenuation at 7GHz fit A7GHz Attenuation at 10GHz fit A10GHz Attenuation at 15GHz fit A15G 10 8 6 4 2 0 0 2 4 6 8 path length (km) 10 12 Figure 5.17 Attenuation for 0.01% of the time at 7, 10, and 15 GHz. Using equation 4.3, the values obtained in equation 4.4, and the values in Table 5.15; the reduction factors for path lengths of 1 to 10-km at frequencies of 7, 10, and 15 GHz are calculated. The reduction factor values are tabulated in Table 5.16 and plotted in Figure 5.18. 95 Table 5.16 Reduction factor (r) values for 1 to 10-km path lengths at frequencies of 7, 10, and 15 GHz for 0.01% of the time. Frequency Path length (km) 7 GHz 10 GHz 15 GHz 1 0.98465 0.98975 0.99594 2 0.91848 0.55489 0.43452 3 0.67875 0.45029 0.3927 4 0.55558 0.36622 0.30256 5 0.46629 0.30735 0.24872 6 0.40773 0.26086 0.21162 7 0.36633 0.23235 0.18946 8 0.33582 0.21194 0.17385 9 0.31044 0.21151 0.15991 10 0.28945 0.17741 0.15029 r7 vs. L curvefit r7 r10 vs. L curvefit r10 r15 vs. L curvefit r15 reduction factor (r) 1 0.8 0.6 0.4 0.2 0 0 2 4 6 8 10 12 14 path length (km) Figure 5.18 Reduction factor (r) plots for 1 to 10-km path lengths at frequencies of 7, 10, and 15 GHz for 0.01% of the time. 96 The best-fit lines for each reduction factor plots in Figure 5.18 follows the power law relationship, given by f(x) = a*xb The a and b values for all the best fit lines are summarized in Table 5.17, and are plotted in Figure 5.19 and Figure 5.20, respectively. Table 5.17 a & b values for best fit lines of reduction factors at 7, 10, and 15 GHz. Frequency a b 7 GHz 1.08 -0.5108 10 GHz 0.9798 -0.7319 15 GHz 0.9654 -0.8629 97 3 a vs. F Curvefit line for a coefficients 2.5 b afit(x) = a*x Coefficients (with 95% confidence bounds): a= 1.455 (-1.474, 4.384) b = -0.1588 (-1.03, 0.7128) coefficient a 2 1.5 2 R = 0.8237 1 0.5 2 4 6 8 10 12 14 16 18 20 Frequency (GHz) Figure 5.19 Best fit line for coefficient a. 0.4 b vs. F curvefit line for b coefficients 0.2 b coefficient 0 bfit(x) = a*xb Coefficients (with 95% confidence bounds): a= -0.159 (-1.003, 0.6845) b= 0.6322 (-1.527, 2.791) -0.2 R2 = 0.9376 -0.4 -0.6 -0.8 -1 -1.2 2 4 6 8 10 12 14 16 Frequency (GHz) Figure 5.20 Best fit line for coefficient b. 18 20 98 From Figure 5.18, Figure 5.19, and Figure 5.20, the reduction factor can be formulated as follows:- reduction factor, r = a L b (5.5) Where a = 1.455 F –0.1588 b = –0.159 F 0.6322 L = path length (km), and F = frequency (GHz). The a and b values for the reduction factor r at 7, 10, and 15 GHz are given in Table 5.18 below. Table 5.18 a and b values at 7, 10, and 15 GHz. Frequency a b 7 GHz 1.078 -0.544 10 GHz 1.009 -0.682 15 GHz 0.946 -0.881 99 The proposed Malaysia reduction factor is given in Table 5.19 for path lengths of 1 to 10-km at 7, 10, and 15 GHz. The proposed reduction factor is compared with other models in Table 5.20, Table 5.21, and Table 5.22 at 7, 10, and 15 GHz, respective. The graphs of all the tables for all the frequencies are plotted in Figures 5.21, Figures 5.22 and Figure 5.23. Table 5.19 The reduction factor (r) for the proposed Malaysia model. path length (km) value of r 7 GHz 10 GHz 15 GHz 1 1.078 1.009 0.946 2 0.739364 0.62891 0.513669 3 0.593014 0.476974 0.359375 4 0.507105 0.392 0.278918 5 0.449137 0.336662 0.229139 6 0.406728 0.297298 0.195137 7 0.374012 0.26763 0.170357 8 0.347807 0.244334 0.15145 9 0.32622 0.225475 0.136522 10 0.308048 0.209841 0.12442 100 path length (km) Lin Moupfouma CETUC Improved CETUC Goddard ITU-R Singapore DAH Proposed Model Table 5.20 Comparison of r from various models at 7 GHz. 1 2 3 4 5 6 7 8 9 10 0.978 0.958 0.938 0.919 0.901 0.883 0.867 0.850 0.835 0.820 0.971 0.943 0.917 0.892 0.868 0.846 0.824 0.804 0.784 0.765 0.914 0.841 0.779 0.725 0.679 0.638 0.601 0.569 0.540 0.514 1.021 0.692 0.591 0.538 0.505 0.481 0.463 0.448 0.437 0.426 0.830 0.680 0.595 0.536 0.490 0.454 0.423 0.396 0.373 0.353 0.886 0.796 0.722 0.661 0.610 0.566 0.527 0.494 0.465 0.438 0.850 0.753 0.679 0.620 0.572 0.531 0.497 0.467 0.440 0.417 0.978 0.891 0.814 0.757 0.712 0.676 0.646 0.621 0.598 0.579 1.078 0.739 0.593 0.507 0.449 0.407 0.374 0.348 0.326 0.308 1.6 1.4 reduvtion factor, r 1.2 Lin Moupfouma 1.0 CETUC Improved CETUC 0.8 Goddard ITU-R Singapore 0.6 DAH Proposed Model 0.4 0.2 0.0 1 2 3 4 5 6 7 8 9 10 path length (km) Figure 5.21 Plots of all the reduction factor models at 7 GHz. 101 Path length (km) Lin Moupfouma CETUC Improved CETUC Goddard ITU-R Singapore DAH Proposed Model Table 5.21 Comparison of r from various models at 10 GHz. 1 2 3 4 5 6 7 8 9 10 0.978 0.957 0.937 0.918 0.899 0.882 0.865 0.848 0.833 0.817 0.971 0.943 0.917 0.891 0.867 0.844 0.822 0.801 0.781 0.762 0.913 0.840 0.778 0.724 0.678 0.637 0.600 0.568 0.539 0.513 1.021 0.692 0.591 0.538 0.505 0.481 0.463 0.448 0.437 0.426 0.829 0.679 0.594 0.534 0.489 0.452 0.421 0.395 0.372 0.351 0.886 0.796 0.722 0.661 0.610 0.566 0.527 0.494 0.465 0.439 0.851 0.754 0.680 0.621 0.573 0.532 0.498 0.468 0.441 0.418 0.869 0.767 0.690 0.634 0.592 0.558 0.530 0.507 0.486 0.469 1.009 0.629 0.477 0.392 0.337 0.297 0.268 0.244 0.225 0.210 1.6 1.4 reduction factor, r 1.2 Lin Moupfouma 1 CETUC Improved CETUC 0.8 Goddard ITU-R Singapore 0.6 DAH Proposed Model 0.4 0.2 0 1 2 3 4 5 6 7 8 9 10 path length (km) Figure 5.22 Plots of all the reduction factor models at 10 GHz. 102 path length (km) Lin Moupfouma CETUC Improved CETUC Goddard ITU-R Singapore DAH Proposed Model Table 5.22 Comparison of r from various models at 15 GHz. 1 2 3 4 5 6 7 8 9 10 0.978 0.958 0.938 0.919 0.901 0.883 0.867 0.850 0.835 0.820 0.971 0.943 0.916 0.890 0.865 0.841 0.817 0.795 0.774 0.753 0.914 0.841 0.779 0.725 0.679 0.638 0.601 0.569 0.540 0.514 1.021 0.692 0.591 0.538 0.505 0.481 0.463 0.448 0.437 0.426 0.830 0.680 0.595 0.536 0.490 0.454 0.423 0.396 0.373 0.353 0.886 0.796 0.722 0.661 0.610 0.566 0.527 0.494 0.465 0.438 0.850 0.753 0.679 0.620 0.572 0.531 0.497 0.467 0.440 0.417 1.098 1.035 0.965 0.910 0.865 0.828 0.797 0.771 0.747 0.726 0.946 0.514 0.359 0.279 0.229 0.195 0.170 0.151 0.137 0.124 1.6 1.4 reduction factor, r 1.2 Lin Moupfouma 1.0 CETUC Improved CETUC 0.8 Goddard ITU-R Singapore 0.6 DAH Proposed Model 0.4 0.2 0.0 1 2 3 4 5 6 7 8 9 10 path length (km) Figure 5.23 Plots of all the reduction factor models at 15 GHz. 103 At 7 GHz, the proposed reduction is similar to the Improved CETUC model and generally follows the same trend as other models, namely the ITU-R, Singapore, and Goddard models, as shown in Figure 5.21. The values for the proposed model are nearer to the Goddard and Improved CETUC models, albeit a little lower. At higher frequencies, especially at 15 GHz, the proposed model shows a much lower reduction factor values. This means that the proposed model will predict a lower attenuation value at higher frequencies. However, models that are dependant on frequency are the Moupfouma and DAH models, beside the proposed model. It has been shown in earlier sections that attenuation due to rain will increase with frequency. The Lin and Moupfouma models consistently exhibit higher values of r as compared to other models. At higher frequency (15 GHz), the DAH models begin to exhibits values that similar to Lin and Moupfouma models. Lin and DAH models were developed using data from the US, while Moupfouma used data from several places. The CETUC, ITU-R, Singapore, Improved CETUC, and Goddard models tend to have values of r, which are close to one another. The proposed model values are always close to the Goddard model in which both uses radar data. The Singapore, Improved CETUC, and the proposed models have values that close to one another, especially at lower frequency. All are done in tropical region, which may explain this. The Improved CETUC has values that are always lower than the older CETUC model. Thus an improvement to the CETUC model will give lower value in the prediction of attenuation. Most models were developed using higher older value of R0.01 from ITU-R, thus it is expected that most models will have lower r values that approaches the values of the proposed model if they were developed using the new value of R0.01 from ITU-R. In some works that have been done in UTM, Islam et al (2000) concluded that the ITU-R model does not reflect the measured attenuation done at several locations in Malaysia. Moupfouma and Lin models were closer to the measured data, but the Moupfouma model underestimated attenuation in some places where 104 measurements were done. Islam (2000) proposed some modifications to the Moupfouma model to suit the measured attenuation data. The Modified Moupfouma Model is as follows. r = 1/[1 + 0.03 α (p/0.01)-βLm] (5.6) where m = 1 + 1.4 x 10-4 F1.76 logeL. α = 2.6 β = 0.34 } L < 3 km } 0.001 ≤ p ≤ 0.01 0.001 ≤ p ≤ 0.02 α = 1.0 β = 0.45 β = 0.6 α = 2.8 β = 0.26 } 0.01 < p ≤ 0.02 } 3 km ≤ L ≤ 6 km 6 km < L ≤ 11.3 km 0.001 ≤ p ≤ 0.02 The performance of the proposed reduction factor is compared with the other models. No actual measurements were done in this study. As a result, attenuation measurement work done by other researcher (Karim, 2000), had to be used eventhough some results from the attenuation measurement were rather questionable. These results are shown in Table 5.23. 105 15 15 5.57 10.26 11.87 3.48 3.96 4.85 5.36 5.83 11.33 125 147 133 107 114 125 125 125 125 B'worth 15 Bahru Temerloh 15 Johor Jitra 15 7 Lumpur 15 7 Kuala Taiping Freq (GHz) Path length (km) R0.01 (mm/hr) Specific attenuation (dB/km) Measured attenuation (dB) Perling Link Tmn 7 Senai Ulu Choh Table 5.23 Comparison of predicted attenuations (dB), with measurements. 1.494 1.494 1.494 9.319 8.325 6.513 6.996 7.762 7.762 8.010 8.220 10.540 29.210 30.140 28.530 29.910 33.400 40.970 7.344 12.311 13.816 29.159 29.788 29.174 33.912 39.718 69.203 Lin Moupfouma 7.118 11.657 12.987 29.331 29.417 27.506 32.201 38.371 64.834 Modified 7.115 8.139 8.76 29.284 29.373 27.418 32.099 38.238 43.004 Moupfouma 5.659 8.640 9.527 18.351 18.557 18.031 20.234 22.915 35.056 DAH CETUC 5.405 7.686 8.247 23.670 23.586 22.088 25.118 28.914 41.915 S'pore 5.670 8.466 9.216 23.472 24.013 23.487 26.692 30.423 46.620 ITU-R 4.86 6.625 7.038 22.445 21.89 I-CETUC 4.024 6.407 7.184 17.030 17.187 16.732 18.912 21.620 35.976 Goddard 3.836 5.195 5.505 17.219 17.150 16.326 18.206 20.438 28.096 Proposed Model 3.492 4.613 4.930 10.232 9.282 7.440 8.087 9.062 9.808 19.5 22.56 25.93 35.91 Table 5.23 shows that the proposed model predicts attenuation values that are lower than other models. It also gives attenuation values that are much lower than the measured attenuation. At 7 GHz, the proposed model is close to the Goddard, Improved CETUC, and ITU-R models. At higher frequency, the proposed model gives a much lower attenuation values than the other models. However, only the Moupfouma and DAH models are dependant on frequency. 106 It must be mentioned here that the measurements done using both Binariang and Digi systems suffers from wet antenna loss and actual attenuation due to rain might be lower (S.K.A.Rahim, 2001). S.K.A.Rahim noted that available attenuation prediction models underestimate the measured attenuation. Spray tests conducted on a microwave system installed in UTM, Skudai campus (Md Rafiqul Islam, 2000) showed that wet antenna loss can be as high as 8 dB at 38 GHz, and 5 dB at 23 GHz. Wet antenna loss measurements cannot be done on the 7 and 15 GHz due to the height of the communication towers. Study in Singapore has shown that wet antenna loss can be as high as 20 dB, and is also affected by wind (Gang et al, 2000). This wind effect is not taken into consideration in the attenuation measurement experiment. Moreover, the wet antenna loss occurs not only during rainfall but also several minutes after the rainfall. This adds more loss to the measured received power. There are also other issues that may effects the measurements. These are discussed here. The path length of the Taman Perling link is almost twice the path length of the Senai link but their measured attenuations have only 0.21 dB difference. Karim suggested that this is due to an equal effective rain cell diameter, which is also to suggest an equal effective path length for both links. This is in contradiction with all the reduction models that have been discussed earlier. All of them have shown that the effective path length is directly proportional to the actual path link as shown in Figure 5.21 – Figure 5.23. In another instance, the length of the Ulu Choh link is comparable to the Taman Perling link but their difference in measured attenuation is 2.32 dB. These results do not tally with long-term measurements taken near each other. For example, rain gauge networks data were collected at different time period. The locations of both rain gauge networks were close to each other. Results from both rain gauge networks however showed similar patterns in rain rate distributions, and rain cell size distributions. 107 The measurement values of attenuation for 0.01 % of the time were rather high for most links. For example, the measured attenuation for Senai link is 8.01 dB. If rain is constant over the link then the total attenuation will be 8.32 dB, for constant specific attenuation along the path. This inferred a large cell of diameter around 5 km at R0.01. Analysis of rain gauge data in this study has shown that for R0.01; the rain cell is about 1.2 km. Thus, it is very unlikely that heavy rain is constant over all the links. The prediction of attenuation values from the Lin and Moupfouma models are very close to the measured values. These models were developed using data from the temperate region. However, it has been widely accepted that tropical region suffers heavier rain than temperate region. The closeness of the attenuation values seemed to suggest otherwise. Thus, these two models are definitely not suitable for the tropical region. They may not even applicable anymore, as newer, more accurate data are available. These two models are discussed here because they can be considered as “classical” models. Moupfouma himself suggested several models for rain attenuation prediction (Chebil, 1997). In a later publication, Moupfouma even used a reduction factor model that was developed by other researcher instead of using one of his own (Moupfouma, 1997). Is not surprising that the Modified Moupfouma model is also close to the measured attenuation values since it was modified according to the measured values. However, due to the issues on the measured data; which makes it questionable, the Modified Moupfouma model also is questionable. The Modified Moupfouma model also suggested different parameters values for different path lengths, which is uncommon with other models. This also points back to the questionability of the measured attenuation values. The DAH model is another model that utilized data from a temperate region. Moreover, it was formulated for slant path attenuation prediction. At high frequency, the predicted attenuation values seemed high since they were higher from models developed in the tropics. Thus, this model is not suitable for terrestrial attenuation prediction in the tropical region. 108 The CETUC and Improved CETUC models were selected because they were developed in Brazil, a tropical region similar to Malaysia. The Improved CETUC showed similar pattern to the proposed model, especially at lower frequency. The Improved CETUC demonstrated that with a larger base and more accurate data, the actual attenuation is lower than that was predicted from previous models. Both CETUC models are, however, insensitive to frequency. This explains the departure of predicted values from the proposed model at higher frequency. Both models also used old data from the ITU-R. Using newer values from ITU-R and incorporating dependence on frequency might make the Improved CETUC model to be much closer to the proposed model. The Singapore model was developed using rain gauge and satellite beacon data. The rain gauges, however, were place quite far apart. The distances were 5.3, 4.9, and 7.1 km apart. It also assumed that rainfall rate is constant over 6 km. This study has shown that rain cell diameters are much smaller than that. Thus, the arrangement of the rain gauge does not give an accurate representation of rain cells across a terrestrial link. As a result, the Singapore model does not predict the attenuation accurately. The ITU-R model was developed from data collected around the world. Studies (Ajayi et al, 1988; Juy et al, 1990; Yagasena et al, 1995) have shown that the model is not applicable everywhere. Characteristics of rain are rather localized and local studies are needed to develop prediction models suitable for the local environment. This has been a motivation for this study. Local studies (Karim, 2000; Islam, 2000) also concluded that the ITU-R model is not suitable for Malaysia. The Goddard and Thurai model is a model developed from radar data. In fact, this study also used the virtual link concept as used by Goddard and Thurai. The values predicted by the proposed model is very close to the values predicted by the Goddard and Thurai model, especially at the lower frequencies. The Goddard and 109 Thurai model however, is insensitive to frequency. Thus, at higher frequencies, the values differ. By incorporating frequency into the equation, the difference may be minimized. To really test the proposed model, measurements must be made for several path lengths and at different frequencies. An elaborate scheme of rain gauge network is also needed. The wet antenna loss must also be taken into consideration. No doubt, this will be an enormous task. Notwithstanding, the proposed model can be used to give an indication of attenuation due to rain, as it was developed using the local rain data. The proposed model not only depends on the length of the link, but also on frequency as well (Assis, 1990; Dissanayake et al, 1997). CHAPTER 6 CONCLUSION AND FUTURE STUDIES 6.1 Conclusion This work is done primarily using weather radar data obtained from the Meteorological Department of Malaysia. Data from weather radar have been utilized in many other studies. The biggest advantage of radar data is the large amount of data that is available in a short period of time. Radar data is also capable of providing areal precipitation where it is almost impossible to do using rain gauges. The previous chapter discussed the inadequacies of available reduction factor models. The proposed reduction model can calculated using Equation 5.5, and is as shown in Figure 6.1 for 7, 10, and 15 GHz. The predicted attenuation due to rain for 0.01 % of the time using the proposed reduction factor is as shown in Figure 6.2 111 reduction factor, r 1.2 r (7 GH) r (10 GHz) r (15 GHz) 1 0.8 0.6 0.4 0.2 0 1 2 3 4 5 6 7 8 9 10 path length (km) attenuation (dB) Figure 6.1 Proposed reduction factor r. 10 9 8 7 6 5 4 3 2 1 0 A (7GHz) A (10 GHz) A (15 GHz) 1 2 3 4 5 6 7 path length (km) 8 9 10 Figure 6.2 Predicted attenuation due to rain for 0.01 % of the time. Figure 6.2 shows that there will be higher attenuation at higher frequencies as expected. It also shows that as the path link gets longer, the amount of attenuation due to rain will increase at a slower rate. This is due to the fact that rain cells will not cover the whole path link. Figure 6.2 also shows that multiple rain cells are taken into consideration in the proposed reduction factor. 112 The proposed reduction factor can be used to calculate the attenuation due to rain. The needed parameters are the length of the link (in km) and the frequency used (in GHz). This is useful microwave link system planning and link budget estimation. Below are the procedures for calculating attenuation due to rain using the proposed reduction factor. First the specific attenuation is calculated using the Equation 3.5. This specific attenuation assumes that rain is constant over a link of 1-km. Parameters k and α in Equation 3.5 are as recommended by ITU-R. The reduction factor r is then calculated using the Equation 5.5. The total attenuation is then calculated using Equation 3.3. This value predicts the attenuation due to rain for 0.01 % of the time. The reliability and the accuracy of the proposed reduction factor model stemmed from the fact that meteorological radar data are highly accurate and reliable. This is demonstrated by the fact that R0.01 extracted from the radar data is similar to the ITU-R, and the long term rain gauge data of Johor Bahru. From the distribution of the rain rate obtained from the radar data, a prediction model for rain rate distribution is also suggested. Equation 5.1 gives the rain rate model from radar data. This model will predict the rain rate, Rp for p percentage of the time. The model for rain rate distribution is given in Figure 5.14. Rain cell diameter will give a good indication for system planners doing attenuation mitigation technique such as site diversity application. Mitigation technique is important when high reliability of microwave link is needed. If 113 attenuation exceeds the link budget of a link, then an alternative technique is surely required to overcome the attenuation problem. This is especially useful for satellite link where it’s application has become more prominent as the communication community moves into higher spectrum of available bandwidth. A model for the distribution of rain cell size obtained from radar data is given by the best-fit curve in Figure 5.15, and is given by Equation 5.2. From this equation, most of the rain cells will have a diameter of 1.65 km. This agrees very well with the rain gauge data. This study also utilized two rain gauge networks that were set-up in UTM, Skudai campus. Analysis of radar data produced the reduction factor; which is the main intention of this study, and prediction models for rain rate distribution, and rain cell size distribution. From rain gauge networks data, rain cell at several rain rates and the profile of rain intensity inside a rain cell have been determined. Results from RGN-UTM 1 and RGN-UTM 2 showed that most rain cells are very convective. Plots of radar data shows that multiple rain cells may exist in a raining region. Widespread rain has low intensity but the rate may be higher during thunderstorms. The rain cell diameter for 120, 90, and 60, mm/hr from RGN-UTM 1 is about 1.2 km. The averaged rain rate cell diameter is also about 1.2 km. For RGN-UTM 2, the average cell diameter is about 1.5 km. From radar data, more than 70% of rain cells have a diameter of about 1 km. Thus, it can be said that most rain cells in Malaysia will have a diameter of around 1.5 km. These rain cells are very convective in nature. 114 One of the most challenging tasks in this study is to determine the proper procedure in formulating the reduction factor. Eventhough there are many reduction factors that were proposed in the literature, none of them really give in detail how the reduction factors were obtained. Also, due to the large amount of data, a very long time is needed to process all the data. Eventhough a personal computer is adequate to do the processing, care must be exercise not to overload the memory of the computer. This means that the data have to be divided into several files and processed separately. A more powerful computing facilities will surely expedite in the processing of data and other works related to this study, such as larger data storage. 6.2 Future Studies Eventhough ground based weather radar is widely used in many studies; Doppler radar and polarimetric radar have been developed and used (Hornbostel et al, 1997). These types of radars allow the determination of the shape, distribution, types of hydrometeors, and rain rate with more precision. Together with other types of data measurements such as rain gauge and beacon receiver, these data can be compared and checked out against each other (Hornbostel et al, 1995). This will raise the degree of confidence in outcomes of researches. Radars of these kinds are not readily available in Malaysia. It is hoped that as a continuation to this study, using experience gained and expertise available, a radar system can be built here in UTM that employs such techniques so that the results found in this study can be refined. However, continuing studies are utilizing the TRMM radar project (Oki et al, 2000; Rincon et al, 2001). Furthermore, due to increasing popularity microwave communication using satellite, vertical path reduction factor must also be studied (Bandera et al, 1999), as a continuation of this 115 study. Other factors that are to be considered are parametrization of dsd, and the relationship between dsd and rain rate. Another important aspect of future studies is to have a longer period of measuring time. Also, different methods of measurements must be done concurrently to enable cross checking of data obtained. As new instruments and systems are being installed, greater accuracy and more reliable data will be available. REFERENCES Ajayi, G.O., Ezekpo, S.U.B. (1988). Development of Climatic Maps of Rainfall Rate and Attenuation for Microwave Applications in Nigeria. The Nigerian Engineer, vol. 23, no. 4, 1988: 13-30. Allnutt, J.E. (1989). Satellite-to-ground radiowave propagation, Theory, practice and system impact at frequencies above 1 GHz, Peter Perigrinus Ltd., London, UK. Assis, M.S. (1990). Path length reduction factor for tropical regions. Proceedings of URSI Commission F Open Symposium on Regional Factors in predicting radiowave attenuation due to rain. Rio de Janeiro, Brazil, 3-7 December, 1990: 69-71. Atlas, D., Rosenfeld, D. and Jameson, A.R. (1995) EVOLUTION OF RADAR RAINFALLMEASUREMENTS: STEPS AND MIS-STEPS. III International Symposium on Weather Radars, São Paulo, Brazil, August 1995. Aydin, K.; Daisley, S.E.A. (2002). Relationships between rainfall rate and 35-GHz attenuation and differential attenuation: modeling the effects of raindrop size distribution, canting, and oscillation. Geoscience and Remote Sensing, IEEE Transactions on, Volume: 40 Issue: 11, Nov. 2002: 2343 –2352. 117 Bandera, J.; Papatsoris, A.D.; Watson, P.A.; Tozer, T.C.; Tan, J.; Goddard, J.W. (1999). Vertical path reduction factor for high elevation communication systems. Electronics Letters, Volume: 35 Issue: 18, 2 Sept. 1999: pp. 1584 – 1585 Barbara, A.K.; Devi, M.; Timothy, K.I.; Sharma, S. (1993). Microwave propagation in relation to atmospheric parameters over different terrains of Assam Valley. Geoscience and Remote Sensing Symposium, 1993. IGARSS '93. 'Better Understanding of Earth Environment'., International , vol.1, 18-21 Aug. 1993: 261 –263. Battan, L.J. (1973) Radar Observation of the Atmosphere, The University of Chicago Press. Bodtmann, W.F. & Ruthroff, C.L. (1974). Radio attenuation on short radio paths, theory, experiment and Design, Bell Sys. Tech. Journal, 1974, 53(7): 13291349. Chebil, J. (1997). Rain Rate and Attenuation Distribution for Microwave Propagation Study in Malaysia. Universiti Teknologi Malaysia: Ph.D. Thesis. Crane, R.K. (1975). Attenuation Due to Rain - A Mini-review. IEEE Trans. On Antennas and Propagation, vol. AP-23, no. 5, September, 1975:750-752. Crane, Robert K.(1980). Prediction of attenuation by rain. IEEE Trans. on Com., 1980, Vol. COM-28, No. 9: 1717-1733. Daisley, S.E.A.; Aydin, K. (2002). Relationships between rainfall rate, attenuation, and reflectivity at 14 and 35 GHz frequencies. Geoscience and Remote Sensing Symposium, 2002. IGARSS '02. 2002 IEEE International, Volume: 4 , 24-28 June 2002: 2009 –2011. 118 Din, J. (1997) Influence of Rain Drop Size Distribution to the Communication Systems Operating at Frequencies Above 10 GHz in Malaysia, PhD Thesis, Universiti Teknologi Malaysia. Dutton, E.J and H.T. Dougherty. (1979). Year-to-year variability of rainfall for microwave applications in the USA. IEEE Trans. on Com, 1979, Vol.COM-27, No. 5, pp. 829-832. Eastment, J.D.; Ladd, D.N.; Thurai, M. (1996). Rain radar measurements in Papua New Guinea and their implications for slant path propagation. Remote Sensing of the Propagation Environment (Digest No: 1996/221), IEE Colloquium on, 19 Nov. 1996: 9/1 -9/6. Filho, F.C.M., Cole, R.S., and Sarma, A.D. (1986). Millimetre-wave Rain Induced Attenuation: Theory and Experiment. IEE Proceedings, vol. 133, pt. H, no. 4, August, 1986: 308-314. Freeman, R.L. (1987) Radio System Design for Telecommunications (1-100 GHz), John Wiley & Sons, Inc., New York. Gang Liu; Ong, J.T.; Choo, E.; Teo, C.G. (2000) The effects of wet radome on a short millimetre-wave link in Singapore. Communication Technology Proceedings, 2000. WCC - ICCT 2000. International Conference on, Volume 1, 21-25 Aug. 2000, pp. 188 – 191. Gibbins, C.J. (1992). Studies of Millimetre-wave Propagation and Related Meteorology Over a 500m Path. URSI Open Symposium, Wave Propagation and Remote Sensing, Ravenscar, North Yorkshire, UK, 8-12 June, 1992:10.6.110.6.8. 119 Goddard, J.W.F. (1991) The Chibolton Rain Radar. IEE Colloquium on National Radio Propagation Programme, 8 January 1991, pp 3/1 – 3/3. Goddard, J.W.F., and Thurai, M. (1997). Radar-Derived Path Reduction Factors for Terrestrial Systems. IEE 10th International Conference on Antenna and Propagation, 14-15 April 1997, Conference Publication No. 436, pp. 2.2182.221. Goldhirsh, J. (1979). A Review on the Application of Nonattenuating Frequency Radars for Estimating Rain Attenuation and Space-Diversity Performance. IEEE Transactions on Geoscience Electronics, October 1979. Vol. GE-17, No. 4: 218-239. Goldhirsh, J. (1983) Rain Cell Size as a Function of Rain Rate for Attenuation Modeling. IEEE Transactions on Antennas and Propagation, vol. AP-31, no. 5, September 1983, pp 799 – 801. Goldhirsh, J., Musiani, B.H. (1992) Dimension Statistics of Rain Cell Cores and Associated Rain Rate Isopleths Derived from Radar Measurements in the MidAtlantic Coast of the United States. IEEE Transactions on Geoscience and Remote Sensing, vol. 30 no. 1, January 1992, pp 28 – 37 Haddad, Z.S.; Short, D.A.; Durden, S.L.; Im, E.; Hensley, S.; Grable, M.B.; Black, R.A. (1997). A new parametrization of the rain drop size distribution. Geoscience and Remote Sensing, IEEE Transactions on, Volume: 35 Issue: 3, May 1997, pp. 532 –539. Hornbostel, A.; Schroth, A. (1995). Comparison of radiometer, radar and rain gauge attenuation predictions with Olympus beacon measurements. Antennas and Propagation, 1995. ICAP '95. Ninth International Conference on (Conf. Publ. No. 407), Volume: 2, 4-7 April 1995, pp. 23 –26. 120 Hornbostel, A.; Schroth, A.; Kutuza, B.G.; Evtuchenko, A. (1997). Dual polarisation and multifrequency measurements of rain rate and drop size distribution by ground-based radar and radiometers. Geoscience and Remote Sensing, 1997. IGARSS '97. 'Remote Sensing - A Scientific Vision for Sustainable Development’, 1997 IEEE International, Volume: 3 , 3-8 Aug. 1997, pp. 1126 –1128. IEEE, 2004 802.16 Broadband Wireless Access System Requirements Islam, M.R.; Tharek, A.R.; Chebil, J. (2000) Comparison between path length reduction factor models based on rain attenuation measurements in Malaysia. Microwave Conference, 2000 Asia-Pacific 3-6 Dec. 2000, pp 1556 – 1560. ITU-R P.530-8 Propagation Data and Prediction Methods required for the design of Terrestrial Line-Of-Sight Systems. ITU-R P.618-5 Propagation Data and Prediction Methods for the design of EarthSpace Telecommunication Systems. ITU-R P.837-4 Characteristics of precipitation for Propagation Modeling ITU-R P.838-1 Specific Attenuation model for rain for use in prediction methods ITU-R P.839-2 Rain Height for Prediction Methods ITU-R SA 1414 Characteristics of Data Relay Systems. Japan Meteorological Agency (1979) Weather Radar Manual, October 1979. Jatila, E., Puhakka, T. (1973). On the accuracy of radar rainfall measurements. Geophysica, 12: 127-140. 121 Joss, J., Thams, J.C., and Waldvogel, A. (1968). The Variation of Raindrop Size Distributions at Locarno. Proceedings of the International Conference on Cloud Physics: pp 26-30. Juy, M., Maurel, R., Rooryck, M., Nugroho, I.A., Hariman, T. (1990). Rain Rate Measurements in Indonesia,” Electronics Letters, vol. 29, no. 9, 26th April 1990: pp. 595-598. Karim, M. (2000) Effective Path Length for Terrestrial Microwave Links-Predictions Based on Rain Attenuation Measurements in Malaysia. Masters Thesis, Universiti Teknologi Malaysia, Skudai, Johor. Khamis, N.H., Din, J., Tharek, A.R. Rain Cell Size Distribution Analysis Using Rain Gauge Network for Attenuation Studies. Proc. Malaysian Science and Technology Congress ’99, Symposium C, Vol. 1, Johor Bahru, Malaysia, 6-8 Dec., 1999, pp. 339-345. Khamis, N.H., Tharek, A.R., Din, J., Awang, M.A., Nawi, S., Yatim. F. (2000). Space Diversity Study for Satellite Communications by Using Radar Data in Tropical Region. Symposium Proceedings, 4th Inter’l Wireless and Telecommunications Symposium/Exhibition, Subang Jaya, Malaysia, 15-19 May 2000, pp. 101-104. Konefal, T.; Spillard, C.; Grace, D. (2002). Site diversity for high-altitude platforms: a method for the prediction of joint site attenuation statistics. Microwaves, Antennas and Propagation, IEE Proceedings -, Volume: 149 Issue: 2, April 2002, pp. 124 –128 Kuhn, U. (1989) Measurement of 18 GHz propagation and of rain cell size. Technische des RFZ, vol. 33, no. 3, August 1989, pp. 63 – 73. 122 Lahaie, P., Lecours, M., Bellon, R.A. (1993). Use of Meteorological Radar Images to Build a Rain Attenuation Prediction Model for Terrestrial Microwave Radiocommunication. Canadian Conference on Electrical and Computer Engineering, 14-17 Sept. 1993, vol.1: 68 - 71. Laws, J.O., and Parsons, D.A. (1943). The Relation of Raindrop Size to Intensity,” Trans. Amer. Geophys. Union, 24, 1943, pp 432-460. Lin, S.H. (1975). A method for calculating rain attenuation distributions on microwave paths. The Bell System Technical Journal, Vol. 54, No. 6, JulyAugust 1975, pp. 1051-1086. Maitra, A. (2000). Three-parameter raindrop size distribution modeling at a tropical location. Electronics Letters, Volume: 36 Issue: 10, 11 May 2000, pp. 906 – 907 Marshall, J.S., and Palmer, W.M.K. (1948). The Distribution of Raindrops with Size,” Journal of Meteorology, 5, 1948: pp 165-166. Md Rafiqul Islam (2000). Rain Attenuation Prediction for Terrestrial Microwave Links Based On Rain Rate and Rain Attenuation Measurements in a Tropical Region.. Universiti Teknologi Malaysia: PhD's Thesis. Meneghini, R.; Kumagai, H.; Wang, J.R.; Iguchi, T.; Kozu, T. (1997). Microphysical retrievals over stratiform rain using measurements from an airborne dualwavelength radar-radiometer,” Geoscience and Remote Sensing, IEEE Transactions on, Volume: 35 Issue: 3, May 1997, pp. 487 –506. Miller, G.M. (1999). Modern Electronic Communication, 6th.ed. Prentice Hall, Inc. New Jersey, USA. 123 Mitnik, L.; Mitnik, M.; Ming-Kuang Hsu. (1998). Satellite X-band real aperture radar signatures of nonprecipitating clouds, rain cells and rain bands,” Geoscience and Remote Sensing Symposium Proceedings, 1998. IGARSS '98. 1998 IEEE International, Volume: 2, 6-10 July 1998, pp. 742 –744. Moupfouma, F. (1984). Improvement of a Rain Attenuation Prediction Method for Terrestrial Microwave Links. IEEE Trans on Antenna and Propagation, vol. AP-32, Pt. H, no. 12, December 1984: 1368-1372 Moupfouma, F. (1997) New theoretical formulation for calculation of the specific attenuation due to precipitation particles on terrestrial and satellite radio links. International Journal of Satellite Communications, v 15, n 3, May-Jun, 1997, p 89-99. Nor Hisham Haji Khamis, Jafri Din, and Tharek Abdul Rahman (2004). Determination of Rain Cell Size Distribution For Microwave Link Design in Malaysia RFM 2004 OCTOBER 5 - 6, 2004, Selangor, MALAYSIA Oki, R., and Kozu, Toshiaki. (2000) TRMM PR rain rate and AMeDAS Rain Gauge Comparison. IEEE, 2000: p. 1355-1357. Olsen, R.L. (1982). A review of theories of coherent radio wave propagation through precipitation media of randomly oriented scatterers, and the role of multiple scattering. Radio Science, October 1982. Vol. 17, No. 5: 913-928. Olsen, R.L., Rogers, D.V., Hodge, D.B. (1978). The aRb Relation in the Calculation of Rain Attenuation,” IEEE Trans. On Antennas and Propagation, vol. AP-26, no. 2, March 1978, pp. 318-328. Ong, J.T., Timothy, K.I., Choo, F.B.L., Carson, W.L. (2000) Effective rain height statistics for slant path attenuation prediction in Singapore,” Electronics Letters, 30th March, 2000, vol. 36, no. 7, pp. 661-663. 124 Ong, J.T.; Shan, Y.Y. (1997). Rain drop size distribution models for Singaporecomparison with results from different regions. Antennas and Propagation, Tenth International Conference on (Conf. Publ. No. 436), Volume: 2, 14-17 April 1997, pp. 281 –285. Pan, Q.W., and Bryant, G.H. (1992) Results of 12 Ghz Propagation Measurements in LAE (PNG). Electronics Letters, 8 October 1992. Vol. 28, no. 21, pp 2022 – 2024. Pan, Q.W., and Bryant, G.H. (1994) Effective Rain-cell Diameters and Rain-column Heights in the Tropics. Electronics Letters vol. 30 no. 21 13 October 1994: pp. 1800-1802 Perez Garcia, N.A., and da Silva Mello, L.A.R. (2004). Improved method for prediction of rain attenuation in terrestrial links. Electronics Letters vol. 40, no. 11, 27 May, 2004: 683-684). Pontes, M.S., and Mello, L.A.R. da Silva. (1993). An improved method for slant path rain attenuation prediction. SBMO 93, International Microwave Conference, Sao Paolo, Brazil, July, 1993:533-538 Pontes, M.S.; da Silva Mello, L.A.R.; Souza, R.S.L. (1995). Statistical behaviour of the effective rain height in the tropics. Antennas and Propagation, 1995. ICAP '95. Ninth International Conference on (Conf. Publ. No. 407), Volume: 2, 4-7 April 1995, pp. 119 –122. Puhakka, T. (1974). On the variability of the Z-R relationship in rainfall related to radar echo pattern. Geophysica, 13: pp 103 –119. 125 Rincon, R.F., Lang, R., Meneghini, R., Bidwell, S., and Tokay, A. (2001). Estimation of Path-Average Rain Drop Size Distribution Using the NASA/TRMM Microwave Link. Geoscience and Remote Sensing Symposium, 2001. IGARSS '01. IEEE 2001 International , Volume: 3 , 9-13 July 2001 p. 1038-1040. Rogers, R.R. (1976). Statistical Rainstorm Models, Their Theoretical and Physical Foundations. IEEE Trans. On Antennas and Propagation, vol. AP-, no., July, 1976: pp. 547-566. Ruthroff, C. L.(1970). Rain attenuation and radio path design. Bell Sys. Tech. Journal: 121-135 Sauvageot, H., Mesnard, F. (1999) The Relation between the Area-Average Rain Rate and the Rain Cell Size Distribution Parameters. Journal of the Atmospheric Sciences, vol. 56, January 1999, pp 57 – 70 Seed, A., Austin, G.L. (1990). Variability of Summer Florida Rainfall and its Significant for the Estimation of Rainfall by Gages, Radar, and Satellite. Journal of Geophysical Research. Vol. 95, No. D3, February 28, 1990: 22072215. Servomaa, H.; Muramoto, K.; Shiina, T. (200). Z-R relationships for precipitation and evaluation using size distribution. Geoscience and Remote Sensing Symposium, 2000. Proceedings. IGARSS 2000. IEEE 2000 International, Volume: 5, 24-28 July 2000, pp. 1810 –1812. Sharul Kamal Abdul Rahim. (2001). Study of Microwave Signal Attenuation Over Terrestrial Link at 26 GHz in Malaysia. Universiti Teknologi Malaysia: Master's Thesis. 126 Silva Mello, L.A.R.; Pontes, M.S.; Souza, R.S.L. (1997). Rain attenuation prediction for the design of site-diversity LEO/SMS Gateway configuration in the tropics. Proceedings Microwave and Optoelectronics Conference, 1997. 'Linking to the Next Century'. 1997 SBMO/IEEE MTT-S International, Volume: 2 , 11-14 Aug. 1997 Pages:729 - 733 vol.2 Watson, P.A, V. Sathiaseelan and B. Potter. (1981)Development of a climatic map of rainfall attenuation for Europe," Interim Report for European Space Agency under ESTEC contract No. 4162n9fNIJDG (SG), Report No. 300, 1981. Wexler et al (1963) Radar Reflectivity and Attenuation of Rain. Journal of Applied Meteorology. Vol. 2, April 1963, pp 276 – 280. Wilson, J.W. (1964). Evaluation of Precipitation Measurements with the WSR-57 Weather Radar. Journal of Applied Meteorology, vol. 3, April 1964: 164-174. Yagasena, A., Hassan, S.I.S., Yusoff, M.M.M. (1995) Rain Attenuation Prediction at 6.75 GHz in Malaysia Using Rain Gauge and Radiometer Measurements. IEEE, 1995, pp. 596-599. Yagasena, A., Hassan, S.I.S., Yusoff, M.M.M. (2000). Rain Attenuation Prediction at 6.75 GHz in Malaysia Using Rain Gauge and Radiometer Measurements. Theme: 'Electrotechnology 2000: Communications and Networks'. [in conjunction with the] International Conference on Information Engineering., Proceedings of IEEE Singapore International Conference on Networks, 2000: pp. 596-599. Zainal, A.R., Glover, I.A., Watson, P.A. (1993). Rain Rate and Drop Size Distribution Measurements in Malaysia. Proceedings of the 1993 International Geoscience and Remote Sensing Symposium (IGARSS ’93), Understanding of Earth Environment, Japan, vol. 1, 1993, pp. 309-311. Better 127 Zhang, G.; Vivekanandan, J.; Brandes, E. (2001). A method for estimating rain rate and drop size distribution from polarimetric radar measurements. IEEE Transactions on Geoscience and Remote Sensing, April 2001. Volume: 39, Issue: 4, pp. 830 – 841. Zhang, G.; Vivekanandau, J.; Brandes, E. (2000). A method for estimating rain rate and drop size distribution from polarimetric radar measurements. Geoscience and Remote Sensing Symposium, 2000. Proceedings. IGARSS 2000. IEEE 2000 International, Volume: 1, 24-28 July 2000, pp. 180 –183. 128 Appendix 1.1 Earth’s Climate and Raindrops 1. Atmospheric Conditions A truly reliable wireless communication system such as satellite and microwave links must take into account the effects of atmospheric conditions. The earth’s atmospheric conditions are very important and must be considered for wireless communication especially for satellite and microwave links. It has a direct effect on frequencies and techniques used for transmission. For example, at 11 GHz, electromagnetic energy will be absorbed by water molecules, resulting in high attenuation. Also, the phenomenon of ‘bending’ or refraction of electromagnetic wave at certain frequencies in earth’s atmosphere must be taken into account for reliable communication link. Figure A1.1.1 shows some division of earth’s atmosphere. For terrestrial links, conditions in the troposphere must be considered while for satellite link, conditions in the neutral and ionized sphere must be taken into account. In nonionized atmosphere, the pressure, temperature, and water vapor content (humidity) all decrease with increasing altitude. The dielectric constant also decreases with altitude. Since electromagnetic waves travel faster in a medium of lower dielectric constant, bending or refraction of the waves will occur (Freeman, 1987). This is a factor to be considered in designing a microwave link. Changes in terrain also have strong effects on ground waves. If the ground is highly conductive, attenuation will be reduced. Ground wave propagation is much better over seawater than very dry desert terrain. Diurnal changes must also be considered. In the ionized atmosphere, the thickness of these layers varies during the day and nighttime. Electromagnetic waves traveling through these layers will be refracted (Miller, 1999). 129 Figure A1.1.2 shows pressure and temperature distribution of the earth’s atmosphere that can have an effect on microwave transmission. Another aspect of the earth’s atmosphere that must be considered is that the potential energy distributed in the atmosphere is not uniform. The difference in pressure, temperature, and energy has an effect on the earth’s weather and the movement of air mass. Figure A1.1.3 shows the changes in the earth’s weather and climatic differences in various parts of the world. This also has an effect on the earth’s weather. Thus, various part of the world has different climatic experiences that give rise to the need for specific studies of local weather conditions for wireless communication applications. Figure A1.1.1 Earth’s atmosphere (Allnutt, 1989) 130 Figure A1.1.2 Pressure and temperature of earth’s atmosphere (Allnutt, 1989) 131 Figure A1.1.3a Mean temperature of the earth in July (Allnutt, 1989) Figure A1.1.3b Mean temperature of the earth in December (Allnutt, 1989) 132 As Malaysia falls into a heavy rainfall region, attenuation of electromagnetic wave propagation due to rain is of paramount concern rather than other atmospheric conditions. This is especially true for terrestrial and satellite microwave links. Various studies and analysis have been conducted to find a suitable model for prediction of rain induced attenuation in the tropical region (Ajayi et al, 1988; Juy et al, 1990; Yagasena et al, 1995; Aydin et al, 2002). 2. Terminal Velocity and Drop Shapes Terminal velocity of a raindrop depends on atmospheric pressure, humidity and temperature. It increases with larger drop size but the rate of increase gradually decreases before reaching its maximum of about 9 m/s (Gibbins, 1992). Figure A1.1.4 shows the terminal velocity profiles of raindrops. 133 Figure A1.1.4 Terminal velocity of a rain drop (Allnutt, 1989) The drop shape of a raindrop will cause depolarization of the electromagnetic wave. An oblate drop with horizontal main axis will attenuate more in the horizontal polarization. However, the difference of attenuation calculations using different drop shapes show variations of less than 15% than using spherical shape (Crane, 1975). The smallest drops are found in clouds and the largest will not exceed 4 mm in radius. Larger drops are unstable and tend to break up. Figure A1.1.5 shows the shapes and sizes of raindrops. 134 Figure A1.1.5 Rain drop shapes and sizes (Allnutt, 1989) 3. Drop Size Distribution (DSD) Drop size distribution (dsd) is very important for rain attenuation calculation. Different region has different dsd while other rainfall microstructure parameters such as terminal velocity, and the shape of the raindrops has less regional dependence (Haddad et al, 1997; Meneghini et al, 1997; Aydin et al, 2002). It is usually expressed as the number of drops per cubic meter of air per unit size interval. Dsd varies considerably with time and locality, even for the same rain rate within the same rain event. At low rain rates, raindrop spectra are about 2 to 3 mm. At the beginning of a rain event, the drop spectrum (at ground level) is dominated by larger sizes because larger drops fall faster and smaller drops easily evaporated. Since dsd is closely related to the rainfall rate, models have been formulated to determine drop size 135 distribution. Common models for dsd are Marshall & Palmer, Gamma, Log-Normal, and Laws & Parsons models. 4. Reference 1. Freeman, R.L. (1987) Radio System Design for Telecommunications (1-100 GHz), John Wiley & Sons, Inc., New York. 2. Miller, G.M. (1999). Modern Electronic Communication, 6th.ed. Prentice Hall, Inc. New Jersey, USA. 3. Allnutt, J.E. (1989). Satellite-to-ground radiowave propagation, Theory, practice and system impact at frequencies above 1 GHz, Peter Perigrinus Ltd., London, UK. 4. Ajayi, G.O., Ezekpo, S.U.B. (1988). Development of Climatic Maps of Rainfall Rate and Attenuation for Microwave Applications in Nigeria. The Nigerian Engineer, vol. 23, no. 4, 1988: 13-30. 5. Juy, M., Maurel, R., Rooryck, M., Nugroho, I.A., Hariman, T., “Rain Rate Measurements in Indonesia,” Electronics Letters, vol. 29, no. 9, 26th April, 1990, pp. 595-598. 6. Yagasena, A., Hassan, S.I.S., Yusoff, M.M.M., “Rain Attenuation Prediction at 6.75 GHz in Malaysia Using Rain Gauge and Radiometer Measurements,” Theme: 'Electrotechnology 2000: Communications and Networks'. [in conjunction with the] International Conference on Information Engineering., Proceedings of IEEE Singapore International Conference on Networks, 1995, pp. 596-599. 136 7. Aydin, K.; Daisley, S.E.A. (2002). Relationships between rainfall rate and 35GHz attenuation and differential attenuation: modelling the effects of raindrop size distribution, canting, and oscillation. Geoscience and Remote Sensing, IEEE Transactions on, Volume: 40 Issue: 11 , Nov. 2002: 2343 –2352. 8. Gibbins, C.J. (1992). Studies of Millimetre-wave Propagation and Related Meteorology Over a 500m Path. URSI Open Symposium, Wave Propagation and Remote Sensing, Ravenscar, North Yorkshire, UK, 8-12 June, 1992:10.6.1-10.6.8. 9. Crane, R.K. (1975). Attenuation Due to Rain - A Mini-review. IEEE Trans. On Antennas and Propagation, vol. AP-23, no. 5, September, 1975:750-752. 10. Haddad, Z.S.; Short, D.A.; Durden, S.L.; Im, E.; Hensley, S.; Grable, M.B.; Black, R.A. (1997). A new parametrization of the rain drop size distribution. Geoscience and Remote Sensing, IEEE Transactions on , Volume: 35 Issue: 3 , May 1997, pp. 532 –539. 137 Appendix 3.1 Specifications of the Casella Rain Gauge 138 Appendix 3.2 Data Sample from RGN-UTM 1 Date 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 08 08 08 08 08 08 08 08 08 08 08 08 08 08 10 10 10 12 12 13 13 09 09 09 09 09 09 09 09 09 09 09 09 09 09 09 09 09 09 09 09 09 09 09 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 Time 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 11 11 11 11 11 11 11 11 11 11 11 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 11 11 13 00 05 10 10 41 42 44 46 47 48 49 51 52 54 56 01 06 10 13 15 19 24 25 27 28 29 30 20 35 37 38 39 40 41 42 43 44 45 47 49 52 07 18 28 37 58 53 54 R (mm/min) 000.5 000.5 000.5 000.5 001.0 000.5 000.5 000.5 000.5 000.5 000.5 000.5 000.5 000.5 000.5 000.5 000.5 000.5 000.5 000.5 000.5 000.5 000.5 000.5 000.5 000.5 000.5 001.0 001.0 000.5 000.5 000.5 000.5 000.5 000.5 000.5 000.5 000.5 000.5 000.5 000.5 000.5 000.5 000.5 13 13 13 13 13 13 13 13 10 10 10 10 10 10 10 10 96 96 96 96 96 96 96 96 10 10 10 10 11 16 17 17 55 56 57 59 08 52 09 30 000.5 000.5 000.5 000.5 000.5 000.5 000.5 000.5 139 Appendix 3.3 Radar Measurement Theory Radar Measurement Theory Amount of power intercepted by a particle is given by Pσ = Pt G A 4 π d 2 (A1) t where Pt = transmitted power, G = gain of antenna, At = cross-sectional area of target, d = range or distance of target. Assuming power is reradiated isotropically by the particle, Pr = Pσ 4 π d 2 A (A2) e where Ae = effective area of antenna. Thus, Now, Pr = Pt G A t ( 4 π )2 d 4 (A3) A e A e = ρ Ap (A4) where Ap = cross-section area of antenna ρ = antenna efficiency. Silver (1951) has shown that A e = G λ2 4 π (A5) 140 For circular parabolic antenna, (A6) 8 π A p 3 λ2 ≈ G Ae = 2/3 Ap which gives Then Pr (A7) Pt G 2 λ2 = (4 π ) 3 A t d 4 P t A 2p ≈ 9 π λ A t 2 d P σ = σ Pi Now (A8) 4 (A9) where σ = backscattering cross-section of particle, and Pi = power intercepted by the particle. Thus σ Pi = 4 π d2 S (A10) where S = backscattered power per unit area at receiving antenna. Then for single scatterer, Pr = Pt G 2 λ2 (4 π ) 3 σ i d (A11) 4 However, radar beam illuminates a large group of raindrops at the same time; and the number is equal to that within a volume defined by the beamwidth and the pulse length of the radar set. If the back-scattered power is averaged over a volume of n numbers of randomly distributed scatterers, then the average received power can be written [1] as Pr = Pt G 2 λ2 (4 n ∑ π )3 d 4 i = 0 σ i (A12) Denoting the back-scattered power of m numbers of uniformly distributed scatterer per unit volume, 141 σ i = Vm m (A13) ∑ σ iv i= 0 where Vm = volume where particles are uniformly distributed, and σiv = backscattering cross-section of a single scatterer per unit volume. The volume Vm occupied by the radar beam of horizontal beamwidth θB and vertical beamwidth φB, approximately can be written [1,3] as Vm = π 4 θB φB d d 2 2 c τ 2 (A14) where c = velocity of light (m/s), and τ = pulse duration (sec). Now, taking σ per unit volume and putting the expression of Vm in eq. A12 above, we get Pr = Pt G 2 λ2 θ B φ B h 512 π 2 d 2 ∑ σ (A15) iv vol where h = c τ = pulse extent, and = radar reflectivity. ∑ σ iv vol In deriving these equations, it is assumed that, across the radar beam between half power points, the transmitted power per unit area has the same value. But, actually, the transmitted power is maximum along the beam axis and decreases to half of the value at the angles corresponding to the half of the beamwidth. So, by considering the power per unit area as Gaussian function within the main lobe, the radar equation can be written as [1,3](from Probert-Jones, 1962) Pr = Pt G 2 λ2 θ B φ B h 5 1 2 (2 ln 2 ) π 2 d 2 ∑ vol σ iv (A16) 142 Substituting for the Gaussian beam shape, G Pr π2 θB φB = (A17) Pt G λ2 c τ = 1 0 2 4 ln 2 d 2 ∑ (A18) σ iv vol When the radar wavelength is large enough compared with the circumference of a scattering particle of diameter D (Rayleigh scattering region), the radar cross-section is [4,5] σ iv π5 D 6 = where λ4 K 2 (A19) K 2 = (ε - 1)/(ε + 1), and ε = permittivity of scattering particle. The value of K 2 for water varies with temperature and wavelength. At 100C and 10 cm wavelength, it is approximately 0.93 [5]. By substituting σ i v into eq. A16, we get Pr = π Pt G c τ 1 0 2 4 ( ln 2 ) d 2 λ 2 K 2 ∑ D 6 (A20) vol The sum of the sixth power of diameter per unit volume is called Z, the radar reflectivity factor or Z = ∑ D6 (A21) Experimental measurements showed that Z is related to rainfall rate R by [3,5,7] 143 Z = aRb (A22) Here, a and b are empirically determined constants. The choice of Z-R relationship can be made on the basis of the type of rain. For different types of rain the Z-R relationship can be given by [5] Z = 200R1.6 (stratiform rain), (A23a) Z = 31R1.71 (orographic rain), and (A23b) Z = 486R1.37 (thunderstorm rain) (A23c) By considering stratiform rain, the average received power can be written as Pr = 2 .4 P t G τ R 1 .6 d 2 λ 2 x 1 0 -8 (A24) where R in mm/hr, λ in meters, τ in sec, and Pt in watts. Reference [1] Skolnik, M.I., Introduction to Radar Systems, McGraw-Hill Book Co., New York, 1980. [2] Rogers, R.R , “Statistical rainstorm models, their theoretical and physical foundations,” IEEE Trans. on Ant. and Prop., July 1976, pp 547 - 566. [3] Battan, L.J., Radar Observation of the Atmosphere, The University of Chicago Press, 1973. [4] Allnutt, J.E., Satellite-to-ground radiowave propagation, Theory, practice and system impact at frequencies above 1 GHz, Peter Perigrinus Ltd., London, UK, 1989. [5] Puhakka, T., “On the variability of the Z-R relationship in rainfall related to radar echo pattern,” Geophysica, 13, pp 103 -119, 1974. 144 Appendix 3.4 Radar Calibration Checklist 145 146 147 148 Appendix 3.5 Kluang Radar Station Specifications Meteorological Radar MR 781 S Stesen Kajicuaca, Kluang with RAPIC Transmitter EH663 v. 8.00 Station ID 3 Kluang Station position latitude = 2.020o; longitude = 103.320o Reflector 12 feet parabolic (3.66m) Frequency 2800 MHz Polarization vertical Gain 38 dB Coverage elevation: -2o to +90o; azimuth: 360o Beamwidth 2.0o Pulse duration 2.2 µs PRF 278 pps Peak power 389 kW STC range 230 km 149 Appendix 3.6 An Example of a Radar Data COUNTRY: 458 NAME: Kluang STNID: 03 DATE: 00798 TIME: 16.22 VERS: 8.04 RNGRES: 1000 ANGRES: 1.0 VIDRES: 16 STARTRNG: 2000 ENDRNG: 256000 PRODUCT: VOLUMETRIC [162200798] PASS: 01 of 15 IMGFMT: PPI ELEV: 000.5 DBZLVL: 11.8 23.0 28.0 31.0 34.0 37.0 40.0 43.0 46.0 49.0 52.0 55.0 58.0 61.0 64.0 %041A10Epp20EA49VA15}IAEA %042A12h21VA48hEi2l11F-j %043A11h4l17EFA47+1-+1-1l8+ %044A36E]o49+1-1l1ll13 %045A10\-24h52l2+1+ %046A3EA4@+-22h44l+-1@7+1 %047A11l13l54v2k2@8 %048A11l7l60lqk10l18 %049A11l69hm-10+ %051A5EA1EFA %052A6F1A38 %053A6EAEA37 %054A6EA@1l13h19 %055A8E %056A4EAFec+ %057A6FAxAl %058A6Eg(p %059A3EA3@m %060A8EAm %061A10 %063A10 %064A7E1 %067A3h %068A5 %069A4l2 %074A2 %093A2l+-1h+ %094A2l1+-1 %095A9hw %096A5h5\ %097A11@2 %098A9@3+ %099A13m2-3l %100A7EA3E1A7 %101A8FA3Ep1k1l3l63vy %102A14@m-1@67Yg %103A12F1j1kv1FA66}1 %104A13F[@2qFA67F %105A6EA3G-tA2qAFA67+ %106A10EAFAEFA1v2nA75 %107A10EA1FA1@+gE1A75 %108A13E+A1+1wA76 %109A5E1A4E2AEdvy %110A13Eha1+2+p52q %111A13EAF-ou-56hm %112A14FEAFe1--51lEFA2 %113A2h2EA4@GAEqdt254G %114A4@7Em1A5l3l39l5@1EGA %115A12@6Eih2vp45YA1E1 %116A11E1A1GA3Eog+vo38l6hv %117A12E1AGA7l1}/45Y1j %118A13FA1HAFA@vkvG[-43Y2 %119A14GA2Ee-EA1lFGA42+uAEk %120A13GAGA2h2EA2BFA42V1i2+-40 %121A10GA6nt-h+AEA40UxCp46 %122A18E-AEAm1-42hFHDo25ll3ll1l4 %123A2@15EkFA@46@F/18l2+5u1l+4-1l2l %124A5h5FA1GAEFAEd1GCp36l1lllEd21+2u2u1u-ll+-l2l %125A14FA3GAVAHEA34l+uWo1q1k19+1u2u1u3-l1l1+1uu1 %126A6@1FA4GAGA2@1IaA32+1U1A1ntgqp16+4u3u-+-+3u-2l25l %127A9FA2FAFA1H1GAEAIAFA33l1hhEA2++p18+1u4uu1-l+7u-23 %128A19GgSAHA37l1+1-2+-15ll+3uuuu10u1u-8ll5l4+1 %129A22EFAFA36+17l17l1l+11uu2+--12+-1l+-4+ %130A21Gt+1A67+2-+1u6u3vs1-7+1-8+u-+3 %131A8FA12EA1GA69+5-3+3u+UB1g+-3+-1+u-ll2l2+u2uu3 %132A9FA10EAEqAl67l+3uuu2u3+g1u2-10ll1l1+u2u1uuu1-1+ %133A21ESvao68l1+9S1t1l5u1u2-ll+Spl+1uu-+u1-l+1 %134A20GA2G+Al68+6uu3-m2-+6u-2+Syt+k2l1+1u3u1-l+2 %135A19HkA2HA1l60l6+2Sp+1u3u2u7u5u2vo2qy-2S-2uu6-+ %136A21FAG1HAHA57l6llVp5-4u1uu8uu6+ukv1y1-3-3+u5u %137A19H1A2GAGA\-54l7+W+-o+3u1-7l+u8-q-+2-g1m2lu3-1l2+2-67+1 %138A17FA2GlA3@65+-E+1jt1u1-3ll1+u3llu6llu1u3u10-+-3+-64+l2 %139A19G2A1Fg1j-64q1UA3hq-1-l2l4+2Skq-u3lg+uu3uuu4-6+1-64+ %140A10FA1GA3F1AGAG-jg-64lwAF1Ar$$kl4u1u4uu2+t1ll1u2-+5uu1-l8+ %141A20EA1GA1hm-65vvA1EAr+1p+(t8u5gqut+-2-1+1u3u-5+1u1 %142A2@9FEA2F1A4qu1p65+v$1AEAE1AEA2EA@+14+1k+u-+-2+2u1l1+u2uu-39+-10+ %143A23EA@3l45l19lr$AFu2[EA2Ep3-5u1uu4uu1-1ll+32l+u6-39+ %144A2h9GA5@1F/h1h45lm-17+W+2AE1-1++-1kt5u4u2-1l1l8+-4ll+2 %145A13Gdp5FAh50\1-18Yy1uar$orAE-1s4-+3-1+1uu1u1u-+-6+2u %146A13FuEA7l50+-20mW$wAF-1AE-on1km-+12-l6l3l2+1 %147A13FEAEa1hFAmu-49l19+SAEv1-EAx2ancq6pl+5-6+2-l1l2+ %148A10EA2FgA1h@2l1+-17GA27h1ll12+1Slqv1FArg+4$uoq2-3u1u-7+1-1l6 150 %149A5h7Eha2hm-1+1-3h41l2l9+++u$1EA1r2-2{1(y2uk+4-+2-1+8u1 %150A7@FA3GAEtsSpl+-5FA5h32ll12+w3A@q4y2yucq2-7-+6-+3u6-24+ %151A14GEA@5l3l10ll30+u-l9v1q1kc1hl1h+gmv2e1+u6u1u4-+3u-27+2 %152A5h5GAGj-1p2l1+-11vc5l22l+-10lhxk-2uu-1l+u1Sphl1+-2+-3l5+ %153A14Gj$1oh18lGB-6l16l2+1-4+-2v2+k5S3t1l1ug3+1 %154A14GA@E1A3lm1-12hh4HA19l2+2u2u1-2vy2p2+gn2lu1tl+p %155A9FA3GjoEA2lh2ll9l1l36+-5v3-1+AE1kcn1-ul1u1-35 %156A9FA4GAFAFA2l1vp47+-11l1Fk2(ycr1(t2u-39+-42 %157A12GA1FuAFA2m-65VFAGe(pE-+kk4 %158A11EGA2Fu2[+u-6l54+1FlEAEA@E2pk3 %159A10EAGA2F-lA1+2u-54+-3U+FA2EA1E1Aqt %160A10@G1A1Gp1k-1ugl54+-3+FpFAE+-Ark %161A10EA4Gd1+o5l43+16mq$mFAEAF %162A4h2EA@1GA2Epc3qpl56l5vyywHA2FA@17m %163A8FA)A2Fco5hh64vp1+1GA1l1n %164A9EA4F1Alx--pEA64we3m %165A17\-@nc77 %166A9@2E1FAmwam1-1l63+-6 %167A13E2lAVa1q1kl63 %168A13ESEA1@rs-1 %169A5h2FAEA3FA1wh(pl %170A5h4r+AEA3F[@1vk31+ %171A11@1E+A2xdEA38 %172A5qAFA1r1wA2@EaEAmgl31 %173A4\g2roE+EA1@2nt-1q1 %174A7@1@1@EA2xA2EA+EA2FA %175A10@5FploFBmrEA1HD %176A6h3h4FAh6EAGlAHEA %177A6@1n1p5@EFA2FA1GA4GAl7+3 %178A9q1p3GAEGAFA5GA6EA7w %179A10hh6G1FA1HA2H1A2Ew[p4vxFAE %180A9qp1nAGA1FA1H1A1G1A5GHA8EA5 %181A9h2l3HA1HAHA2HA1GA8l3l2E %182A10h3hEA1H1A4GA2FA1FA2h5l1GAh2 %183A11q1krUAHAHA2HAHAHAHA3@l7h2E %184A10ll1hh2GA5HA1HIA4@1l7l1 %185A10l5h1HA1HA12EA10 %186A10l4mYEA2H1A7n1 %187A10l3+1-10EA1+2u %188A12@+2g1EAGAFA2EAh4 %189A10l3+1SvAEA1+] %190A2@12+S{( %191Aq1p17 %194B %195A2 %196A1h7 %199B %202B3 %205A11l+ %206A9l3 %207A8 %208A6h9 %209A12q %210A5l13 %212A13l1 %213A5h5l1l1 %214A13l %215A13+ %216A5l4hl %217A11 %218A11\2 %219A12 %220A5l5l4l85 %221A8l3+-4 %222A5h5l+-2+ %223A4l5Eo%224A4l9+-1 %225A11 %226A14v %227A3l8l %228A3 %229A11 %231A110l %232A109+2u %233A37l69lh1EA %234A106ll@EAEe %235A28l12l63+gx1AFEA+u1-l2 %236A110qw1AGA1qyupl3l+ %237A4l10l90l@EFAH1A1Fpoqt1-3 %238A14ll90vmqGA1HA1G1--al1+St2 %239A4h10l89+-V1AGlv-k1AEal1+-1+1 %240A16l23@53lll5l1woHAHA1IAI1p1eo5l+ %241A10l2m122l56l8UwS8yGA1l4+u %242A14ll27EA46l1hll8)+FAH-1-AFAFd %243A15+8l16h48lh+-6UUFA2FGA1FuGAh6+ %244A3m-4+-+-76v1p11EGA3EAEAE3kk4 %245A4h9+-22l54l10l@h1qkro %246A13+-24@66lm2-1+1Spn %247A14l2l21GA68l1l4+AGAFAh6+ %248A4EA7l3l15l74l+-2l3F1Al5Y1+A2 %249A15l26@69+-2+Syp7G+1Ars %250A12ll19l6+-65l2l16CGAHGA@E %251A12+1-98+-15Ul1 %252A12+1-18h63+2-32l1m %253A4h3l1lEA3h3+-72+1 %254A12BFA1FA6l23+ %255A12BGA1h1l26l+-109qp %256A23l23l+103l1vmuol2 %257A14EA1h4l12l4+1-4+1uu-h+1g90l1lmghllh34ll6l5 %258A14@2l10l5l11+1uu2g1xk1k95+Skv1k34+1-16 %259A14FA2l17h9h1+hqcEAEArj102l55+1 %260A17m2-23lhq2ykhEu1lal158+u %261A12EA3m17EA5+Sp1qcq1EA4@1h157+1 %262A10lEA3l24+2-4E1-(AEAFAEA3h76l77+1 %263A9+AFA3l4l19l1v1AEFAr1$1EA85 %264A23l18m-3VFA1EA4E 151 %265A23l20l3)A4EA3EA2 %266A17l26llh2GkA11 %267A5l2+-4l25Eo4H1A1FA3EA@hqk %268A9h3HA1l26l+-GA4E1A1@4 %269A17l26+1 %270A9h6ll10l8l2+1-EA4F1EA4h1l1 %271A43FA+g4Gt1AEA9 %272A8@8l11l8l2qy4FAGkFAEA4h2v %273A8wEA30vupqtVAGkA1@2@1qt-1 %274A4l3)GA5h22ll2FAwoF1-A6h1hh2l %275A9EA7h5@2ll6l1+2G(pvgAFA3l5qp %276A9@7h5EA2lvp4+u1-3h@@1Goo4l3l3 %277A14EA2h5h15vp3m+yVEA2h1qt1-hl %278A14GAEA14l5ll6mScEg1u(p1h3+ %279A32+u-7h4l1E-e1@hh6+1 %280A32l1l8\-m1+pmg1hqph1+1u1 %281A12GA19+u1u-3FA4v2p3qu2yt1-l %282A13EA2l12l1vk+-3HA2m1Slph1+2+2t-3 %283A3l5EA5h8h3Uk2l-6+1-q1{p5l+2 %284A12GA3l5l7rk1Sp7+1+pqkh+uu2u %285A19ll6l2v1-Sk4l2m1v-1ut32+u %286A16h2l5@4v1p1m-8+S2p1l+1 %287A16h9h1+-2l1h3HA6qut3u1 %288A16l12hl1+Sk6l2\l1u5-l %289A16l6lUFA2mu-l7h2hq-uu1u-l %290A14HEA5EA+vm-s-3h6@l+u1-3+-1+ %291A26F1-A2m-2h9+Sk6+1u-+ %292A15FA6lh1EaF1i1-10l+u1-5l2 %293A20\-2FA@EA2+-10mg+-7l3+1 %294A16FA3ll2HA2Egj6xA4l+4-+1u-1+ %295A21l+gI-1Del6h6+2-l3l %296A18h1hl1GAHAG/u-1l3l5+-7+u %297A21hl4GA+-8h6l9ll20 %298A21@5HC1p10@41 %299A17GAEAh1+AIFB1-10m1-38 %300A17EA3l3GAl12 %301A17EA2h+1Wj15FAh36E %302A13GA5mu1ug18@G %303A18FAhm-+-17EA21 %304A6h12hll24 %305A27l13l6FA2l14+ %306A18EA1npl2l16FA5l15 %307A6h4FA2EA4n1g26l10+-+1 %308A14G1A5@1lhl22h11+-1q1 %309A21Ea1vg-35+-llm %310A10GA4Ea1@@1h2l37l1qy1t %311A24qc+1-30l2l2m %312A23l1qt-4l25+l2 %313A15EA7l1qp5l29l+3S--69 %314A20@1lh1m-2ll27lll2l3l3h %315A24l2h6l27lll2+2u1-2 %316A25l6l3l1l30+uu2Sh %317A14EA10lll3h4l32l+v2-u1 %318A35h1h+1-31lm1u2v %319A13@1@2qp1vp2l2h3+1-22+-5l+-ll@ %320A24h4l2h4+-32l+u2u1-2 %321A21qp2nt1-2lll+-32+5-2+u %322Cp36l2+329+u4-4 %323A19FA+-10ll3+2-34+2u %324A12FA8l13ll2l1+1-33l1+1 %325A18hEA12l1m2-l1m-l33+4 %326A15GA3@12+Sp7l26l8l1+2 %327A11GA7h11@5l47+ %328A19h14l46+ %329A12FA3v1p3@4F/nk5@38+112 %330A7EA2FA2EA2h3VA7@l3l4l34+ %331A11EA2EA1h2EA8E-1jh7h33+ %332A9FAFA2@l1@2@7Ep1yp6h30+1u-3 %333A10FEA6wo1F1A4EAEA1@9h28+-+-5 %334A17l6GA4h9h33+-l+3 %335A11EA2l25Edl2l26l6mS %336A11EA4l2h9U1o5FA2l1+-21+-6l+2 %337A10GA2EA1l7FA4l8IA4+2-21l6+1u1 %338A15lUo26+2-34 %339A20h20GAFA2 %340A20wo7h8GCp %341A13EA5h9l8GBg1m-9m%342A21h8l8E$a+-9mg1 %343A20+-7l8hF1Ah12vph4l1+1 %344A12@6+17l@15m-l6r1(3 %345A11G!5Eo1-14ll3GA7m-1l7lVaF[@m %346A12ro4l5l12l12m-10+U3lA\1 %347A13@5l4l11lEA3l7l2l9h1Ekpqt %348A11GAEA8l11+-l17l4l3++{e %349A13EA3l8lGA4l2h2@12l7m1-1vpmu1 %350A13EA1ll4l10h6+-1h5l7q{k1u2Sk+ %351A18+u-l10l8l1EA5lh6x1tk2U1AEk %352A12h6+-+-3h10vp2FAl3l8EAFEA1+SqAGek %353A12@1\-+1u11l1@12Ea1@EA12@1FArp2W)dA %354A3wo5EA1\uugl20EA2lFA12w(-14GHAEo %355A17l1@+1-18m-17vp7vh %356A3h10h4l+1 %357A7EA7EA4+ %358A14GCmo1l7l7E %359A7EAFA5r1o31 %000A12FA1r2e1@h3 %001A7@6wo3@l3hl16l1 %002A7rEA4@2EA2\-2h12l5+g64 %003A4@1EyA2E1A3FA2h1l1lh10h6m %004A4h1EyA4Ecp1FAEill1l17q %005A7EAEA2hxa5m-10l3@5 %006A7nwA1EA3G1HA2nt-11hFA2l6 %007A7nwA6EFA1EFAwo1l15hlv1 %008A4@1rAEFA3@ElAFA1qp+14h2EFAE %009A4h1n{o4E1AEA3lll16h1lGAE %010A7@h3lrvA1h2+1- 152 16@1UEA\ %011A6Edh@h1hG1A3mu-17lFB1- %012A6EA2wphHGA1l15l9m1S %013A7EA1wtp@GA2hh13h13 %014A3EA6hh2GAFAF/hl8h1EA7 %015A9qy1ph3FA2 %016A9hhmSpHA2h+30 %017A3@3qpvpCGA4+2 %018A11qp1@6l32 %019A3@5hq1pFA4ll14l15E %020A11hh7l34 %021A10h1h6m-108 %022A5EA3h1h1qEA1l110 %023A8Ekpl2Y %024A1@EA4Edm-2EA1 %025A9n1--2 %026A9n1t-2 %027A7E1A1h2 %028A3EA3EA1h2 %029A3@3Ed4 %030A8ro3@75+-14l61 %031A4EA2Ekt-5EA11+-55m-8+2-63 %032A10ll20l54hh11 %033A9h21+150vSs %034A9qpl18l52m1+p17 %035A3EA4qp7l9vp2+-46+2-14l+1 %036A9@20h5h57l4+3S %037A4EA3rol5l13qyp49+1u-6vy2u %038A9EA1h6l12GEA1l45l1v1--7lwom2 %039A10@8l12HGA50vl--7+v2(%040A5EA2re7l12GHAl47vms-10xAFkp END RADAR IMAGE 153 Appendix 3.7 Radar Data Encoding Format 154 155 156 157 Appendix 4.1 MATLAB programs %filename : Radselec.m %This program will select the Kluang Station radar data filen='F:\13\02579803.220'; %filename of raw radar data in cd fid=fopen(filen,'r'); select=[];testsc=0; Nsize=1e6;kmax=50; %read the 1st 1Mb chars, 50x. serN=1; %read data for k=1:kmax data=fscanf(fid,'%c',Nsize); sp=findstr(data,'PASS: 01 of 15'); %find loc/add of 1st char in string (of 0.5 deg) sc=findstr(data,'COUNTRY'); if (testsc==1) %read more data if only part was read before select=[select, data(1:sc(1)-1)]; testsc=0; sklg=findstr(select,'Kluang'); if (isempty(sklg)==0) eval(['klg',num2str(serN),'=select;']); filen1=['C:\Radar Data 0.5 deg\12\klg',num2str(serN),'.txt']; fid1=fopen(filen1,'w'); fprintf(fid1,'%c',select); fclose(fid1); serN=serN+1; end select=[]; end for i=1:length(sp) differ=abs(sc-sp(i));%find difference between "COUNTRY" and "KLUANG" n=find(differ< 200); if (isempty(n)==0)%1(true) if empty and 0 if nonempty if (length(sc)> n) select=[select,data(sc(n):sc(n+1)-1)]; sklg=findstr(select,'Kluang'); if (isempty(sklg)==0) eval(['klg',num2str(serN),'=select;']); filen1=['C:\Radar Data 0.5 deg\12\klg',num2str(serN),'.txt']; fid1=fopen(filen1,'w'); fprintf(fid1,'%c',select); fclose(fid1); serN=serN+1; end 158 select=[]; else select=[select,data(sc(n):length(data))]; testsc=1; end end end if(k<kmax),fseek(fid,k*Nsize,-1);end end fclose(fid); clear fid filen testc clear i k kmax n sp sc select differ fid1 filen1 Nsize testsc ans clear data 159 %File name - Decoder. This program will decode the Kluang data into corresponding numerical values Code=['ABCDEFGHIJKLMNOP']; %Data code level=[0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15]; %Corresponding absolute level %Deviation encoding characters Devencod=['![abc]@';'/defgh\';'ijk<lmn';'op.+qr';'stu>vwx';'(ySTUV)';'${WXY}&']; rangbin1=[-3 -2 -1 0 1 2 3]; %Deviation values rangbin2=[-3 -2 -1 0 1 2 3]; %read data from file serN=1; filen=['c:\Radar Data Samples\radar',num2str(serN),'.txt'];fid=fopen(filen,'r'); data=fscanf(fid,'%c');fclose(fid); n=findstr(data,'%'); levklg=[]; for ind=1:length(n) if (ind<length(n)) s0=data(n(ind)+1:n(ind+1)-2); else s0=data(n(ind)+1:length(data)-17); % the array data contains one section only end deg=s0(1:3); lev=[]; i=4; while (i<=length(s0) ) Lr= s0(i); mult=str2num(Lr); if(isempty(mult)==1) nc=find(Code==Lr); if (isempty(nc)==0) lev=[ lev level(nc)]; else [Ic,Jc]=find(Devencod==Lr); lev=[ lev lev(length(lev))+rangbin1(Jc)]; lev=[ lev lev(length(lev))+rangbin2(Ic)]; end else if(i<length(s0)) if (isempty(str2num(s0(i+1)))==0), mult=str2num(s0(i:i+1));i=i+1;end end lev=[lev lev(length(lev))*ones(1,mult)]; end i=i+1; end if (length(lev)>=100) lev=[str2num(deg) lev(1:100)]; else lev=[str2num(deg) lev zeros(1,100-length(lev))]; 160 end levklg=[levklg;lev]; end temp=[]; for i=0:359 ni=find(levklg(:,1)==i); if(isempty(ni)==1) temp=[temp; i]; end end n=length(temp); temp(1:n,2:101)=zeros(n,100); levklg=[levklg;temp]; [Y,Isort] = sort(levklg(:,1)); eval(['levklg',num2str(serN),'=levklg(Isort,:);']); clear Code Devencod Ic Jc Lr fid filen rangbin1 rangbin2 nc ind i level clear s0 lev lev deg n mult ans Y Isort clear data levklg temp n ni serN 161 %filename : DataSelect [Select data for virtual path links from specific area.] % # of files in each data folder d1=175;d2=648;d3=949;d4=906;d5=721;d6=1041;d7=955;d8=792;d9=869; d10=777;d11=785;d12=893;d13=807;d14=662;d15=888;d16=1049;d17=996 ; for m=1:17 eval(['load data',num2str(m),';']); % load data eval(['p=d',num2str(m),';']); % # of files in data for j=1:p eval(['A=levklg',num2str(j),';']); B1=A([3 6 9 12 15],32:51); B2=A([238 241 244 247 250],32:51); B3=A([294 297 300 303 306],32:51); B4=A([334 337 340 343 346],32:51); eval(['D',num2str(j),'=[B1;B2;B3;B4];']); clear B* A end eval(['save SD',num2str(m),' D*;']); clear p levklg* D* end 162 %Filename : CellSize - rain cell size for selected virtual paths data % # of files in each data folder d1=175;d2=648;d3=949;d4=906;d5=721;d6=1041;d7=955;d8=792;d9=869; d10=777;d11=785;d12=893;d13=807;d14=662;d15=888;d16=1049;d17=996 ; clock for ind=1:17 eval(['load SDR',num2str(ind),';']); eval(['q=d',num2str(ind),';']); % load data % # of files in data X1=[];X2=[];X3=[];X4=[];X5=[];X6=[];X7=[];X8=[];X9=[];X10=[]; X11=[];X12=[];X13=[];X14=[];X15=[];X16=[];X17=[];X18=[];X19=[];X20=[ ]; C1=0;C2=0;C3=0;C4=0;C5=0;C6=0;C7=0;C8=0;C9=0;C10=0; C11=0;C12=0;C13=0;C14=0;C15=0;C16=0;C17=0;C18=0;C19=0;C20=0;Cbi g=0; T=[];count=0;total=0; for i=1:q eval(['A=DR',num2str(i),';']); for j=1:20 for k=1:20 temp=A(j,k);t1=A(j,1);t2=A(j,20); if t1 + t2 == 0 if temp ~= 0 T=[T temp]; count=count + 1; end if temp == 0 if count==1 X1=[X1;T];C1=C1 + 1;T=[];count=0;total=total + 1; elseif count==2 X2=[X2;T];C2=C2 + 1;T=[];count=0;total=total + 1; elseif count==3 X3=[X3;T];C3=C3 + 1;T=[];count=0;total=total + 1; elseif count==4 X4=[X4;T];C4=C4 + 1;T=[];count=0;total=total + 1; elseif count==5 X5=[X5;T];C5=C5 + 1;T=[];count=0;total=total + 1; elseif count==6 X6=[X6;T];C6=C6 + 1;T=[];count=0;total=total + 1; elseif count==7 X7=[X7;T];C7=C7 + 1;T=[];count=0;total=total + 1; elseif count==8 X8=[X8;T];C8=C8 + 1;T=[];count=0;total=total + 1; elseif count==9 X9=[X9;T];C9=C9 + 1;T=[];count=0;total=total + 1; elseif count==10 X10=[X10;T];C10=C10 + 1;T=[];count=0;total=total + 1; elseif count==11 X11=[X11;T];C11=C11 + 1;T=[];count=0;total=total + 1; elseif count==12 X12=[X12;T];C12=C12 + 1;T=[];count=0;total=total + 1; 163 elseif count==13 X13=[X13;T];C13=C13 + 1;T=[];count=0;total=total elseif count==14 X14=[X14;T];C14=C14 + 1;T=[];count=0;total=total elseif count==15 X15=[X15;T];C15=C15 + 1;T=[];count=0;total=total elseif count==16 X16=[X16;T];C16=C16 + 1;T=[];count=0;total=total elseif count==17 X17=[X17;T];C17=C17 + 1;T=[];count=0;total=total elseif count==18 X18=[X18;T];C18=C18 + 1;T=[];count=0;total=total elseif count==19 X19=[X19;T];C19=C19 + 1;T=[];count=0;total=total elseif count==20 X20=[X20;T];C20=C20 + 1;T=[];count=0;total=total elseif count > 20 Cbig=Cbig + 1; eval(['Xbig',num2str(Cbig),'=[T];']); T=[];count=0;total=total + 1; end end end end end end eval(['save CSDR',num2str(ind),' X* C* total;']); clear temp t1 t2 X* C* DR* end clock + 1; + 1; + 1; + 1; + 1; + 1; + 1; + 1; 164 %Filename : R1km - rain distribution for 1-km paths %Get 1-km data Onekm=[ ];OnekmNew=[ ];count=1; for i=1:17 eval(['load CSDR',num2str(i),';']); Onekm=[Onekm;X1]; clear X* C* total end %sort data temp = sortrows(Onekm,1); for j=1:(length(temp)-1) a=temp(j,1); if a==temp(j+1,1) count=count+1; else OnekmNew=[OnekmNew;a count]; count=1; end if j==(length(temp)-1) if a==temp(j+1,1) OnekmNew=[OnekmNew;a count]; else OnekmNew=[OnekmNew;temp(j+1,1) 1]; end end end %save Onekm OnekmNew save Onekm OnekmNew -ascii clear all 165 %filename : Path1km7GAt [Total att dist for links of 1-km length] clock d1=175;d2=648;d3=949;d4=906;d5=721;d6=1041;d7=955;d8=792;d9=869; d10=777;d11=785;d12=893;d13=807;d14=662;d15=888;d16=1049;d17=996 ; total=0;% total # of 1-km paths count=0;% # of 1-km paths which have rain/attenuation X=[];% temp to store att values for m=1:17 eval(['load At7G',num2str(m),';']); eval(['p=d',num2str(m),';']); for j=1:p eval(['B=At',num2str(j),';']); for i=1:20 for k=1:20 Att=B(i,k); total=total + 1; if Att~=0 count=count + 1; X=[X; Att]; end end end end end clear d* m p i j B Att % count1 total temp=sortrows(X,1); T=temp(:,1); clear temp X countAt=1;% total count of each att value Att1=[];% att dist for 1-km paths - [AttValue events] for i=1:(length(T)-1) t=T(i,1); if t==T(i+1,1) countAt=countAt + 1; else Att1=[Att1;t countAt]; countAt=1; end if i==(length(T)-1) if t==T(i+1,1) Att1=[Att1;t countAt]; else Att1=[Att1;T(i+1,1) 1]; end end end save P1km7GAttDist Att1 count total; clear all;clock 166 Appendix 4.2 Radar Rain Rate Distribution Table A4.2.1 Rain rate distribution for range-bin size of 1-km from radar data. Rain rate (mm/hr) 0 0.2 1 2.1 3.2 4.9 7.5 11.5 17.8 27.3 42.1 64.8 99.9 153.8 236.8 364.6 Total events/occurrence of rain rate value 4743844 236698 124172 114667 125677 65680 54020 44683 33062 17019 2818 1427 727 378 203 125 5565200 percentage (%) 85.24121 04.25318 02.231223 02.060429 02.258266 01.180191 00.970675 00.8029 00.594085 00.305811 00.050636 00.025641 00.013063 00.006792 00.003648 00.002246 100 The percentage is calculated by, dividing the number of events for a particular rain rate by the total number of events and then multiplying them by 100%. For example, for 11.5 mm/hr rain rate percentage = 44683/5565200 * 100 % = 00.8029 % (A4.2.1) This is easily done using EXCEL or MATLAB. Using the Curve Fitting Toolbox available in Matlab, the best-fit equation for the data in Table A4.2.1 is obtained. The Curve Fitting Tool window is as shown in Figure A4.2.1. This window is invoked using the “cftool” command in Matlab’s Command Window. 167 Figure A4.2.1 Matlab’s Curve Fitting Tool window. Data is chosen and put into the Curve Fitting Tool via the Data option. The Data option window is shown in Figure A4.2.2. It shows the options for choosing data sets for the x and y-axes. It also shows the preview of the data plot. 168 Figure A4.2.2 The window for inputting data in Curve Fitting Toolbox in Matlab. The View Data Set window for curve fitting is as shown in Figure A4.2.3. It shows the data sets for the y and x-axes. As in Figure A4.2.2, it also shows the preview plot of the data. This window is opened by clicking the View button in the Data window. 169 Figure A4.2.3 View Data Set window in Matlab. Fitting is done using the Fitting option. Several fitting types are available in Matlab. It has been found out that the Power fit fitting type gives the best result based on the goodness-of-fit criterions. The Fitting window and its results are shown Figure A4.2.4. 170 Figure A4.2.4 Fitting editor and the result of curve fitting. Two important parameters from this curve fitting exercise is the R-square value and the RMSE or the root mean square error value. These parameters indicate the goodness-of-fit for the curve or equation found from the curve fitting exercise. The R-square statistics measures how successful the fit is in explaining the variation of the data. It is the square of the correlation between the response values and the predicted response values. R-square has the value between 0 and 1. A value of R-square closer to 1 indicates a good fit. In this case, the value of 0.99496 is the best from any types of fit, and it is a good value. The RMSE is the fit standard error or the standard error of regression. A low value closer to 0 indicates a good fit. The result of 7.6482 for RMSE is rather good for this fit. The confidence bound defines the lower and upper values of the fitted coefficients. The bounds define the level of certainty for the fitted coefficients. Matlab has a default value of 95 % as this is the value that is often used. The plot of the results is shown in Figure A4.2.5. 171 Figure A4.2.5 The plots of original data and the curve-fit line. Using the Analysis option, several values or a range of values for the curvefit equation can be found. This is as shown in Figure A4.2.6. 172 Figure A4.2.6 Analysis from curve fitting. The most important result from the analysis of the curve fitting exercise is that the rain rate for 0.01 % of the time. It is shown here that from the radar data, the rain rate for 0.01 % of the time is 120.907 mm/hr. This is very close to the value that is obtained from rain gauge measurements and from ITU-R. This value will be used in the mathematical model that is used to obtain the reduction factor. The results are summarized below. General model Power1: powerfit1(x) = a*xb Coefficients (with 95% confidence bounds): a= 4.382 (2.959, 5.806) b = -0.7204 (-0.777, -0.6637) rsquare: 0.9950 rmse: 7.6482 Thus, the equation for the best curve-fit line is y = 4.382 x -0.7204 (A4.2.2) where y is the rain rate in mm/hr and x is the percentage of occurrence. This procedure is followed for every curve fitting exercise in this study. 173 Appendix 4.3 Rain Cell Diameter from Radar Data diameter SDR1 SDR2 SDR3 SDR4 SDR5 SDR6 SDR7 SDR8 SDR9 SDR10 1 3159 10987 14667 15016 8015 16193 12889 11838 5459 4247 2 371 1858 1755 4480 2676 2432 2978 3062 1063 912 3 134 692 548 1826 1020 831 1022 1270 830 810 4 56 305 216 978 603 297 507 730 316 360 5 22 190 171 693 399 206 333 445 193 185 6 11 75 61 259 262 95 141 172 37 58 7 5 47 25 118 126 34 82 96 82 88 8 2 27 39 149 110 43 62 83 31 32 9 1 4 6 15 19 2 5 28 45 44 10 0 3 1 16 12 3 7 25 18 22 11 2 3 4 1 4 1 1 8 13 26 12 2 3 0 7 5 2 3 14 24 14 13 0 0 1 5 6 0 2 6 10 16 14 0 0 0 11 7 1 0 8 6 9 15 0 1 1 20 3 0 3 10 3 6 16 0 1 0 0 1 0 0 1 2 7 17 0 0 0 0 0 0 0 3 5 6 18 0 0 0 0 2 0 0 1 2 8 total 3765 14196 17495 23594 13270 20140 18035 17800 8139 6850 SDR11 5235 1158 892 364 297 71 103 30 37 22 18 17 11 11 5 6 6 1 8284 SDR12 SDR13 SDR14 SDR15 SDR16 5576 5591 3743 3826 7243 1247 916 748 700 1291 1091 834 703 513 797 583 239 294 187 256 288 185 167 95 177 73 39 45 22 49 121 75 57 37 57 42 18 20 11 13 65 34 36 23 18 34 12 14 8 11 23 13 17 11 10 41 16 16 6 6 18 7 16 10 8 37 10 6 4 6 13 2 5 7 1 21 6 7 1 3 8 2 7 2 2 2 1 2 3 2 9283 8000 5903 5466 9950 SDR17 Total percentage 12977 146661 70.507 2957 30604 14.713 998 14811 7.120 454 6745 3.243 230 4276 2.056 115 1585 0.762 56 1209 0.581 26 738 0.355 11 393 0.189 5 213 0.102 2 157 0.075 2 178 0.086 2 118 0.057 0 116 0.056 2 82 0.039 1 57 0.027 0 41 0.020 0 24 0.012 17838 208008 100.000 174 Appendix 4.4 1 to 10-km path attenuations at 7, 10, and 15 GHz Table A4.4.1a 1-km path attenuation at 7 GHz Attenuation (dB) 0.0000 0.0823 0.6310 1.1135 1.9612 3.4549 6.0862 No. of occurences 4743800 818500 1427 727 378 203 125 Total Percentage 85.2410 14.7070 0.0256 0.0131 0.0068 0.0036 0.0022 5565160 99.9994 att1 vs. percent1 fit 1 0 attenuation (dB) 10 -1 10 b -2 10 fittedmodel1(x) = a*x Coefficients (with 95% confidence bounds): a = 0.02367 (-0.01339, 0.06072) b = -0.8873 (-1.133, -0.6411) 0.01==>1.40835 -3 10 2 R = 0.9815 -2 10 -1 10 0 10 percentage Figure A4.4.1a 1-km path attenuation at 7 GHz 1 10 175 Table A4.4.2a 2-km path attenuation at 7 GHz Attenuation (dB) 0.0000 0.1365 0.6956 1.1944 1.9442 2.1629 2.5923 3.4175 3.6405 4.0859 4.5684 5.4161 6.1611 6.8135 7.1997 No. of occurences 3824100 1193700 2481 1213 526 94 10 267 47 2 2 1 190 3 2 Percentage 76.1380 23.7660 0.0494 0.0242 0.0105 0.0019 0.0002 0.0053 0.0009 0.0000 0.0000 0.0000 0.0038 0.0001 0.0000 Total 5022638 100.0003 0 10 Att2c vs. percentc fit 1 attenuation (dB) -100 10 -200 10 fittedmodel1(x) = a*exp(b*x) Coefficients (with 95% confidence bounds): a = 4.9 (3.769, 6.031) b = -62.32 (-135.4, 10.76) 0.01==>2.62741 -300 10 0 2 R = 0.6576 2 4 6 percentage Figure A4.4.2a 2-km path attenuation at 7 GHz 8 10 176 Table A4.4.3a 3-km path attenuation at 7 GHz Attenuation (dB) 0.0000 0.1822 0.7157 1.2134 1.8766 2.1725 2.6383 3.4139 3.6305 4.1182 4.5942 5.4175 6.1627 6.8052 7.2104 No. of occurences 3131400 1369800 3440 1594 591 232 21 295 109 6 3 2 250 5 5 Total 4507753 Percentage 69.4670 30.3880 0.0763 0.0354 0.0131 0.0051 0.0005 0.0065 0.0024 0.0001 0.0001 0.0000 0.0055 0.0001 0.0001 100.0004 0 10 atten3 vs. percent fit 1 attenuation (dB) -100 10 -200 10 -300 fittedmodel1(x) = a*exp(b*x) Coefficients (with 95% confidence bounds): a = 5.105 (3.941, 6.27) b = -56.12 (-116.6, 4.334) 0.01==>2.91245 2 R = 0.6849 10 0 5 10 percentage Figure A4.4.3a 3-km path attenuation at 7 GHz 15 177 Table A4.4.4a 4-km path attenuation at 7 GHz Attenuation (dB) 0.0000 0.2135 0.7084 1.2174 1.8699 2.1394 2.6496 3.4502 3.6204 4.1653 4.5942 5.4501 6.1606 6.7334 7.2054 No. of occurences 2576600 1436900 3938 1707 598 306 27 300 134 9 3 3 283 7 10 Total 4020825 Percentage 64.0810 35.7370 0.0979 0.0425 0.0149 0.0076 0.0007 0.0075 0.0033 0.0002 0.0001 0.0001 0.0070 0.0002 0.0002 100.0002 0 10 Atten4 vs. percent fit 1 attenuation (dB) -100 10 -200 10 -300 10 fittedmodel1(x) = a*exp(b*x) Coefficients (with 95% confidence bounds): a = 5.166 (3.992, 6.341) b = -48.57 (-98.78, 1.627) 0.01==>3.17858 2 R = 0.6922 0 2 4 6 8 10 12 percentage Figure A4.4.4a 4-km path attenuation at 7 GHz 14 16 18 178 Table A4.4.5a 5-km path attenuation at 7 GHz Attenuation (dB) 0.0000 0.2196 0.6928 1.2104 1.8797 2.1345 2.6507 3.4590 3.5928 4.1773 4.5943 5.4446 5.5973 6.1623 6.7234 7.2450 No. of occurences 2126700 1427300 4239 1741 577 350 28 276 153 11 3 3 1 286 6 12 Total 3561686 Percentage 59.7110 40.0730 0.1190 0.0489 0.0162 0.0098 0.0008 0.0077 0.0043 0.0003 0.0001 0.0001 0.0000 0.0080 0.0002 0.0003 99.9998 0 10 atten5 vs. percent fit 1 attenuation (dB) -100 10 -200 10 fittedmodel1(x) = a*exp(b*x) Coefficients (with 95% confidence bounds): a = 5.268 (4.215, 6.321) b = -45.73 (-88.42, -3.044) 0.01==>3.33466 2 R = 0.7131 -300 10 0 5 10 15 percentage Figure A4.4.5a 5-km path attenuation at 7 GHz 20 179 Table A4.4.6a 6-km path attenuation at 7 GHz Attenuation (dB) 0.0000 0.2198 0.6832 1.2108 1.8934 2.1311 2.6638 3.4622 3.6023 4.1794 4.6068 5.4575 5.5999 6.1633 6.6890 7.2470 No. of occurences 1749900 1372700 4438 1734 536 391 27 242 179 11 3 3 1 283 5 13 Percentage 55.8990 43.8490 0.1418 0.0554 0.0171 0.0125 0.0009 0.0077 0.0057 0.0004 0.0001 0.0001 0.0000 0.0090 0.0002 0.0004 Total 3130466 99.9993 0 10 atten6 vs. percent fit 1 attenuation (dB) -100 10 -200 10 -300 10 fittedmodel1(x) = a*exp(b*x) Coefficients (with 95% confidence bounds): a = 5.293 (4.245, 6.34) b = -41.38 (-79.1, -3.667) 0.01==>3.49907 2 R = 0.7191 0 5 10 15 percentage Figure A4.4.6a 6-km path attenuation at 7 GHz 20 180 Table A4.4.7a 7-km path attenuation at 7 GHz Attenuation (dB) 0.0000 0.2216 0.6780 1.2121 1.8926 2.1226 2.6745 3.4623 3.6081 4.1796 4.6132 5.4599 5.5999 6.1684 6.6869 7.2746 No. of occurences 1434300 1284600 4653 1712 487 439 28 202 202 11 3 3 1 279 6 13 Percentage 52.5960 47.1090 0.1706 0.0628 0.0179 0.0161 0.0010 0.0074 0.0074 0.0004 0.0001 0.0001 0.0000 0.0102 0.0002 0.0005 Total 2726939 99.9998 0 10 atten7 vs. percent fit 1 attenuation (dB) -100 10 -200 10 -300 10 fittedmodel1(x) = a*exp(b*x) Coefficients (with 95% confidence bounds): a =5.31 (4.261, 6.359) b =-37 (-70.42, -3.584) 0.01==>3.66776 2 R = 0.7207 0 5 10 15 percentage Figure A4.4.7a 7-km path attenuation at 7 GHz 20 181 Table A4.4.8a 8-km path attenuation at 7 GHz Attenuation (dB) 0.0000 0.2245 0.6711 1.2117 1.8843 2.1224 2.6828 3.4683 3.6157 4.1872 4.6156 5.4743 5.6121 6.1756 6.6941 7.2785 No. of occurences 1163500 1179600 4910 1657 424 490 29 162 223 11 3 3 1 275 6 13 Percentage 49.4820 50.1690 0.2088 0.0705 0.0180 0.0208 0.0012 0.0069 0.0095 0.0005 0.0001 0.0001 0.0000 0.0117 0.0003 0.0006 Total 2351307 100.0000 0 10 atten8 vs. percent fit 1 attenuation (dB) -100 10 -200 10 fittedmodel1(x) = a*exp(b*x) Coefficients (with 95% confidence bounds): a = 5.307 (4.255, 6.359) b = -32.29 (-61.67, -2.913) 0.01==>3.84259 2 R = 0.7185 -300 10 0 5 10 15 percentage Figure A4.4.8a 8-km path attenuation at 7 GHz 20 182 Table A4.4.9a 9-km path attenuation at 7 GHz Attenuation (dB) 0.0000 0.2281 0.6640 1.2151 1.8795 2.1256 2.6902 3.4694 3.6218 4.1676 4.6079 5.4523 5.5648 6.1813 6.7121 7.2866 No. of occurences 935230 1059900 5133 1602 363 522 30 139 229 9 4 2 2 264 6 13 Percentage 46.6810 52.9040 0.2562 0.0800 0.0181 0.0261 0.0015 0.0069 0.0114 0.0004 0.0002 0.0001 0.0001 0.0132 0.0003 0.0006 Total 2003448 100.0002 0 10 atten9 vs. percent fit 1 attenuation (dB) -100 10 -200 10 fittedmodel1(x) = a*exp(b*x) Coefficients (with 95% confidence bounds): a = 5.28 (4.221, 6.338) b = -27.85 (-53.87, -1.83) 0.01==>3.99616 2 R = 0.7125 -300 10 0 5 10 15 percentage Figure A4.4.9a 9-km path attenuation at 7 GHz 20 183 Table A4.4.10a 10-km path attenuation at 7 GHz Attenuation (dB) 0.0000 0.2293 0.6603 1.2160 1.8651 2.1274 2.6917 3.4650 3.6249 4.1899 4.6326 5.4523 5.5755 6.1894 6.7328 7.2952 No. of occurences 738210 936980 5231 1529 318 530 32 123 234 9 4 2 2 248 6 13 Percentage 43.8500 55.6580 0.3107 0.0908 0.0189 0.0315 0.0019 0.0073 0.0139 0.0005 0.0002 0.0001 0.0001 0.0147 0.0004 0.0008 Total 1683471 99.9999 0 10 atten10 vs. percent fit 1 attenuation (dB) -100 10 fittedmodel1(x) = a*exp(b*x) Coefficients (with 95% confidence bounds): a = 5.285 (4.221, 6.349) b = -24.42 (-47.44, -1.398) -200 10 0.01==>4.14 2 R = 0.7108 -300 10 0 5 10 15 percentage Figure A4.4.10a 10-km path attenuation at 7 GHz 20 184 Table A4.4.1b 1-km path attenuation at 10 GHz Attenuation (dB) 0.0000 0.0978 0.5797 1.0023 1.7288 2.9880 5.1551 8.8949 15.3480 Total No. of occurences 4743800 798660 17019 2818 1427 727 378 203 125 Percentage 85.2410 14.3510 0.3058 0.0506 0.0256 0.0131 0.0068 0.0036 0.0022 5565157 99.9998 1 atten1 vs. percent fit 1 10 attenuation (dB) 0 10 b -1 fittedmodel1(x) = a*x Coefficients (with 95% confidence bounds): a = 0.07532 (0.009986, 0.1407) b = -0.8494 (-0.9867, -0.7121) -2 0.01==>3.7641 10 10 2 R = 0.9899 -2 10 -1 10 0 10 percentage Table A4.4.1b 1-km path attenuation at 10 GHz 1 10 185 Table A4.4.2b 2-km path attenuation at 10 GHz Attenuation (dB) 0.0000 0.1576 0.6552 1.0963 1.7620 2.1266 2.9263 3.1654 3.7790 4.7168 5.2421 5.8554 6.1574 6.8839 8.1430 8.9178 9.2012 9.8973 10.6240 11.8830 14.0500 15.3840 15.7190 16.3510 17.0770 17.7900 18.3360 Total No. of occurences 3824100 1155100 33465 5083 2256 201 609 568 60 10 577 20 13 10 4 247 58 5 2 2 1 171 17 2 2 1 2 5022586 Percentage 76.1380 22.9980 0.6663 0.1012 0.0449 0.0040 0.0121 0.0113 0.0012 0.0002 0.0115 0.0004 0.0003 0.0002 0.0001 0.0049 0.0012 0.0001 0.0000 0.0000 0.0000 0.0034 0.0003 0.0000 0.0000 0.0000 0.0000 99.9998 186 atten2 vs. percent fit 1 1 attenuation (dB) 10 b 0 10 fittedmodel1(x) = a*x Coefficients (with 95% confidence bounds): a = 1.597 (0.1777, 3.016) b = -0.212 (-0.3053, -0.1187) 0.01==>4.23897 2 R = 0.6513 -4 10 -2 10 0 percentage 10 Figure A4.4.2b 2-km path attenuation at 10 GHz 187 Table A4.4.3b 3-km path attenuation at 10 GHz Attenuation (dB) 0.0000 0.2138 0.7081 1.1688 1.7855 2.1921 2.7955 3.1768 3.6812 4.1113 4.7732 5.2670 5.7212 6.2024 6.8719 7.2308 8.1972 8.9350 9.2036 9.8071 10.0110 10.7200 11.9020 12.0770 14.0540 15.4070 15.6640 16.3960 17.1100 17.8090 18.3690 Total No. of occurences 3131400 1315800 46856 7320 2866 405 576 943 109 24 20 685 67 37 18 4 6 263 117 16 2 6 2 1 2 200 45 5 3 2 5 4507805 Percentage 69.4670 29.1890 1.0394 0.1624 0.0636 0.0090 0.0128 0.0209 0.0024 0.0005 0.0004 0.0152 0.0015 0.0008 0.0004 0.0001 0.0001 0.0058 0.0026 0.0004 0.0000 0.0001 0.0000 0.0000 0.0000 0.0044 0.0010 0.0001 0.0001 0.0000 0.0001 100.0004 188 atten3 vs. percent fit 1 1 attenuation (dB) 10 b 0 10 fittedmodel1(x) = a*x Coefficients (with 95% confidence bounds): a = 2.232 (0.5273, 3.937) b = -0.1804 (-0.2659, -0.09495) 0.01==>5.12344 2 R = 0.5620 -4 10 -2 10 0 percentage 10 Figure A4.4.3b 3-km path attenuation at 10 GHz 189 Table A4.4.4b 4-km path attenuation at 10 GHz Attenuation (dB) 0.0000 0.2560 0.7290 1.2047 1.7800 2.2147 2.7699 3.1964 3.6746 4.1515 4.7544 5.2839 5.7127 6.2056 6.8689 7.1616 7.9089 8.3221 8.9294 9.1862 9.7559 10.1180 10.7380 11.8230 12.0770 14.1520 15.4110 15.6900 16.3680 16.6660 17.1290 17.8090 18.3420 18.7400 Total No. of occurences 2576600 1372500 55652 8950 3140 528 491 1133 124 39 25 716 98 43 18 10 1 3 259 151 16 5 8 3 1 3 214 65 4 2 3 2 8 2 4020817 Percentage 64.0810 34.1350 1.3841 0.2226 0.0781 0.0131 0.0122 0.0282 0.0031 0.0010 0.0006 0.0178 0.0024 0.0011 0.0004 0.0002 0.0000 0.0001 0.0064 0.0038 0.0004 0.0001 0.0002 0.0001 0.0000 0.0001 0.0053 0.0016 0.0001 0.0000 0.0001 0.0000 0.0002 0.0000 99.9996 190 atten4 vs. percent fit 1 1 attenuation (dB) 10 b 0 10 fittedmodel1(x) = a*x Coefficients (with 95% confidence bounds): a = 2.52 (0.7578, 4.281) b = -0.1723 (-0.2506, -0.094) 0.01==>5.57111 2 R = 0.5590 -4 10 -2 10 0 percentage 10 Figure A4.4.4b 4-km path attenuation at 10 GHz 191 Table A4.4.5b 5-km path attenuation at 10 GHz Attenuation (dB) 0.0000 0.2772 0.7334 1.2086 1.7694 2.2064 2.7675 3.1932 3.6504 4.1514 4.7520 5.2849 5.6854 6.2301 6.8415 7.1737 7.9101 8.3445 8.9419 9.1612 9.7659 10.0630 10.7770 11.3000 11.8230 12.0780 14.1370 14.5820 15.4150 15.6780 16.3200 16.8350 17.1290 17.8280 18.3600 18.6590 19.0780 Total No. of occurences 2126700 1355700 61822 10043 3292 608 381 1260 144 46 26 715 112 44 19 12 1 3 230 174 18 4 9 1 3 1 3 1 206 73 7 2 3 1 8 3 1 3561676 Percentage 59.7110 38.0630 1.7357 0.2820 0.0924 0.0171 0.0107 0.0354 0.0040 0.0013 0.0007 0.0201 0.0031 0.0012 0.0005 0.0003 0.0000 0.0001 0.0065 0.0049 0.0005 0.0001 0.0003 0.0000 0.0001 0.0000 0.0001 0.0000 0.0058 0.0020 0.0002 0.0001 0.0001 0.0000 0.0002 0.0001 0.0000 99.9997 192 atten5 vs. percent fit 1 1 attenuation (dB) 10 b 0 10 fittedmodel1(x) = a*x Coefficients (with 95% confidence bounds): a = 2.738 (1.037, 4.439) b = -0.1646 (-0.2327, -0.09657) 0.01==>5.84443 2 R = 0.5783 -4 10 -2 10 0 percentage 10 Figure A4.4.5b 5-km path attenuation at 10 GHz 193 Table A4.4.6b 6-km path attenuation at 10 GHz Attenuation (dB) 0.0000 0.2865 0.7303 1.2094 1.7694 2.2035 2.7571 3.2012 3.6605 4.1617 4.7479 5.2920 5.6765 6.2279 6.8271 7.1643 7.9412 8.2957 8.9372 9.1710 9.7061 10.0710 10.7850 11.3010 11.7740 12.0570 14.1750 14.5910 15.4160 15.6700 16.3080 16.8550 17.1440 18.3850 18.6600 19.0780 Total No. of occurences 1749900 1296300 65715 11001 3348 665 322 1306 161 46 28 695 125 49 18 12 2 2 197 192 23 7 9 1 2 2 3 1 187 86 10 2 3 9 3 1 3130433 Percentage 55.8990 41.4100 2.0992 0.3514 0.1070 0.0212 0.0103 0.0417 0.0051 0.0015 0.0009 0.0222 0.0040 0.0016 0.0006 0.0004 0.0001 0.0001 0.0063 0.0061 0.0007 0.0002 0.0003 0.0000 0.0001 0.0001 0.0001 0.0000 0.0060 0.0027 0.0003 0.0001 0.0001 0.0003 0.0001 0.0000 99.9997 194 atten6 vs. percent fit 1 1 attenuation (dB) 10 b 0 10 fittedmodel1(x) = a*x Coefficients (with 95% confidence bounds): a = 2.76 (1.029, 4.491) b = -0.1669 (-0.2381, -0.09567) 0.01==>5.95246 2 R = 0.5666 -4 10 -2 10 0 percentage 10 Figure A4.4.6b 6-km path attenuation at 10 GHz 195 Table A4.4.7b 7-km path attenuation at 10 GHz Attenuation (dB) 0.0000 0.2931 0.7322 1.2133 1.7679 2.1926 2.7385 3.2044 3.6688 4.1652 4.7664 5.2992 5.6692 6.2289 6.8065 7.1495 7.8735 8.3156 8.9434 9.1740 9.6867 10.1320 10.7730 11.3010 11.7740 12.0880 14.1840 14.5910 15.4190 15.6690 16.2940 16.8900 17.1720 18.4430 18.6600 19.0780 Total No. of occurences 1434300 1203400 69428 12215 3411 731 308 1292 190 49 30 653 155 53 16 15 3 2 167 188 35 12 9 1 2 2 3 1 170 95 14 3 3 9 3 1 2726969 Percentage 52.5960 44.1300 2.5460 0.4479 0.1251 0.0268 0.0113 0.0474 0.0070 0.0018 0.0011 0.0239 0.0057 0.0019 0.0006 0.0006 0.0001 0.0001 0.0061 0.0069 0.0013 0.0004 0.0003 0.0000 0.0001 0.0001 0.0001 0.0000 0.0062 0.0035 0.0005 0.0001 0.0001 0.0003 0.0001 0.0000 99.9996 196 atten7 vs. percent fit 1 1 attenuation (dB) 10 b 0 10 fittedmodel1(x) = a*x Coefficients (with 95% confidence bounds): a = 2.867 (1.138, 4.595) b = -0.167 (-0.2372, -0.09685) 0.01==>6.18554 2 R = 0.5689 -4 10 -2 10 0 percentage 10 Figure A4.4.7b 7-km path attenuation at 10 GHz 197 Table A4.4.8b 8-km path attenuation at 10 GHz Attenuation (dB) 0.0000 0.2996 0.7353 1.2151 1.7604 2.1896 2.7316 3.2118 3.6827 4.1832 4.7924 5.3064 5.6804 6.2429 6.7782 7.1737 7.8393 8.1631 8.9463 9.1717 9.6861 10.1440 10.8080 11.3240 11.7930 12.0990 14.2280 14.6290 15.4220 15.6750 16.2480 16.9010 17.2070 18.4570 18.5960 19.0780 Total No. of occurences 1163500 1094600 72089 13432 3501 821 289 1247 217 57 33 610 166 62 15 18 2 3 126 196 45 16 9 1 2 2 3 1 148 109 18 3 3 7 5 1 2351357 Percentage 49.4820 46.5520 3.0659 0.5713 0.1489 0.0349 0.0123 0.0530 0.0092 0.0024 0.0014 0.0259 0.0071 0.0026 0.0006 0.0008 0.0001 0.0001 0.0054 0.0083 0.0019 0.0007 0.0004 0.0000 0.0001 0.0001 0.0001 0.0000 0.0063 0.0046 0.0008 0.0001 0.0001 0.0003 0.0002 0.0000 100.0002 198 atten8 vs. percent fit 1 1 attenuation (dB) 10 b 0 10 fittedmodel1(x) = a*x Coefficients (with 95% confidence bounds): a = 3.028 (1.266, 4.789) b = -0.1642 (-0.2334, -0.09494) 0.01==>6.4482 2 R = 0.5635 -4 10 -2 10 0 percentage 10 Figure A4.4.8b 8-km path attenuation at 10 GHz 199 Table A4.4.9b 9-km path attenuation at 10 GHz Attenuation (dB) 0.0000 0.3059 0.7382 1.2149 1.7515 2.1902 2.7222 3.2194 3.6837 4.1738 4.8129 5.3117 5.6854 6.2674 6.7076 7.1784 7.7892 8.1675 8.9440 9.1774 9.6900 10.1740 10.8270 11.3250 11.8830 12.0800 14.2610 14.6290 15.4220 15.6730 16.2540 16.8350 17.2720 18.4490 18.6280 19.0870 Total No. of occurences 935230 974180 72093 14198 3531 919 287 1187 239 66 31 558 185 64 14 20 3 3 95 203 49 19 8 1 1 3 3 1 121 122 19 5 3 7 5 1 2003474 Percentage 46.6810 48.6240 3.5984 0.7087 0.1762 0.0459 0.0143 0.0592 0.0119 0.0033 0.0015 0.0279 0.0092 0.0032 0.0007 0.0010 0.0001 0.0001 0.0047 0.0101 0.0024 0.0009 0.0004 0.0000 0.0000 0.0001 0.0001 0.0000 0.0060 0.0061 0.0009 0.0002 0.0001 0.0003 0.0002 0.0000 100.0000 200 0 10 atten9 vs. percent fit 1 -100 attenuation (dB) 10 -200 fittedmodel1(x) = a*exp(b*x) Coefficients (with 95% confidence bounds): a = 12.71 (10.79, 14.62) b = -56.25 (-99.77, -12.74) 0.01==>7.23961 -300 R = 0.5669 10 10 2 0 5 percentage 10 Figure A4.4.9b 9-km path attenuation at 10 GHz 15 201 Table A4.4.10b 10-km path attenuation at 10 GHz Attenuation (dB) 0.0000 0.3085 0.7372 1.2137 1.7444 2.1851 2.7147 3.2223 3.6803 4.1980 4.8220 5.3119 5.6890 6.2554 6.7096 7.1986 7.7289 8.2178 8.9452 9.1855 9.7021 10.1940 10.8980 11.1870 11.9210 12.1670 14.2610 14.6960 15.4230 15.6860 16.2200 16.6680 17.1300 17.5380 18.4490 18.6760 19.1250 Total No. of occurences 738210 853550 69554 14467 3498 989 289 1119 250 74 29 504 192 67 17 19 4 3 82 200 53 20 7 2 1 3 3 1 101 123 22 3 4 1 7 5 1 1683474 Percentage 43.8500 50.7020 4.1316 0.8594 0.2078 0.0587 0.0172 0.0665 0.0149 0.0044 0.0017 0.0299 0.0114 0.0040 0.0010 0.0011 0.0002 0.0002 0.0049 0.0119 0.0031 0.0012 0.0004 0.0001 0.0001 0.0002 0.0002 0.0001 0.0060 0.0073 0.0013 0.0002 0.0002 0.0001 0.0004 0.0003 0.0001 99.9999 202 atten10 vs. percent fit 1 1 attenuation (dB) 10 b fittedmodel1(x) = a*x Coefficients (with 95% confidence bounds): a = 3.031 (1.402, 4.661) b = -0.1737 (-0.2395, -0.1079) 0 10 0.01==>6.74703 2 R = 0.6055 -4 10 -2 10 0 percentage 10 Figure A4.4.10b 10-km path attenuation at 10 GHz 203 Table A4.4.1c 1-km path attenuation at 15 GHz Attenuation (dB) 0.0000 0.1277 0.6937 1.3952 2.2743 3.6992 6.0279 9.8071 15.9570 25.9650 Total No. of occurences 4743800 720910 77745 17019 2818 1427 727 378 203 125 5565152 atten1 vs. percent1 fit 1 1 10 attenuation (dB) Percentage 85.2410 12.9540 1.3970 0.3058 0.0506 0.0256 0.0131 0.0068 0.0036 0.0022 99.9998 0 10 b -1 10 fittedmodel1(x) = a*x Coefficients (with 95% confidence bounds): a = 0.235 (0.06574, 0.4043) b = -0.7504 (-0.8666, -0.6342) 0.01==>7.44543 2 R = 0.9890 -2 10 -2 10 -1 10 0 10 percentage 1 10 Figure A4.4.1c 1-km path attenuation at 15 GHz 204 Table A4.4.2c 2-km path attenuation at 15 GHz Attenuation (dB) 0.0000 0.1982 0.7266 1.2982 1.6960 2.3348 2.7300 3.1355 3.7581 4.1248 4.5545 5.0945 5.9735 6.1374 6.7216 7.4108 8.3022 9.8339 10.1580 10.6680 11.2020 12.0690 13.5060 15.9360 16.2080 16.8190 17.3520 18.2320 19.6560 21.9850 25.9240 26.2160 26.8260 27.3600 28.2390 29.6640 31.9540 Total No. of occurences 3824100 1021500 126470 32154 6905 5430 991 160 2000 256 143 58 24 1003 140 49 21 450 107 30 14 19 10 153 129 14 13 5 2 2 111 70 6 2 2 2 3 5022548 Percentage 76.1380 20.3390 2.5180 0.6402 0.1375 0.1081 0.0197 0.0032 0.0398 0.0051 0.0028 0.0012 0.0005 0.0200 0.0028 0.0010 0.0004 0.0090 0.0021 0.0006 0.0003 0.0004 0.0002 0.0030 0.0026 0.0003 0.0003 0.0001 0.0000 0.0000 0.0022 0.0014 0.0001 0.0000 0.0000 0.0000 0.0001 100.0000 205 atten2 vs. percent2 fit 1 1 attenuation (dB) 10 b fittedmodel1(x) = a*x Coefficients (with 95% confidence bounds): a = 2.04 (0.5344, 3.545) b = -0.2516 (-0.3336, -0.1695) 0 10 0.01==>6.4967 2 R = 0.6679 -4 10 -2 10 0 percentage 10 Figure A4.4.21c 2-km path attenuation at 15 GHz 206 Table A4.4.3c 3-km path attenuation at 15 GHz Attenuation (dB) 0.0000 0.2556 0.7554 1.2448 1.7141 2.3179 2.7383 3.2026 3.8073 4.2044 4.6493 5.2202 5.8675 6.2190 6.7507 7.3211 7.6662 8.3518 8.6942 9.1634 9.8582 10.1800 10.7240 11.2570 11.5950 12.1930 12.7160 13.4510 13.5800 14.2710 14.9020 15.9320 16.2160 16.7320 17.3130 17.7780 18.2800 18.5560 19.8020 20.5180 21.9850 22.1090 22.5110 25.9210 26.2300 26.7300 27.3130 No. of occurences 3131400 1143700 156650 43962 14440 8037 2388 668 2351 607 282 133 39 1172 240 114 40 31 11 5 443 216 73 23 11 39 4 1 16 4 1 118 203 41 28 2 10 2 5 1 1 1 1 92 124 21 10 Percentage 69.4670 25.3720 3.4751 0.9752 0.3203 0.1783 0.0530 0.0148 0.0522 0.0135 0.0063 0.0030 0.0009 0.0260 0.0053 0.0025 0.0009 0.0007 0.0002 0.0001 0.0098 0.0048 0.0016 0.0005 0.0002 0.0009 0.0001 0.0000 0.0004 0.0001 0.0000 0.0026 0.0045 0.0009 0.0006 0.0000 0.0002 0.0000 0.0001 0.0000 0.0000 0.0000 0.0000 0.0020 0.0028 0.0005 0.0002 207 28.3340 28.5640 29.7650 31.9540 32.1160 Total 4 1 3 4 3 4507776 0.0001 0.0000 0.0001 0.0001 0.0001 100.0006 atten3 vs. percent3 fit 1 1 attenuation (dB) 10 b 0 10 fittedmodel1(x) = a*x Coefficients (with 95% confidence bounds): a= 4.203 (1.898, 6.509) b = -0.1606 (-0.2208, -0.1005) 0.01==>8.80735 2 R = 0.5157 -4 10 -2 0 10 10 percentage Figure A4.4.3c 3-km path attenuation at 15 GHz 208 Table A4.4.4c 4-km path attenuation at 15 GHz Attenuation (dB) 0.0000 0.2992 0.7596 1.2424 1.7258 2.2887 2.7388 3.2302 3.8058 4.2133 4.6861 5.2578 5.7755 6.2404 6.7347 7.2631 7.6824 8.3353 8.7025 9.1938 9.8719 10.2130 10.7260 11.2790 11.6240 12.2430 12.5870 13.4840 13.6580 14.2950 14.7330 15.9590 16.2240 16.7480 17.2660 17.6600 18.2760 18.6650 19.8090 20.1820 20.5230 21.9850 22.0820 22.5110 25.9560 26.2250 26.6750 No. of occurences 2576600 1175900 173070 52139 20906 9970 3761 1152 2505 814 372 171 53 1195 301 123 64 34 18 9 417 272 95 29 19 46 5 1 17 8 2 97 227 54 37 5 10 4 6 1 2 1 2 1 78 152 33 Percentage 64.0810 29.2460 4.3044 1.2967 0.5199 0.2480 0.0935 0.0287 0.0623 0.0202 0.0093 0.0043 0.0013 0.0297 0.0075 0.0031 0.0016 0.0008 0.0004 0.0002 0.0104 0.0068 0.0024 0.0007 0.0005 0.0011 0.0001 0.0000 0.0004 0.0002 0.0000 0.0024 0.0056 0.0013 0.0009 0.0001 0.0002 0.0001 0.0001 0.0000 0.0000 0.0000 0.0000 0.0000 0.0019 0.0038 0.0008 209 27.2590 27.5430 28.3010 28.7650 29.4250 29.8240 31.9540 32.1130 32.9000 33.0550 Total 18 1 4 1 1 3 3 6 1 2 4020818 0.0004 0.0000 0.0001 0.0000 0.0000 0.0001 0.0001 0.0001 0.0000 0.0000 100.0002 atten4 vs. percent4 fit 1 1 attenuation (dB) 10 b 0 10 fittedmodel1(x) = a*x Coefficients (with 95% confidence bounds): a= 3.975 (2.011, 5.939) b = -0.1786 (-0.2324, -0.1249) 0.01==>9.04752 2 R = 0.6012 -4 10 -2 10 0 percentage 10 Figure A4.4.4c 4-km path attenuation at 15 GHz 210 Table A4.4.5c 5-km path attenuation at 15 GHz Attenuation (dB) 0.0000 0.3210 0.7597 1.2469 1.7297 2.2686 2.7395 3.2315 3.7865 4.2164 4.7129 5.2527 5.7175 6.2475 6.7180 7.2457 7.6962 8.2861 8.6820 9.1896 9.8769 10.1980 10.7320 11.2650 11.6290 12.2510 12.6450 13.2680 13.6740 14.3200 14.7880 15.9620 16.2510 16.7260 17.2310 17.6740 18.2790 18.6510 19.8410 20.1280 20.5370 21.3840 21.9850 22.0820 22.5170 25.9550 26.2360 No. of occurences 2126700 1145000 180270 58188 25928 11522 4841 1593 2511 1027 422 199 54 1152 344 149 82 35 24 12 359 324 105 35 23 44 8 1 18 9 3 66 241 63 38 6 11 5 3 3 3 1 1 2 1 45 175 Percentage 59.7110 32.1460 5.0613 1.6337 0.7280 0.3235 0.1359 0.0447 0.0705 0.0288 0.0118 0.0056 0.0015 0.0323 0.0097 0.0042 0.0023 0.0010 0.0007 0.0003 0.0101 0.0091 0.0029 0.0010 0.0006 0.0012 0.0002 0.0000 0.0005 0.0003 0.0001 0.0019 0.0068 0.0018 0.0011 0.0002 0.0003 0.0001 0.0001 0.0001 0.0001 0.0000 0.0000 0.0001 0.0000 0.0013 0.0049 211 26.6790 27.2770 27.7200 28.2940 29.4310 29.7740 31.9930 32.1940 32.7880 33.0550 33.9160 Total 43 19 2 6 1 4 2 6 2 2 1 3561734 0.0012 0.0005 0.0001 0.0002 0.0000 0.0001 0.0001 0.0002 0.0001 0.0001 0.0000 100.0001 atten5 vs. percent5 fit 1 1 attenuation (dB) 10 b 0 10 fittedmodel1(x) = a*x Coefficients (with 95% confidence bounds): a= 3.975 (2.061, 5.89) b = -0.1845 (-0.238, -0.1309) 0.01==>9.297 2 R = 0.6182 -4 10 -2 10 0 percentage 10 Figure A4.4.5c 5-km path attenuation at 15 GHz 212 Table A4.4.6c 6-km path attenuation at 15 GHz Attenuation (dB) 0.0000 0.3326 0.7616 1.2490 1.7361 2.2573 2.7401 3.2300 3.7715 4.2240 4.7167 5.2435 5.7143 6.2557 6.7269 7.2376 7.6821 8.2615 8.6793 9.2106 9.8862 10.2020 10.7490 11.2300 11.6560 12.2460 12.6690 13.2090 13.7260 14.2630 14.7030 15.9720 16.2540 16.7230 17.2250 17.6530 18.2980 18.7490 19.8420 20.1730 20.5740 21.3900 21.9850 22.2450 22.5220 25.9730 26.2280 No. of occurences 1749900 1080200 180850 61507 29476 12825 5817 2009 2448 1227 456 221 68 1064 390 165 98 43 24 13 301 351 123 40 30 42 11 2 17 6 5 40 236 75 41 12 12 7 3 3 3 1 1 2 1 29 170 Percentage 55.8990 34.5060 5.7770 1.9648 0.9416 0.4097 0.1858 0.0642 0.0782 0.0392 0.0146 0.0071 0.0022 0.0340 0.0125 0.0053 0.0031 0.0014 0.0008 0.0004 0.0096 0.0112 0.0039 0.0013 0.0010 0.0013 0.0004 0.0001 0.0005 0.0002 0.0002 0.0013 0.0075 0.0024 0.0013 0.0004 0.0004 0.0002 0.0001 0.0001 0.0001 0.0000 0.0000 0.0001 0.0000 0.0009 0.0054 213 26.6750 27.2740 27.8070 28.3060 28.5170 29.6980 30.0200 32.2590 32.7910 33.0550 33.9160 Total 56 21 3 7 1 4 1 8 2 2 1 3130471 0.0018 0.0007 0.0001 0.0002 0.0000 0.0001 0.0000 0.0003 0.0001 0.0001 0.0000 100.0000 atten6 vs. percent6 fit 1 1 attenuation (dB) 10 b 0 10 fittedmodel1(x) = a*x Coefficients (with 95% confidence bounds): a= 3.984 (2.179, 5.788) b = -0.1885 (-0.2398, -0.1373) 0.01==>9.492 2 R = 0.6425 -4 10 -2 10 0 percentage 10 Figure A4.4.6c 6-km path attenuation at 15 GHz 214 Table A4.4.7c 7-km path attenuation at 15 GHz Attenuation (dB) 0.0000 0.3400 0.7661 1.2492 1.7390 2.2529 2.7403 3.2351 3.7547 4.2321 4.7115 5.2366 5.7215 6.2723 6.7241 7.2266 7.7021 8.2741 8.6551 9.2110 9.8961 10.2070 10.7610 11.2450 11.6930 12.2590 12.6870 13.2130 13.7570 14.2420 14.6770 15.8960 16.2580 16.7070 17.2280 17.7050 18.3140 18.8240 19.3100 19.8840 20.2170 20.5990 21.3900 21.9850 22.0550 22.5790 25.9700 No. of occurences 1434300 986480 177220 64289 32618 14434 6959 2540 2423 1427 511 252 87 955 429 195 113 53 29 14 240 370 143 45 40 39 16 3 15 7 6 25 208 87 38 22 15 10 2 3 3 3 1 1 1 2 20 Percentage 52.5960 36.1750 6.4987 2.3575 1.1961 0.5293 0.2552 0.0931 0.0889 0.0523 0.0187 0.0092 0.0032 0.0350 0.0157 0.0072 0.0041 0.0019 0.0011 0.0005 0.0088 0.0136 0.0052 0.0017 0.0015 0.0014 0.0006 0.0001 0.0006 0.0003 0.0002 0.0009 0.0076 0.0032 0.0014 0.0008 0.0006 0.0004 0.0001 0.0001 0.0001 0.0001 0.0000 0.0000 0.0000 0.0001 0.0007 215 26.2390 26.7100 27.1990 27.7610 28.3340 28.6770 29.7750 30.0970 32.3270 32.6820 33.0550 33.9160 Total 166 51 31 5 7 3 5 1 9 1 2 1 2726975 0.0061 0.0019 0.0011 0.0002 0.0003 0.0001 0.0002 0.0000 0.0003 0.0000 0.0001 0.0000 99.9993 atten7 vs. percent7 fit 1 1 attenuation (dB) 10 b 0 10 fittedmodel1(x) = a*x Coefficients (with 95% confidence bounds): a= 4.188 (2.4, 5.976) b = -0.1871 (-0.2365, -0.1378) 0.01==>9.91478 2 R = 0.6501 -4 10 -2 10 0 percentage 10 Figure A4.4.7c 7-km path attenuation at 15 GHz 216 Table A4.4.8c 8-km path attenuation at 15 GHz Attenuation (dB) 0.0000 0.3447 0.7700 1.2484 1.7411 2.2473 2.7398 3.2337 3.7480 4.2375 4.7086 5.2276 5.7223 6.2849 6.7221 7.2149 7.7040 8.2392 8.6788 9.2542 9.8832 10.2280 10.7460 11.2270 11.7350 12.3050 12.7350 13.1570 13.7840 14.2190 14.7260 15.8970 16.2700 16.7290 17.2200 17.6780 18.3470 18.7610 19.2940 19.8290 20.1320 20.6010 21.4670 21.9850 22.1790 22.6180 25.9790 No. of occurences 1163500 883020 168460 65524 35256 15923 8079 3162 2453 1578 624 279 119 814 453 235 126 63 40 16 177 386 152 50 47 38 24 5 13 7 9 14 179 90 40 30 18 12 5 2 3 4 1 1 1 2 13 Percentage 49.4820 37.5540 7.1646 2.7867 1.4994 0.6772 0.3436 0.1345 0.1043 0.0671 0.0265 0.0119 0.0051 0.0346 0.0193 0.0100 0.0054 0.0027 0.0017 0.0007 0.0075 0.0164 0.0065 0.0021 0.0020 0.0016 0.0010 0.0002 0.0006 0.0003 0.0004 0.0006 0.0076 0.0038 0.0017 0.0013 0.0008 0.0005 0.0002 0.0001 0.0001 0.0002 0.0000 0.0000 0.0000 0.0001 0.0006 217 26.2480 26.7330 27.2130 27.6910 28.2570 28.7010 29.7610 30.1820 32.3700 32.6250 33.0880 33.9160 Total 152 56 36 10 8 4 4 2 7 3 2 1 2351332 0.0065 0.0024 0.0015 0.0004 0.0003 0.0002 0.0002 0.0001 0.0003 0.0001 0.0001 0.0000 99.9996 atten8 vs. percent8 fit 1 1 attenuation (dB) 10 b 0 10 fittedmodel1(x) = a*x Coefficients (with 95% confidence bounds): a= 4.337 (2.528, 6.146) b = -0.1899 (-0.2399, -0.1398) 0.01==>10.3971 2 R = 0.6535 -4 10 -2 10 0 percentage 10 Figure A4.4.8c 8-km path attenuation at 15 GHz 218 Table A4.4.9c 9-km path attenuation at 15 GHz Attenuation (dB) 0.0000 0.3485 0.7732 1.2508 1.7417 2.2474 2.7393 3.2347 3.7440 4.2366 4.7191 5.2191 5.7266 6.2876 6.7310 7.2277 7.7093 8.2615 8.7040 9.2449 9.8820 10.2460 10.7450 11.2010 11.7530 12.3200 12.7280 13.1230 13.7700 14.2140 14.7620 15.2650 15.9620 16.2690 16.7420 17.1910 17.6730 18.3110 18.7740 19.2470 19.7100 20.0430 20.6100 21.4730 21.9850 No. of occurences 935230 773440 156130 64729 35992 16744 8907 3655 2543 1657 720 308 164 694 470 267 135 71 51 16 136 363 157 69 53 30 33 9 7 12 7 3 9 150 89 48 31 23 13 6 3 2 5 1 1 Percentage 46.6810 38.6050 7.7928 3.2308 1.7965 0.8358 0.4446 0.1824 0.1269 0.0827 0.0359 0.0154 0.0082 0.0346 0.0235 0.0133 0.0067 0.0035 0.0025 0.0008 0.0068 0.0181 0.0078 0.0034 0.0026 0.0015 0.0016 0.0004 0.0003 0.0006 0.0003 0.0001 0.0004 0.0075 0.0044 0.0024 0.0015 0.0011 0.0006 0.0003 0.0001 0.0001 0.0002 0.0000 0.0000 219 22.6200 23.0410 25.9790 26.2470 26.7310 27.2210 27.7280 28.2110 28.7550 29.8590 30.0670 30.8240 32.3450 32.7180 33.2130 33.9500 Total 2 1 11 127 64 42 9 10 5 4 1 1 7 3 2 1 2003473 0.0001 0.0000 0.0005 0.0063 0.0032 0.0021 0.0004 0.0005 0.0002 0.0002 0.0000 0.0000 0.0003 0.0001 0.0001 0.0000 100.0004 atten9 vs. percent9 fit 1 1 attenuation (dB) 10 b 0 fittedmodel1(x) = a*x Coefficients (with 95% confidence bounds): a= 4.538 (2.769, 6.307) b = -0.1874 (-0.2344, -0.1405) 10 0.01==>10.7589 2 R = 0.6649 -4 10 -2 10 0 percentage 10 Figure A4.4.9c 9-km path attenuation at 15 GHz 220 Table A4.4.10c 10-km path attenuation at 15 GHz Attenuation (dB) 0.0000 0.3531 0.7747 1.2546 1.7446 2.2479 2.7388 3.2332 3.7439 4.2351 4.7204 5.2035 5.7344 6.2956 6.7276 7.2137 7.7116 8.2581 8.7183 9.2250 9.8600 10.2450 10.7370 11.2150 11.7480 12.3090 12.6970 13.1740 13.6670 14.2920 14.8090 15.2200 15.9020 16.2760 16.7480 17.1920 17.6920 18.3000 18.6930 19.2370 19.6900 20.2050 20.6210 21.4730 21.5640 22.1090 22.5170 23.2460 No. of occurences 738210 666800 142230 61850 35375 16976 9275 4021 2588 1698 785 330 204 602 464 273 159 71 53 21 103 341 155 81 53 32 35 11 7 12 7 3 8 134 83 53 34 23 16 7 3 2 5 1 1 1 1 2 Percentage 43.8500 39.6090 8.4488 3.6740 2.1013 1.0084 0.5509 0.2389 0.1537 0.1009 0.0466 0.0196 0.0121 0.0358 0.0276 0.0162 0.0094 0.0042 0.0031 0.0012 0.0061 0.0203 0.0092 0.0048 0.0031 0.0019 0.0021 0.0007 0.0004 0.0007 0.0004 0.0002 0.0005 0.0080 0.0049 0.0031 0.0020 0.0014 0.0010 0.0004 0.0002 0.0001 0.0003 0.0001 0.0001 0.0001 0.0001 0.0001 221 25.9800 26.2490 26.7260 27.2330 27.6640 28.1280 28.7690 29.8920 30.1910 31.0250 32.3450 32.6610 33.2880 34.0740 Total 8 106 65 45 10 12 6 2 3 1 7 2 3 1 1683470 0.0005 0.0063 0.0039 0.0027 0.0006 0.0007 0.0004 0.0001 0.0002 0.0001 0.0004 0.0001 0.0002 0.0001 100.0000 atten10 vs. percent10 fit 1 1 attenuation (dB) 10 b fittedmodel1(x) = a*x Coefficients (with 95% confidence bounds): a= 4.829 (3.016, 6.643) b = -0.1834 (-0.2295, -0.1372) 0.01==>11.2355 2 R = 0.6607 0 10 -4 10 -2 10 0 percentage 10 Figure A4.4.10c 10-km path attenuation at 15 GHz 222 Appendix 4.5 Rec. ITU-R P.838-1 Regression coefficients for estimating specific attenuation Frequency (GHz) 1 2 4 6 7 8 10 12 15 20 25 30 35 40 45 50 60 70 80 90 100 120 150 200 300 400 k 0.0000352 0.000138 0.000591 0.00155 0.00265 0.00395 0.00887 0.0168 0.0335 0.0691 0.113 0.167 0.233 0.310 0.393 0.479 0.642 0.784 0.906 0.999 1.06 1.13 1.27 1.42 1.35 1.31 α 0.880 0.923 1.075 1.265 1.312 1.310 1.264 1.200 1.128 1.065 1.030 1.000 0.963 0.929 0.897 0.868 0.824 0.793 0.769 0.754 0.744 0.732 0.711 0.690 0.689 0.684