The Rainfall Estimation using Remote Sensing in Thailand Miss Preeyaphorn Kosa Ph.D. Student, Department of Water Resource Engineering, Kasetsart University, Bangkok 10900, Thailand. E-mail: P.Kosa@cgiar.org Dr. Kobkiat Pongput Associate Professor, Department of Water Resource Engineering, Kasetsart University, Bangkok 10900, Thailand. E-mail: firstname.lastname@example.org ABSTRACT: Rainfall consists of both temporal and spatial variability. Rain gauge supports temporal resolution, on the other hand it is weakness in the quality of spatial resolution. However, remote sensing provides good spatial resolution. According to the advantages of these measurement methods, more precious rainfall data should be estimated from the combination of existent rain gauge and remote sensing observation. The methodology of this paper is review the measurement and evaluation of rainfall, which are well known, using both rain gauge and remote sensing observation. The results are that infrared thermal and radar have the capability in order to estimate rainfall for Thailand. Infrared thermal band is used for the estimation of rainfall by the relationship of cloud top temperature from satellite image and rainfall from rain gauge. Radar band is served to calculate rainfall from the relationship between radar-measure reflectivity and rain rate. KEYWORDS: Rainfall Estimation, Infrared Thermal, Radar 1. Introduction Both the temporal and the spatial variability are the elements of rainfall estimation. Temporal variability is caused by the climatic factors governing rainfall: solar radiation, wind, atmospheric humidity, land and seas surface temperatures and global circulation patterns. Ordinary spatial variability is because of (a) landscape topography and (b) the climatic variables listed above . These climatic parameters (including rainfall) can be recorded by meteorological stations on the ground and by various forms of remote sensing from satellites. At present, rainfall is estimated by using rain gauge and remote sensing observation. Rain gauge supports temporal resolution on the other hand it is weakness in the quality of spatial resolution. However, remote sensing provides good spatial resolution. Therefore, the combination of rain gauges and remote sensing should be used to estimate rainfall. The purpose of this paper is so as to improve rainfall estimation by using both rain gauge and remote sensing observation. Since this paper is literature review of rainfall estimation, the results from this paper are guideline to estimate rainfall. These results have to be utilized in study area to modify the results. Thereafter, the detail of these results may be improved if they are not suitable for rainfall estimation in Thailand. 2. Methodology The methodology of this paper is review the measurement and evaluation of rainfall, which are well known, using both rain gauge and remote sensing observation. Thereafter, they are analyzed to create methods that are related to existent rain gauge and meteorological satellites. The detail of review is explained as following. 2.1 The Nature of Rainfall Zone There are three climatic zones on the world: polar, equatorial, and the mid-latitude zone. First, the polar zone locates between 600N to 900N and 600S to 900S. Second, the equatorial zone is from 300N to 300S. The tropical zone ranges between 23.50N to 23.50S of the equator. Last, the mid-latitude zone or temperature zone locates between 300N to 600N and 300S to 600S. , , ,  Patterns of rainfall change with season, latitude, and the movement of the sun. In temperate zones, there is the annual rainfall about 100 inches or 2,500 millimeters on western facing coastlines, and decreases within the interiors of major landmasses. The tropical zone, which has strong heating from the sun, has the highest rainfall. Since twothirds of global rainfall occurs in the tropical zone, this zone has great importance for the global hydrological cycle. The process of rainfall, which releases the latent heat of condensation, transfers heat energy to the atmosphere. Three-fourths of the atmosphere’s heat energy comes from the tropical zone. The total of average annual rainfall is usually more than 100 inches or 2,500 millimeters, and can increase to 400 inches or 10,000 millimeters if it is influenced by the monsoons. Since Thailand locates on the tropical zone, annual average rainfall is more than 1,000 mm. In the rainy season, heavy rainfalls often occur. The duration and volume of heavy rainfalls are often unknown. Then, flood damage can be high in some areas, because flood forecasting has not been implemented for the whole area of Thailand. For this reason, rain gauges should cover all of Thailand and should have high efficiency for rainfall estimation. Remote sensing technology can play a key role in improving rainfall estimates. 2.2 Rainfall Estimation Using Rain Gauge Rain gauge is applied for the estimation of average daily rainfall. Ideally, an adequate density and distribution of rain gauges have to be in any study area, but it is rare in many part of the world. Spatial interpolation techniques are utilized for spatial averaged rainfall over a well-gauged area. Interpolation techniques can allow the estimation of rainfall in an un-gauged area, which is surrounded by sufficiently wellsited rainfall stations, but, due to insufficient density and distribution of rain gauges, the estimation of average rainfall is often in error.  As far as possible, rain gauges are sited in representative locations although it is common to find significant variation even over a separation distance of a few kilometers . There are three main types of rain gauges: non-recording gauges, weighing gauges, and tipping bucket gauges. First, non-recording gauges are obtained by a small graduated wedge or cylinder that is placed at some height above the ground. These gauges have to be read manually once or twice per day. Second, a weighing gauge includes a collection bucket, and the depth of water is estimated from the weight in the bucket. Therefore, rainfall is calculated from the change of water depth. Last, a tipping bucket gauge consists of a funnel and a tipping bucket that is calibrated so that a number of tips correspond to 1 mm of rainfall (typically 4 – 10). This gauge estimates rainfall by using the number of tips in a given period of time and is usually equipped with a built in data logger  2.3 Rainfall Estimation Using Remote Sensing Remote sensing is new technology for rainfall estimation, because rainfall is estimated by rain gauge from past to present. The number of technologies has been created to estimate rainfall, and most of these technologies are based on the detection of rain clouds and determine rainfall using a variety of more or less refined algorithm. At present, wavelengths of microwave and infrared are often applied for rainfall estimation. In microwave channel, groundbased radar can measure rainfall information at spatial and temporal scales suitable for such applications. Ground-based radar has become an important tool for meteorologists over the past 20 years. Radar can continuously scan an area about 200 km in radius day and night and in all weather conditions. The major limitation is that its field of view can be blocked by high terrain. There are three parts to work: (1) an antenna or receiver, (2) computers that process the raw radar data, and (3) an interactive workstation the forecasters can use to display the processed radar data. Ground-based radar detects information at different scanning elevations and patterns, in addition to varying pulse frequencies. Radar sends out a pulse, and listens for a return signal. The return observed in a radar sounding will be computed from the shape of the radar antenna, the wavelength, frequency and energy transmitted. Radar based reflectivity images are normally presented by color in dbZ, which is a logarithmic scale of the target’s reflectivity. The reflectivity, which is received by radar, can present precipitation type, storm physics, and a good sense of the intensity of storm patterns. The properties of Z-R relationship (where radar-measured reflectivity is Z and rain rate is R) relates to the reflectivity of radar, so this relationship is also suitable for analysis rainfall using ground-based radar , , . To estimate rainfall, near infrared and thermal infrared are utilized in infrared channel. Rain drops cannot be sensed directly using near infrared wavebands. This wavelength is completely filtered by cloud, but cannot detect aerosols easily. However, the thermal infrared bands provide very good detection of cloud and aerosols (and are therefore used in cloud filtering and correction algorithms for NIR bands). The thermal band will of course give good estimates of the temperature at the top of a cloud. Thus, one needs to relate the independent measurements of rainfall to the properties of a cloud measured by infrared remote sensing. Cloud temperatures are used for rainfall calculation. For the infrared channel, cloud and the earth’s surface reflect in this channel, and the intensity of radiance depends on the temperature of the radiation source. The higher the temperature, the more intense the infrared radiation. High clouds are invariably very cold, while low clouds are much warmer . Therefore, the intensity of the infrared radiation in this channel can infer both the temperature and the height of the cloud top. The characteristic of cloud can be classified by shape and cloud height. Stratus and cumulus are the types of cloud which are grouped by shape of cloud. The characteristic of a stratus cloud is being layered or sheetlike, while a cumulus cloud is puffy and heaped. Cloud height can be classified in three levels - high, middle, and low , . The temperature of the atmosphere above the Earth decreases with altitude, at a rate of - 7 degrees Celsius per kilometer. A cloud’s altitude is given as “ceiling”, which is the measurement at the top of cloud and is part of the set of upper air measurements . Table 1 Relationship between cloud height and cloud temperature. Table 1 Relationship between cloud height and cloud temperature Cloud Height (meters) 0 1,250 2,500 3,750 5,000 6,250 7,500 8,750 10,000 11,250 Temperature (0 Celsius) 15 8 -1 -6 -13 -24 -34 -41 -50 -57 Satellite image is represented by various shapes of gray after satellite measures the temperature of clouds. Whiter areas are the coldest part, while darker areas are the warmest part. The apparent temperature of clouds depends on many factors that include cloud height, its size and make-up of the particles that form the cloud . 2.4 Rainfall Measurement in Thailand Presently, rainfall is measured by rain gauge, while remote sensing technology is used to detect meteorological information for weather forecasting. However, researchers are attempting to create suitable method for rainfall estimation using remote sensing observation. The equipment used at a climate station includes: Thermometer screen, MaximumMinimum Thermometer, Dry-Wet Bulb Psychrometer, Thermo-Hygrograph, American Class A Evaporation Pan, and Rain gauge. Many rainfall stations have some of these instruments as well. Daily rainfall is measured by the Thailand Meteorological Department (TMD), Royal Irrigation Department (RID), and Royal Forest Department (RFD) . RID and RFD normally use rain gauges, while TMD uses rain gauges to measure rainfall and uses ground-based radar to collect intensity of rainfall and meteorological information. For ground based radar, X-Band, C-Band, and SBand are used to measure the intensity of rainfall. The frequencies of these bands are 6,200-10,900 MHz, 3,900-6,200 MHz, and 1,550-3,900 MHz, respectively. The wavelengths of these bands are 2.75-4.84 cm, 4.84-7.69 cm, and 7.69-19.3 cm, respectively. TMD’s ground-based radar stations are located in Chiang Rai, Mea Hong Son, Chiang Mai, Phitsanulok, Sakon Nakhon, Khon Khaen, Ubon Ratchathani, Surin, Don Muang, Rayong, Hua Hin, Sum Porn, Surat Thani, Phuket, Ranong, Trung, Song Kha, Nara Thiwas, and Thailand Meteorological Department. There is also a mobile radar unit. These stations nearly cover all of Thailand. Ground-based radar is measured every hour during regular weather conditions and every 30 minutes in unusual weather conditions. Also, it is used to determine cloud movement, and the location of rainfall . Meteorological satellites, which are used by TMD, include GMS-5. This satellite is used for meteorology information by remote sensing and direct broadcast, such as cloud top temperature, cloud movement, temperature, sea surface temperature and Earth surface temperature, radiation from sun, moisture, ozone, and transfer information to member or user. The bands of GMS 5 include water vapour band (6.5 – 7.0 micron) and infrared band (10.5 – 11.5 and 11.5 – 12.5 micron) . 3. Result According to the review of rainfall estimation using rain gauge and remote sensing in Thailand, the study results are that rain gauge and remote sensing technology, which is in the wavelengths of infrared and radar, should be combined to estimate rainfall. Rain gauge measures rainfall in the point of rain gauge station, and rainfall between rain gauge stations are interpolated by using spatial interpolation technology that errors often occur. Normally, rain gauges are located in everywhere except in mountain and forest, where are primary water resource. While remote sensing technology help estimate rainfall overall area, there are less errors than spatial interpolation technology. Infrared thermal band is used for the estimation of rainfall by a non-linear equation ( P aT b ), where T is cloud top temperature (K) from GMS satellite image and P is rainfall (mm) from rain gauge. In the target area, cloud top temperature and rainfall are collected, and they are plotted to find a and b coefficients. Thereafter, the equation between P and T is created for each characteristic area. This equation is used to predict rainfall in that area when cloud top temperature is determined from GMS satellite imagery. The confidence of this equation can be defined from the coefficient of determination, or r 2 . The selected a and b parameter are importance for the accuracy of created equation. Addition to the a and b parameter, P and T are also significance, because all of these parameters have influence to the coefficient of determination. Thereafter, this equation should be calibrated by using rain gauge. When temperature from GMS satellite is applied with P-T equation, rainfall from this equation is calculated. Then, rainfall from rain gauge and this equation are compared. If there are many different values, the a and b coefficients have to be adjusted. There are two researches about above equation in Thailand. First, the cloud top temperature from GMS 4 and rainfall from ground-based rainfall stations (20 stations) are used to calculate a non-linear regression equation of relationship between cloud top temperature (T) and rainfall (P) in Northern Thailand. The equations are 58 24.80 2 P 4.3932 10 T ( r = 0.97) in the dry season (March – May 1993), and P 1.8462 10 22 T 8.93 ( r 2 = 0.62) in the rainy season (August – September 1994) . Secondary, the equations for Northern, Northeastern, Eastern, and Southern Thailand from January to December 1994 are: P 1.1102 1013 T 4.8288 , P 4.4905 1010 T 3.8012 , P 5.2384 1012 T 4.6683 , and 11 4.0840 P 2.4916 10 T , respectively, with coefficients of determination ( r 2 ) being 0.71, 0.73, 0.70, and 0.71, respectively . On the other hand, radar band or groundbased radar can be used to analyze rainfall by using Z-R relationship, which represents rainfall intensity. The equation between Z and R is created for each ground-based radar station, and this equation is representation for that station. The duration of rainfall is measured to apply volume of rainfall. Therefore, this relationship can be used for rainfall estimation when the duration of rainfall is not unknown. After these two equations are created for the target area, rainfall can be estimated from P-T relationship and/or Z-R relationship depending on the suitable information in requirement time of rainfall estimation. If it’s raining when there is not GMS satellite information, Z-R relationship has to use to estimate rainfall. On the other hand, it’s raining when GMS satellite passes the target area, so P-T relationship and/or Z-R relationship can be utilized. Finally, maps of rainfall and rainfall intensity are created, and more accurate rainfall is profit for many relative fields. Also, rainfall estimation will be developed to produce more efficiency. 4. Conclusion The characteristic of rainfall includes spatial and temporal variability. At present, rainfall is estimated by using rain gauge and remote sensing observation. Rain gauge supports temporal resolution, on the other hand it is weakness in the quality of spatial resolution. However, remote sensing provides good spatial resolution. According to the advantages of these measurement methods, more precious rainfall data should be estimated from the combination of existent rain gauge and remote sensing observation. In Thailand, there are more than 600 rainfall stations operated by TMD, RID, and RFD. Now, meteorological satellites and ground-based radar are used together with rain gauges to make more comprehensive rainfall measurements. TMD uses rain gauges to measure rainfall and uses meteorological satellite and ground-based radar to collect meteorological information. GMS satellites is used by TMD and collect cloud top temperature, cloud movement, temperature, sea surface temperature, Earth surface temperature, radiation from sun, moisture, and ozone. At the moment this satellite is not used to directly estimate rainfall in Thailand. At the same time, the ground-based radar collects the meteorology information for air transportation and weather forecasting. Since the ground-based radar can present rainfall intensity, cloud movement, and location of rainfall, it is important that ground-based radar relates to rainfall measurement. Infrared thermal and radar band have capability to estimate rainfall, while rain gauge should be utilized for calibration. Infrared thermal band is used for the estimation of rainfall by the relationship of cloud top temperature from satellite image and rainfall from rain gauge. Radar band is served to calculate rainfall from the relationship between radar-measure reflectivity and rain rate. Since there are more 600 rain gauge stations in Thailand, data from these stations are enough for the calibration of estimated rainfall. Therefore, the accurate equation is produced to estimate rainfall. Since the relationship is usually a nonlinear equation, coefficients of the equation are estimated using accurate ground data. The estimated rainfall can be calibrated by rain gauge. Rain gauges are normally used to calibrate rainfall (mm). Finally, the results of rainfall estimation using ground-based data and remote sensing observations provide an accurate rainfall map. Acknowledge We thank The Royal Golden Jubilee Ph.D. Program (RGJ) for supporting budget and Dr. Hugh Turral from IWMI (International Water Management Institute) for useful suggestions. Reference  Agata Yasushi. 2002. Thailand Routine Observation. Rainfall in Thailand. Available Source http://hydro.iis.u-tokyo.ac.jp/GAMET/GAIN-T/routine/thailand/, July 16, 2003.  Chatham County Flood Forecasting and Warning System (CCFFWS). 2003. Radar body. Radar. Available Source http://www.dhi.dk/dhiproj/Country/USA/Cha thamdemo/Radar_body.htm, May 13, 2003.  Chaubey I., C. T. Haan, J. M. Salisbury, and S. Grunwald. 1999. Quantifying Model Output Uncertainty due to Spatial Variability of Rainfall. Journal of the American Water Resources Associate, Vol. 35, No. 5, October.  Costulis, P. Kay. 2002. Background: Clouds. Cloud. Available Source http://asd_www.larc.nasa.gov/edu_act/bkg_cl ouds.html, May 2, 2003.  Costulis, P. Kay. 2002. Tropical storm formation. Tropical and rainfall. Available Source http://asdwww.larc.nasa.gov/asd_over/glossary/t.html, May 2, 2003.  FAS. 2003. Groun Based Radar (GBR) X-band Radar (XBR). Ground Based Radar. Available Source http://www.fas.org/spp/starwars/program/gbr .htm, September 4, 2003.  Kuligowski Robert J. 1996. An Overview of National Weather Service Quantitative Precipitation Estimates. Precipitation. Available Source http://126.96.36.199/pub/im/tdl96-3.pdf, April 20, 2003.  NASA. 1998. Clouds and the Energy Cycle. NASA Facts. Available Source http://www.gsfc.nasa.gov/gsfc/service/galler y/fact_sheets/earthsci/eos/clouds.pdf, October 13, 2003.  Nichols Jim, Andrew Durante, and Brian Frugis. 2003. Applications of Weather Radar (WSR-88D) in Operational Forecasting. Ground Based Radar. Available Source http://www.envsci.rutgers.edu/~group5/radar. htm, September 5, 2003.  NWS (NAtional Weather Service Forecast Office). 2003. WSR-88D The National Weather Service’s Doppler Weather Radar. Ground Based Radar. Available Source http://www.srh.noaa.gov/tlh/tlh/wsr88d.html, September 5, 2003.  Saovaphak Theerapong. 1996. Flood Induced Rainfall Assessment from Meteorological Satellite Data. Thesis of mater degree. Kasetsart University. Bangkok.  Senges Keri. 2003. What does it mean to have a Temperate Climate?. Temperate zone. Available Source http://www.wfu.edu/~sengka03/sample%20e ssay.pdf, May 5, 2003.  Sivapalan M. Siva. 2003. Hydrologic cycle and water balance. Hydrologic cycle. Available Source http://www.cwr.uwa.edu.au/cwr/teaching/ole 107/OLE_MSpart1.pdf, April 11, 2003.  Srinivasan Margaret. 2002. Ocean Surface Topography from Space-Overview. Climate zone. Available Source http://sealevel.jpl.nasa.gov/overview/climate. html, May 8, 2003.  Tatsuo Yanagita. 2003. Morphology of Cloud Phase Diagram of Cloud Pattern. Cloud. Available Source http://www.aurora.es.hokudai.ac.jp/yanagita/j ob/cloud/html/souken.html, May 2, 2003.  Thailand Meteorological Department (TMD). 2001. Meteorological satellite. Rainfall in Thailand. Available Source http://www.tmd.go.th/~satellite/section.html, July 16, 2003.  Udomchoke, V. and P. Aungsuratana. 1995. The Prediction on Flood Induced Rainfall from Meteorological Satellite Data. Proceedings The Second International Study Conference on GEWEX in Asia and GAME Pattaya, Thailand. Japan National Committee for GAME and National Research Council of Thailand. 598p.  Van der Meulen Jitze P. 2002. Review and Refine Practices for Reporting the Instantaneous Precipitation Intensity, Total Cloud Amount and Cloud Height. Cold cloud temperature. Available Source http://www.wmo.ch/web/www/OSY/ETAWS/Dos7/pdf, May 6, 2003.  Washington Richard and Todd Martin. 1999. Tropical-Temporal Links in Southern African and Southwest Indian Ocean Satellite-Derived Daily Rainfall. International Journal of Climatology. Vol. 19, pp. 1601-1616.