\ PERGAMON Renewable Energy 07 "0888# 064Ð078 www[elsevier[com:locate:renene Potential for wind generation on the Guyana coastlands Shashi Persauda\\ Damian Flynnb\ Brendan Foxb a b Department of Electrical En`ineerin`\ University of Guyana\ Geor`etown\ Guyana School of Electrical and Electronic En`ineerin`\ The Queen|s University of Belfast\ Ashby Buildin`\ Stranmillis Road\ Belfast BT8 4AH Received 2 September 0887^ accepted 11 October 0887 Abstract Guyana|s dependence upon imported petroleum fuels can only be o}set by the sustained exploitation of its indigenous resources[ With its populated coastlands exposed to the northeast trade winds and a history of small!scale wind energy utilisation wind is one such potential energy source[ In this study\ the coastal wind regime is analysed and historical data from a coastal weather station are used to estimate the potential for wind generation[ It is found that a hybrid Weibull probability density function best describes the annual wind speed frequency distribution at the reference height of 09[56 m[ With an annual mean wind speed of 4[7 m:s\ an energy pattern factor of 0[30\ and an annual average power density of 048 W:m1\ this distribution represents a class!2 wind resource\ suitable for most wind turbine applications[ Site analysis and observed trends in coastal wind availability suggest the strong likelihood of a greater wind resource in more open locations[ In view of its apparent potential for wind farm operation\ a comprehensive\ wind resource assessment programme is recommended for the Guyana coastlands[ Þ 0888 Elsevier Science Ltd[ All rights reserved[ 0[ Introduction Guyana is heavily dependent upon imported fossil fuels to meet its energy needs[ Petroleum products currently account for nearly half of the total energy needs and utilise 19) of the income derived from the export of goods and services ð0Ł[ The single largest consumer of petroleum products is the electricity sector with the lone utility Corresponding author 9859!0370:88:, ! see front matter Þ 0888 Elsevier Science Ltd[ All rights reserved PII] S 9 8 5 9 ! 0 3 7 0 " 8 7 # 9 9 6 8 2 ! 8 065 S[ Persaud et al[ : Renewable Ener`y 07 "0888# 064Ð078 accounting for approximately a third of total fuel imports[ With the exception of biomass fuels used in the sugar industry\ and on a small scale in sections of the rice and timber industries\ diesel generation is the norm[ This reliance upon imported fuels makes Guyana|s fragile economy very vulnerable to price ~uctuations in the international petroleum market[ To cushion the e}ects of inevitable price increases in the face of a diminishing resource\ greater utilisation of its indigenous resources must be e}ected[ A potential resource\ that has received little attention\ is wind energy[ Situated on the north eastern edge of the South American mainland\ the Guyana coastlands are exposed to the northeast trade winds\ known for their steady wind potential[ With a history of small scale wind energy utilisation and established electricity networks\ the coastal potential for wind generation appears promising[ Historical records suggest a long!term mean annual wind speed of 4[7 m:s at 09 m ð1Ł[ These winds are moderate by international standards but many wind farms have been designed to operate under similar conditions[ Wind predictability and its correlation with load demand may also allow for high penetration and make wind energy an economic supplement to expens! ive diesel generation[ However\ _rm decisions regarding the exploitation of the coastal wind resource must be based upon accurate knowledge of the region|s wind regime\ which can only be gained from analyses of detailed long!term wind speed records[ The available data have\ to date\ limited the e}ectiveness of wind assessment studies[ All previous studies have restricted themselves to qualitative analyses of meteorological\ topographical and historical information\ citing an insu.cient dat! abase and emphasising the need for additional data[ Whilst recognising the apparent coastal wind potential and providing valuable general information on wind patterns\ availability and utilisation\ they have been unable to quantify the local wind potential and remain generally inconclusive ð2Ð4Ł[ In the absence of new data\ quanti_cation of the coastal wind!potential remains elusive\ which is unfortunate\ given the devel! opments in wind energy technology and utilisation[ This paper attempts to classify the coastal wind resource by correlating the results of qualitative studies with statistical analysis of available records[ The coastal wind regime is qualitatively analysed and favourable zones identi_ed[ Data from a par! ticular site are then used to quantify the wind potential in these areas[ It is assumed that the length of the observational period is su.cient to accurately identify long! term means and that\ despite its age\ the data are still relevant and representative of shoreline coastal sites today[ 1[ The Guyana wind resource Guyana is situated on the north eastern edge of the South American mainland between the approximate latitudes 0Ð8> N of the equator "see Fig[ 0#[ These latitudes lie within the ranges of two predominant weather zones\ the North East Trade Winds and the Equatorial Doldrums or Inter Tropical Convergence Zone "ITCZ#[ The northeast trade winds are steady winds with good all!year energy potential\ whilst the ITCZ is an equatorial belt of calm winds resulting from the convergence of the S[ Persaud et al[ : Renewable Ener`y 07 "0888# 064Ð078 066 Fig[ 0[ Windy regions of Guyana "not to scale#] "0# Northwest coast^ "1# Essequibo coast^ "2# West Demerara coast^ "3# East Demerara:Berbice coast^ "4# Corentyne coast^ "5# Rupununi[ northern and southern trade winds[ The annual northÐsouth movement of the ICTZ is held responsible for the general variation in weather and wind patterns experienced over Guyana[ Distribution of the wind resource is further characterised by topogra! phy[ The hilly and mountainous regions that constitute the bulk of Guyana|s land mass exhibit little wind potential\ whilst the ~at lands of the Rupununi savannahs and the coastal plains are relatively windy[ The Rupununi is a large\ undeveloped and sparsely populated region located in the country|s southwest[ With an approximate population density of two persons per square kilometre this geographically remote region has limited access to conventional resources\ such as roads\ fuel\ electricity\ water\ etc[ Only its capital\ Lethem\ has any provision for electricity*a few hours daily[ Local knowledge suggests that the average wind potential of the Rupununi is greater than that of the coastlands\ but veri_cation is impossible at present due to extremely limited data[ Wind power for water pumping applications in this region is well established\ however\ with many villages receiving potable water from multi!vaned wind pumps[ A scarcity of investment capital\ high transportation costs and limited access to spares and supplies are some of the main factors that have contributed to the limited utilisation of the wind resource for electricity generation[ 067 S[ Persaud et al[ : Renewable Ener`y 07 "0888# 064Ð078 The coastal plain is a narrow strip of arable land\ some three hundred miles long and three to _ve miles in depth[ It is segmented into _ve distinct regions* the Northwest\ Essequibo\ West Demerara\ East Demerara:Berbice and Corentyne coastlands*by four large rivers "Fig[ 0#[ With the exception of the northwest region\ the coastlands are well developed with established infrastructure*electricity\ roads\ water\ etc[\ and are well populated\ with 74) of the country|s population[ Most of these regions have a history of wind energy utilisation for both water pumping and small!scale electricity generation\ but such applications have largely been displaced by established regional electricity and water distribution networks[ The northeast trade winds predominate across the coastlands\ their in~uence being greatest at the shoreline and reducing signi_cantly as they progress inland[ The e}ects of the ITCZ\ greatest in the northwest region\ are generally only experienced for a few months of the year ð3Ł[ The resulting seasonal pattern in wind availability is characterised by high mean wind speeds during the northern winter and low means during the summer[ Superimposed upon the seasonal periodicity are diurnal cycles\ characterised by early morning lows and afternoon highs[ These cycles\ attributed to changes in momentum transfer due to solar radiation\ are most distinctive during the summer when mean wind speeds are low[ Figures 1 and 2 illustrate the seasonal and diurnal phenomena[ Knowledge of the coastal wind patterns suggests that the most promising wind sites occur close to the ocean|s edge\ with the populated central and eastern coastal regions being most favourable[ Fig[ 1[ Monthly mean wind speeds[ Old Ri~e Range\ Georgetown\ 0857Ð0863[ S[ Persaud et al[ : Renewable Ener`y 07 "0888# 064Ð078 068 Fig[ 2[ Annual and seasonal diurnal cycles[ Old Ri~e Range\ Georgetown\ 0857Ð0863[ 1[0[ The wind database The Guyana Hydrometeorological Service has on record\ wind speed measurements from nine weather stations scattered across the country[ With the exception of two coastal stations\ however\ wind speeds were only sampled twice daily and are of limited utility value[ Hourly measurements were made at the Old Ri~e Range and the Botanic Gardens stations^ both located in the capital city\ Georgetown[ The Old Ri~e Range is situated on the Atlantic shoreline\ whilst the Botanic Gardens is located some two miles inland\ in central Georgetown[ Because of its sheltered location\ wind speed recordings from the latter site are not considered representative of the coastal wind resource and\ in the absence of a known correction factor\ cannot be used to estimate the coastal wind potential[ Data from the Old Ri~e Range station\ located very close to\ and in open view of\ the Atlantic Ocean are\ however\ considered indicative of the wind regimes along the Demerara and Berbice coastlands[ These were recorded during the period 0857Ð0863\ at an anemometer height of 24 ft "09[56 m#\ using a Dines tube anemometer and a continuous data recorder[ Hourly mean wind speeds were found by {striking| the average of the previous hour|s continuous recordings[ From the strip chart records\ monthly averages of hourly mean wind speeds "Fig[ 1#\ monthly and annual means "Figs 2 and 3\ respectively# were computed[ Wind speed and wind direction frequency distribution statistics were only compiled for the period 0860Ð0862[ These\ published in the Annual Climatological Summaries S[ Persaud et al[ : Renewable Ener`y 07 "0888# 064Ð078 079 Fig[ 3[ Variation in annual mean wind speeds[ of the Guyana Hydrometeorological Service ð3Ł\ have been used to compute the seasonal and annual distributions shown in Table 0[ Approximately 7[8) of the hourly data for this period were lost due to equipment and other failures[ Seasonal and annual mean wind speeds and standard deviations\ computed for the longer period of 0857Ð0863\ are shown in Table 1[ Wind speeds were originally recorded in knots[ Table 0 Wind speed frequencies[ Old Ri~e Range\ Georgetown\ 0860Ð0862 Wind speed bins "knots# "m:s# JanÐMar ")# AprÐJun ")# JulÐSep ")# OctÐDec ")# Year ")# 9Ð1 1Ð2 2Ð5 5Ð09 09Ð05 05Ð10 10Ð16 16Ð22 9[99Ð0[92 0[92Ð0[43 0[43Ð2[97 2[97Ð4[97 4[97Ð7[11 7[11Ð09[68 09[68Ð02[77 02[77Ð05[81 1[91 9[96 9[88 8[30 51[43 12[96 0[77 9[91 5[10 9[00 1[76 05[19 42[95 19[32 0[00 9[99 02[63 9[06 7[32 16[75 34[46 2[73 9[24 9[92 7[85 9[96 2[84 14[27 44[28 4[70 9[33 9[99 6[62 9[00 3[95 08[60 43[03 02[18 9[84 9[90 S[ Persaud et al[ : Renewable Ener`y 07 "0888# 064Ð078 070 Table 1 Seasonal and annual mean wind speeds[ Old Ri~e Range\ Georgetown\ 0857Ð0863 Mean wind speed "m:s# Standard deviation "m:s# JanÐMar AprÐJun JulÐSep OctÐDec Year 6[03 9[28 4[88 0[92 3[50 9[57 4[31 9[22 4[68 9[31 2[ Method of data analysis The three!year seasonal and annual wind speed frequency distribution statistics presented in Table 0 and the seven!year seasonal and annual mean wind speeds presented in Table 1 are used to estimate the expected long!term distribution of wind speeds at the Old Ri~e Range site and to quantify its energy potential[ A probability model is selected that is representative of the recorded frequency distribution data and its seasonal and annual distribution parameters and mean wind speeds estimated using the procedure outlined in section 2[0[ Relationships are then identi_ed\ between the estimated parameters and mean wind speeds\ and used to predict the distribution parameters corresponding to the long!term mean wind speeds[ The energy pattern factors and power densities associated with these distributions are then computed as shown in section 2[1[ 2[0[ Wind speed frequency distribution The probability models that have evolved as standard in wind resource assessment studies are the Weibull and Raleigh distribution ð5Ł[ The general form of the Weibull distribution is fw"v# "k:v#"v:c#k−0 exp "−"v:c#k# "0# where fw"v# is the probability of observing wind speed v\ k is a dimensionless shape factor\ and c\ referenced in the units of wind speed\ is the scale factor[ The Raleigh distribution is a special case of the Weibull distribution\ with k 1[ The cumulative Weibull distribution is given by Fw"v# exp "−"v:c#k# "1# A number of methods are available for estimating the factors\ c and k\ depending upon the available data[ Where sampled wind speed frequency distribution statistics are available\ the {least!squares| method is preferred ð5Ł[ Here it is assumed that the recorded cumulative frequencies are noise!corrupted samples of the cumulative Weibull function\ such that F"v# Fw"v#¦r"v# "2# where F"v# is the recorded cumulative frequency corresponding to the wind speed v and r"v# is the associated noise residual[ Determination of the parameters of the 071 S[ Persaud et al[ : Renewable Ener`y 07 "0888# 064Ð078 Weibull distribution requires a good _t of eqn "1# to the recorded discrete cumulative frequency distribution[ Equation "1# is linearised by taking the natural logarithm of both sides twice\ to give ln "− ln "F"v### k ln "v#−k" ln "c## "3# and k and k" ln "c## are subsequently determined as the least squares solution to the resulting over!determined system of eqn "4#[ A common approach is to _t a least squares straight line to a scatter plot of ln "−ln "F"v### vs ln "v# and determine its slope*k and intercept with the ln "v# axis−k ln "c#[ K ln "− ln "F"v0### L K ln "v0#−0 L H H H H k H ln "− ln "F"v1### H H ln "v1#−0 H H H H H * * k" ln "c## H H H H kln "− ln "F"vn−0###l kln "vn−0#−0l $ % "4# With c and k determined\ the likelihood of wind speed occurrences within a given interval "v0 ³ v ³ v1# can be estimated as Fw"v0 ³ v ³ v1# Fw"v0#−Fw"v1# "5# and the probabilities of wind speed occurrences determined from eqn "0#[ The mean vmw and standard deviation sw of the Weibull distribution can then be computed from ð6Ł vmw cG"0¦0:k# "6# sw z"c1ðG"0¦1:k#−G1"0¦0:k#Ł# "7# and where G" # is the Gamma function[ The main limitation of the Weibull model is that it does not accurately represent the probabilities of observing zero and very low wind speeds\ given that fw"9# 9[ This limitation can be overcome\ however\ through the use of a hybrid model of the form Fh"v# F"9#¦ð0−F"9#ŁFw"v# "8# where F"9# is the cumulative probability of observing very low wind speeds[ The distribution of very low wind speeds can further be modelled separately[ Reference ð6Ł suggests the use of the Dirac Delta function[ For the purposes of estimating wind potential\ however\ this is usually unnecessary\ as the energies avail! able at low wind speeds are negligible and outside the operating range of wind turbines[ From eqn "8#\ the cumulative frequencies needed for the determination of the Weibull parameters can be calculated as] S[ Persaud et al[ : Renewable Ener`y 07 "0888# 064Ð078 Fw"v# ðFh"v#−F"9#Ł:ð0−F"9#Ł 072 "09# and the mean of the hybrid distribution as vm F"9# vm9¦ð0−F"9#Ł vmw "00# where vm9 and vmw are the means of the low wind speed and Weibull distributions respectively[ 2[1[ Ener`y availability A key indicator to the size of an available wind resource is its annual average power density[ Computation of this factor at a reference height of 09 m or 49 m allows for a general classi_cation of the available resource and the subsequent estimation of its potential for wind energy applications ð3Ł[ The wind power density\ PD\ associated with a wind of velocity v m:s and density r kg:m2 is ð5Ł PD 9[4rv2 W:m1[ "01# The annual average power density may then be estimated as PD"avg# 9[4rÐv2f "v# dv 9[4rKev2m W:m1 "02# where Ke is the annual energy pattern factor that accounts for all variations in wind strength over the year[ At sea level and assuming an average temperature of 16>C\ the density of air is 0[066 kg:m2 and eqn "02# becomes PD"avg# 9[4774 Kev2m W:m1 "03# The energy pattern factor is computed as Ke energy available in the wind per year energy available at mean wind speed g 9 v2f "v# dv v2m "04# where f "v# is the known wind speed probability density function at the site[ Sub! stituting fw"v#\ from eqn "0#\ for f "v# into eqn "04# gives Ke "k:ck−0# v2m g vk¦0 exp "−"v:c#k# dv[ "05# 9 Recognising the de_nite integral ð8Ł g vx−0 exp "−uvp# dv "0:=p=#u−"x:p#G"x:p# "06# 9 allows for the simpli_cation Ke c2G"0¦1:k# v2m "07# S[ Persaud et al[ : Renewable Ener`y 07 "0888# 064Ð078 073 In the case of the hybrid Weibull distribution\ the energy pattern factor can be approximated as Ke "0−F9#c2G"0−1:k# v2m "08# From eqn "03# the annual energy availability is determined as E 7659PD"avg# Wh:m1:yr "19# 3[ Results The adoption of irregular wind speed bins in the compilation of wind speed fre! quency distributions by the Guyana Hydrometeorological Service\ prevents an accu! rate representation of the data by way of frequency histograms[ To overcome this problem\ approximate distributions were created\ by regrouping frequencies into uniform 0[92 m:s "1!knot# intervals[ The frequency histogram corresponding to the annual distribution presented in Table 0 is shown in Fig[ 4[ Examination of Fig[ 4 and the seasonal histograms reveals an approximately Weibull distribution except for the signi_cant occurrences of wind speeds in the 9Ð0[92 m:s bin[ A hybrid Weibull probability density function was selected to represent the continuous wind speed Fig[ 4[ Hybrid and approximate annual frequency distributions "0860Ð0862#[ S[ Persaud et al[ : Renewable Ener`y 07 "0888# 064Ð078 074 frequency distributions recorded in Table 0\ as it accounted for both the high and near zero frequencies observed in the 9Ð0[92 m:s and 0[92Ð0[43 bins\ respectively[ In the determination of the Weibull factors in the hybrid function\ frequencies in the 9Ð0[92 m:s bin were removed from the recorded distributions\ the remaining data scaled to represent 099) of useful data\ and the factors estimated using the procedure outlined in section 2[0[ For simplicity\ wind speed occurrences within the 9Ð0[92 m:s interval have not been modelled separately[ It was found that the best _t of the cumulative hybrid distribution to the recorded data occurred when {outliers| were ignored in the Weibull parameter estimation process[ Figure 5 illustrates the goodness! of!_t for the case of the annual distribution[ The results for the seasonal and annual distributions are presented in Table 2[ Fig[ 5[ Hybrid and recorded annual cumulative frequency distributions "0860Ð0862#[ Table 2 Three!year "0860Ð0862# seasonal and annual distribution parameters at 09[56 m F9")# c"m:s# k vm"m:s# JanÐMar AprÐJun JulÐSep OctÐDec Year 1[91 6[85 3[68 6[05 5[10 6[37 2[82 5[23 02[63 5[05 2[12 3[72 7[85 5[69 2[77 4[45 6[62 6[97 2[69 4[83 S[ Persaud et al[ : Renewable Ener`y 07 "0888# 064Ð078 075 Fig[ 6[ Distribution parameters vs mean wind speed[ Examination of the results\ presented in Table 2\ indicates that the distribution parameters vary with mean wind speed[ The scatter plots of Fig[ 6 further illustrate the nature of these relationships\ which have been identi_ed as c"vm# 9[687vm¦1[147 "10# k"vm# 9[204vm¦9[143 "11# F9"vm# exp ð−"vm:2[15#0[53Ł "12# F9 was non!linearly modelled as it must satisfy the conditions F9"9# 0 and F9"# 9[ From the relationships identi_ed and the long!term mean wind speeds presented in Table 0\ seasonal and annual long!term distribution parameters have been estimated[ These are shown in Table 3 along with the corresponding values of energy pattern factor and average power density[ The seasonal and annual distributions of wind speeds are shown in Fig[ 7[ 4[ Discussion The foregoing analysis has revealed that a hybrid Weibull distribution best charac! terises the distribution of wind speeds at 09[56 m[ With an expected annual mean wind speed of 4[68 m:s at 09[56 m and a corresponding annual average power S[ Persaud et al[ : Renewable Ener`y 07 "0888# 064Ð078 076 Table 3 Estimated seasonal and annual longÐterm distribution parameters at 09[56 m vm "m:s# F9 ")# c "m:s# k Ke PD"avg# "W:m1# Wind power class JanÐMar AprÐJun JulÐSep OctÐDec Year 6[03 1[56 6[85 3[51 0[11 146[6 4 4[88 5[50 6[93 2[81 0[26 061[0 2 3[50 06[04 4[83 2[96 0[53 82[1 0 4[31 09[95 5[47 2[46 0[36 024[7 1 4[68 6[60 5[77 2[79 0[30 047[5 2 Fig[ 7[ Estimated seasonal and annual frequency polygons "0857Ð0863#[ density of 048 W:m1\ these winds correspond to a class!2 wind resource at 09 m "049 ³ PD"W:m1# ³ 199# when measured on the Bagattelle wind power scale ð5Ł[ This available resource exhibits signi_cant seasonal variation and ranges from a class!4 "149 ³ PD"W:m1# ³ 299# resource during winter to a class!0 "9 ³ PD"W:m1# ³ 099# resource in the summer[ As a consequence\ energy availability during the period\ JulyÐSeptember\ will be minimal "approx[ 03) of annual available energy# providing a convenient period for wind turbine maintenance[ A class!2 resource is generally considered suitable for most wind turbine appli! S[ Persaud et al[ : Renewable Ener`y 07 "0888# 064Ð078 077 cations but for economic operations of large wind turbines a greater resource at its operating hub height is required ð5Ł[ In the absence of measured data\ the available resource at an extended height can be estimated from the Hellman|s wind shear formula v v9"h:h9#a "13# where v9 is the known velocity at the reference height h9\ v is the desired value at the new height h and a is Hellman|s exponent\ usually 9[032 for ~at coastal regions[ On the Guyana coastlands very little is known about the distribution of wind speeds with regard to height and extrapolation of mean wind speeds cannot be performed with con_dence[ Two factors suggest\ however\ a larger value for the Hellman|s exponent at the Old Ri~e Range site than the general coastal _gure of 9[032[ Firstly\ the Guyana coastlands are approximately 0[4 m below sea level and separated from the Atlantic Ocean by sea defence structures that project another metre[ With the Old Ri~e Range station situated only a very short distance from the ocean\ it is likely to have been partially shielded from the predominant northeasterly trade winds\ blowing inland from the Atlantic[ Secondly\ this site is located on the outskirts of the city of Georgetown\ whose pro_le must serve to restrict wind ~ow at low altitudes[ The foregoing discussion suggests that higher wind speeds are likely at shoreline locations backed by ~at open spaces\ and that the wind resource may exceed a minimum class!2 rating at the hub height of a large wind turbine[ A similar phenom! enon has been observed in Ref[ ð7Ł[ Such an eventuality should make wind farm generation an economically viable proposition\ especially when compared to the expensive diesel generation that currently prevails[ The fact that diurnal and seasonal patterns in wind availability correlate strongly with load demands and are very predictable\ will also serve to enhance wind penetration levels and improve economy[ 5[ Conclusions The coastal potential for wind generation has been qualitatively assessed and quanti_ed using the available historical records[ This investigation has revealed that the most promising wind sites are located along the Atlantic shoreline in the regions of the Essequibo\ Demerara:Berbice and Corentyne coasts[ At such locations\ a minimum class!2 wind resource can be expected at 09 m[ This establishes the potential for small!scale wind generation and water pumping applications\ but such utilisation of the wind resource is likely to remain minimal\ due to the presence of electricity and water networks[ Wind farm generation\ on the other hand\ can contribute signi_cantly to national energy needs by feeding directly into the existing electricity networks[ Such a scenario would only be possible\ however\ if the coastal wind regimes were strong enough to support and make wind farm operation viable[ Extrapolation of the available data suggests that this is the case\ but such analysis is questionable and the results cannot be used with con_dence[ Given the likely bene_ts that could result from economic S[ Persaud et al[ : Renewable Ener`y 07 "0888# 064Ð078 078 wind farm operation it is important that the potential for such operation be accurately established[ A comprehensive wind resource assessment programme is therefore rec! ommended for the Guyana coastlands\ the results of which can be made available to potential interest groups[ Acknowledgements The authors wish to thank the Guyana Hydrometeorological Service for infor! mation provided[ S[ Persaud thanks the Commonwealth Scholarship Commission for funding[ References ð0Ł Government of Guyana[ National development strategy[ Part V\ 0886[ ð1Ł Guyana Hydrometeorological Service[ Annual climatological summary Ministry of Works and Com! munication[ Guyana\ 0860Ð0862[ ð2Ł Raghunandan S[ Solar and wind energy potential in Guyana[ A paper presented at the training workshop on meteorological data for solar and wind energy applications[ Barbados\ December 0873[ ð3Ł Lamming SD[ Report on a fact _nding mission to Guyana[ Consultancy Services Wind Energy Developing Countries\ October 0878[ ð4Ł Liu N[ Requirements for the development of wind energy in Guyana[ Guyana National Resources Agency\ 0877[ ð5Ł Frost W\ Aspliden C[ Characteristics of the wind[ In] Wind turbine technology[ ASME Press\ 0883[ p[ 260Ð289[ ð6Ł Eastern Caribbean wind resource assessment project[ Regional summary[ Caribbean Development Bank\ vol[ 0\ 0883[ ð7Ł Hossain A[ A preliminary account of wind speed data in the coastal regions of Bangladesh[ Wind Energy Conversion\ 0885[ p[ 284Ð8[ ð8Ł Gradshteyn IS\ Ryzhik IM[ Table of integrals\ series and products[ 4th ed[ Academic Press\ 0883[ p[ 275[