Quantifying water and energy

Quantifying water and energy budgets and the impacts of climatic and human factors in the Haihe River Basin,China:2.Trends and implications to waterresourcesYing Guo,Yanjun Shen ⇑Key Laboratory of Agricultural Water Resources,Hebei Key Laboratory of Agricultural Water-Saving,Center for Agricultural Resources Research,Institute of Genetics and Developmental Biology,Chinese Academy of Sciences,Shijiazhuang 050021,Chinaa r t i c l e i n f o Article history:Received 31December 2014Received in revised form 3March 2015Accepted 28April 2015Available online 5May 2015This manuscript was handled by Corrado Corradini,Editor-n-Chief,with theassistance of Magdeline Laba,Associate EditorKeywords:Evapotranspiration Climate impacts Human impacts Haihe River BasinSpatial and temporal change Water and energy fluxs u m m a r yIn this paper,spatial and temporal changes in water and energy fluxes were investigated over the past 28years along with relevant factors in the Haihe River Basin in North China,which is currently facing an increasingly severe water shortage.Water and energy fluxes in different topographical regions have been affected by both climate change and human activities.In the plain regions,actual evapotranspira-tion (ET a )has generally decreased during the study period,which can be attributed to decreased net radi-ation,precipitation and rapid urbanization.The water deficit has continued in the plains regions due to extensive pumping for irrigation to meet the deficit of crop water.However,irrigation has led to signif-icant groundwater depletion,which poses a significant challenge to the sustainability of water resources.In the mountain regions,decreases in sensible heat flux and increases in ET a have caused a decrease in the Bowen ratio.These changes in water and energy fluxes have brought a positive ecological effect,but the increase in ET a in the mountain regions has resulted in a substantial reduction of water yield and conse-quently a negative effect on water resources in this basin.These changes in water and energy fluxes across the Haihe River Basin have made the water issue more complex and require integrated river basin management.Ó2015Elsevier B.V.All rights reserved.1.IntroductionClimate changes and human activities have substantial impacts on land surface processes,such as the hydrological cycle,water and energy balance,and eco-environment at various spatial and tem-poral scales (Piao et al.,2010).Regional and global climate change and human activities are interrelated.Human activities have a pro-found influence on the hydrologic cycle and the surface energy budgets and consequently regional climate.Climate changes com-bined with human activities result in massive changes to eco-hydrological patterns,which lead to changes in hydrological processes at basin scales globally,causing water resource and eco-logical environment problems in some regions (Li et al.,2007;McVicar et al.,2007;Zhang et al.,2001).In semi-arid,semi-humid and drought-prone areas,ecosystems are more vul-nerable to climate change and human disturbance (Liu and Xia,2004a;Xia and Zhang,2008).For instance,irrigation from ground-water pumping may result in changes in the regional hydrological cycle,including increasing ET a rates and consequently a regional climate with a higher humidity,more frequent extreme precipita-tion events and land surface cooling (Kueppers et al.,2007;Lo and Famiglietti,2013;Long et al.,2012;Ozdogan and Salvucci,2004).A regional climate change may then lead to further changes in the hydrological cycle,including the reduction or increase of ET rates and therefore changes in runoff (Liu and Yang,2010;Lo and Famiglietti,2013).Based on a quantification study of hydrological and energy ele-ments,water and energy changes in time and space can be detected and analyzed to understand the hydrological trends and environmental changes (Kumar and Merwade,2011;Leblanc et al.,2012;Moran-Tejeda et al.,2011;Long et al.,2014).Changes in water and energy fluxes are primarily impacted by cli-mate change and human activities (Liu and Xia,2004a ).Changes in meteorological variables may have a combined effect on water and energy fluxes,e.g.,changes in air temperature,wind speed,and sunshine duration,which have affected the evaporative capability of the atmosphere and ET a in the Yellow River Basin of China over the past 50years (Liu and Yang,2010;Wang et al.,2012).Increased air temperature and decreased relative humidity can enhance the/10.1016/j.jhydrol.2015.04.0710022-1694/Ó2015Elsevier B.V.All rights reserved.⇑Corresponding author.Tel./fax:+8631185872248.E-mail addresses:guoy@ (Y.Guo),yjshen@ (Y.Shen).evaporative capability and decrease the actual vapor,which,in turn,caused an increase in ET a in arid and semi-arid regions of Iran from1951through2005(Eslamian et al.,2011).Sunshine duration directly affects the solar radiation;a decrease in sunshine duration caused a decrease in solar radiation in the HRB from1957 to2008(Liu et al.,2010).Similar decreasing trends in sunshine duration were found in other places globally(Stanhill and Cohen, 2001;Wild,2009;Wild et al.,2005).For example,in the United States,solar radiation decreased from1961to1990due to decreased sunshine duration resulting from an increase in cloud cover(Liepert,2002).The influences of human activities on water and energy budgets are primarily manifested by land use/cover change,agricultural irrigation,and urbanization(Mao and Cherkauer,2009; Nakayama,2011;Ozdogan and Salvucci,2004;Randhir,2003). For instance,in an agricultural catchment on the Loess Plateau of China,conversion from shrub land and sparse woodland to med-ium and high grassland resulted in decreased runoff,decreased soil water contents,and increased ET a during1981–2000,(Li et al., 2009).In the Great Lakes region of the USA,both conversion from deciduous forests to wooded grasslands and row crop agriculture, and conversion from a majority of evergreen plants to a majority of deciduous forests has resulted in a decrease in ET a and an increase in total runoff in the central and northern areas,respec-tively.Conversion from prairie grasslands to row agriculture crop has also resulted in an increase in ET a and a decrease in total runoff in the southern and western parts(Mao and Cherkauer,2009).Since the1950s,irrigated agriculture has expanded by174% globally,accounting for90%of global freshwater consumption (Scanlon et al.,2007).Surface water irrigation has reduced stream flows,raised water tables and consequently resulted in waterlog-ging of many regions and countries(e.g.,China,India,and the United States).However,marked increases in groundwater-fed irrigation in the last few decades in these regions have lowered water tables(1m/yr)and reduced streamflow(Scanlon et al., 2007).Irrigation also leads to increased ET a.Simulations over a 20-year period in the Colorado and Mekong River Basins indicate irrigation water requirements of10km3/yr and13.4km3/yr, respectively,corresponding to streamflow decreases of37%and 2.3%;an increase in latent heatflux is also accompanied (Haddeland et al.,2006).Urbanization is manifested primarily by construction of imper-vious built-up areas(e.g.,buildings,road networks,and drinking water,rainwater and drainage networks).These changes have impacts on the water cycle throughout a region due to increases in surface runoff,decreases in groundwater recharge,and modifi-cations of natural water pathways.For instance,discharge was affected significantly by rapid development of cities and industries and increase in farmland irrigation,resulting in groundwater degradation(Braud et al.,2013).In recent decades,with the development of agriculture,indus-try and urbanization,water resources safety and ecological envi-ronment problems in the Haihe River Basin of North China have become serious(Liu and Xia,2004a).In order to manage water resources in a sustainable manner,it is imperative to study the water and energy budget changes associated with the water prob-lems to understand the causes of the existing issues of water resources and to try to increase water safety.Therefore,a quantita-tive study of the spatial and temporal changes in water and heat fluxes and an analysis of their causes in the Haihe River Basin are essential to understand the mechanisms of these changes in the hydrological cycle.It will also be helpful for water resources man-agement and safeguarding food security in such regions.The objectives of this study were therefore to(1)examine the spatial and temporal trends of climate change and land use in the HRB over the past28years,(2)identify spatial and temporal trends in water/energy budgets and water/heat environment,(3) investigate the effects of climate change and human activities on land surface water and energy processes,and(4)provide implica-tions of such changes and effects for the hydrological cycle and water resources.2.Materials and methods2.1.Study areaThe Haihe River Basin(HRB)is surrounded by the Bohai Sea to the east,Taihang Mountain to the west,the Mongolia Plateau to the north and the lower reaches of the Yellow River to the south. The topography descends gradually from the plateau and moun-tainous regions in the northern and western areas to the plain region in the east.The HRB is divided into5regions(i.e.,the coastal plain,the low plain,the piedmont plain,the hilly mountains and the high mountains),as shown in the topography of Fig.1.In the plateau and mountainous regions,lands are primarily covered by forest and shrubs,and partly by grasslands and meadow.In the plain regions,lands are primarily covered by crops(Gao et al., 2012;Zhu et al.,2010).Most irrigated croplands are distributed in the piedmont and low plain regions,and the rain-fedfields are primarily distributed in the coastal plain regions.Over the past few decades,water shortage has become a serious problem in the HRB.Frequent droughts and sharply growing water shortages in this basin hindered its economic development and resulted in severe environmental problems(Liu and Xia,2004b), which were partly due to rapid social and economic development; the problem was further aggravated by climate change(Gao et al., 2012).The annual precipitation in the Haihe River Basin has also been decreasing(Wang et al.,2011).Runoff in the basin has also exhibited a steadily declining trend,which can be attributed to increased human activity and possibly to climate change as well (Liu and Xia,2004b;Ren et al.,2002;Zhang et al.,2011).A signif-icant decline in runoff was found infive of eight studied sub-basins and abrupt changes in runoff occurred in1978–1985for most studied sub-basins in the HRB(Yang and Tian,2009).Air tempera-ture has also increased across the majority of the HRB(Tang et al., 2011).Rapid agricultural developments such as desalination and intensifying irrigation in coastal croplands have resulted in over-exploitation of groundwater resources.Rapid urban expan-sion and other projects also resulted in land use changes(Liu and Xia,2004b).2.2.DataThe net radiation(R n),sensible heatflux(H),potential evapo-transpiration(ET p),actual evapotranspiration(ET a),and soil mois-ture in the HRB during1981–2008have been simulated using the developed model described and validated in the companion paper to this study.Annual water and energyfluxes were calculated using daily simulations at a daily time step.The annual climate data were calculated using the spatially interpolated daily temper-ature(T),precipitation(P),relative humidity(rh),wind speed(ws), and sunshine duration(sd),which were used for analysis of the influential factors due to climate nd use changes reflect human activities.In this study,land use data based on TM (Thematic Mapper)images were used.Ten land use types were combined intofive groups(i.e.,cropland,forest,grassland,imper-vious,water body and unused land)at30m resolution.Area per-centages within each8kmÂ8km grid of these land use types were calculated for the analysis on the influential factors due to human activities on water and energyfluxes.The annual aridity index and Bowen ratio and their trends in the HRB were calculated252Y.Guo,Y.Shen/Journal of Hydrology527(2015)251–261for the analysis of the water and heat environment changes during 1981–2008.The aridity index is defined as the ratio of annual potential evapotranspiration to precipitation (Koster and Suarez,1999).A higher aridity index represents drier air because the evap-orative demand cannot be met by precipitation.The Bowen ratio is defined as the ratio of sensible heat flux to latent heat flux,which reflects the energy state of the ecosystem.2.3.Methods2.3.1.Trend analysisIn this study,two trend analysis methods are employed to investigate the changes of meteorological variables and water/energy fluxes at regional scale.The Mann–Kendall nonparametric test (Sprent,1976,2009)is a common method to search for a trend in a time series without specifying the distribution of the data and is less sensitive to outliers.For a time series X ={x 1,x 2,...,x n },the standard normal variate Z (Eslamian et al.,2011;Khaliq et al.,2009)is given by:Z ¼s ffiffiffiffiffiffiffir 2r Ãp ð1Þs ¼4pn ðn À1ÞÀ1ð2Þr 2s ¼2ð2n þ5Þ9n ðn À1Þð3Þwhere p is the number of pairs of data (x i ,x j ,j >i ,x j >x i )and n is the number of data in the time series.The significance of trends can be tested by comparing the standardized variable Z in Eq.(1)with the standard normal variate at the desired significance level a .If the absolute value of Z is greater than the theoretical value Z a /2,the time series have a trend at the significant level of a .A positive value of Z indicates an increasing trend in the time series,while a negative value describes a decreasing trend.A simple linear regression method is also highly recommended for general use by the World Meteorological Organization;this method is typically used to determine to the trends in evapotran-spiration,radiation and other meteorological elements and to test their significance (Mitchell et al.,1966).This method determines whether there is a linear trend by examining the relationship between time (x )and the variable of interest (y )using least square linear regressions:y ¼a þb Áx þeð4Þwhere a and b are parameters and e is the regression residual.The trend value b is defined as the regression gradient of the least-squares linear function fitting the interannual variability of water and energy flux value,which is estimated by:b ¼P ni ¼1ðx i À x Þðy i À yÞP ni ¼1ðx i À x Þ2ð5ÞParameter b also reflects the change rate of y at the time scale ofx .A parametric T -test method was employed to test the statistical significance of the trends (Santer et al.,2000).2.3.2.Coefficient of variationThe coefficient of variation (CV)is a normalized measure of dis-persion of a probability distribution or frequency distribution.It is also known as unitized risk or the variation coefficient,which is defined as the ratio of the standard deviation r to the mean l (Abdi,2010).In this study,it was used to represent the variations of water and energy fluxes over the past 28years.2.3.3.Correlation analysisTo investigate the effects on water and energy flux changes due to climate changes,three types of correlation coefficients were employed,which are the Pearson correlation coefficient,Spearman’s rank correlation coefficient (Khaliq et al.,2009)and Kendall’s tau (s )coefficient.The Pearson correlation coefficient is a measure of the linear correlation between two variables x and y .It is widely used in the sciences as a measure of the degreeoftopographic regions (coastal plain,low plain,piedmont plain,hilly mountain (150–500m)and high mountain elevation in the Haihe River Basin and its location in China.linear dependence between two variables.(Ahlgren et al.,2003) and is a value betweenÀ1and1(inclusive),where1is a com-pletely positive correlation,0is no correlation,andÀ1is a com-pletely negative correlation.The Spearman correlation coefficient r src is defined as the Pearson correlation coefficient between the ranked variables.It is described as being non-parametric,which meanings a perfect Spearman correlation results when x and y are related by any monotonic function,and can be contrasted with the Pearson correlation(Khaliq et al.,2009).Kendall’s tau(s)coef-ficient is a statistic used to measure the association between two measured quantities,which is a measure of rank correlation(i.e., the similarity of the ordering of data when ranked by a given quan-tity)(Kendall,1938;Sprent,1990).Kendall’s s coefficient is defined in Eq.(2);the coefficient s is betweenÀ1and1.If the agreement between the two rankings is perfect(i.e.,the two rankings are the same),the coefficient has a value of1;if the disagreement between the two rankings is perfect(i.e.,one ranking is the reverse of the other),the coefficient has a value ofÀ1;if x and y are inde-pendent,then the authors would expect the coefficient to be approximately zero.variations of T were very slight with CV value between0.002and 0.005(Fig.3b);this result for T is similar to that of Tang et al. (2011)from1960to2002.Annual precipitation P decreased in most parts of the HRB except in the high mountain regions,where P increased at a rate between1mm/yr and5mm/yr.However, both the decrease and increase trends are shown to be insignificant (Figs.2b and3c).The CV of P ranged from0.1to0.21in the basin and was higher in the plain regions than in the high mountain regions(Fig.3d).The pattern is generally similar to the results of Gao et al.(2007)and Tang et al.(2011).The general decreasing trend in P can be partly explained by the weakening of the Asian summer monsoon over the past decades(Wang and Zhou,2005). Sunshine duration determines solar radiation,and consequently, impacts the total net radiation.The HRB was dominated by a sig-nificant decreasing trend in annual sd during the study period (Figs.2c and3e).The negative trends for sd were less than À10h/yr in the plain regions;the positive trends for sd are only shown in a few high mountain regions at elevations above 2000m,(Fig.3e).Annual rh decreased throughout the basin;the magnitudes of the decreasing trends magnified from the southeast parts to the northwest parts in the basin(Figs.2d and3g). However,the variations of both sd and rh are not significantMann–Kendall trends(SI:significant increasing,SD:significant decreasing,NSI:nonsignificant increasing,NSD:nonsignificant decreasing;significant level=0.05) temperature,(b)precipitation,(c)sun duration,(d)relative humidity and(e)wind speed during1981–2008in the HRB.254Y.Guo,Y.Shen/Journal of Hydrology527(2015)251–261proportion is more than70%of the mountainous areas;cropland and a few cities scattered in the mountain basins make up the remaining proportion.In the plain regions in the eastern and southern parts of the HRB,cropland occupies more than70%of the area.Most urban and rural areas in the HRB are located in this region,which are primarily covered by impervious surfaces.During 1980–2008,the cropland had the largest area ratio change(from 48.5%to42.9%),followed by forest(from31.2%to31.4%)and grass-land(from10.5%to12%)throughout the basin.The urban and rural areas constituted5.7–9.2%of the entire basin,while water areas only constituted 2.6–2.8%and unused land occupied 1.5–1.7% (Fig.4).In the past28years,the land use types in the HRB have changed significantly(Table1).Croplands have significantly reduced in all regions,particularly in the piedmont and coastal plain,where the crop area decreased by10%.The forests slightly increased in the high mountains and decreased in the hilly mountains.The grass-lands increased in all mountain regions.Impervious surfaces increased significantly in all sub-regions due to rapid urbanization. Particularly,in the piedmont plain region,impervious surface area has increased by8.7%from1980to2008.The water body increased by5.6%in the coastal plain but did not change significantly in other regions.3.3.Spatial and temporal variations of water and energy budgetsThe water and energyfluxes for the HRB changed in the past 28years due to the effects of climate and land use changes.To reveal the spatial and temporal variations in water and energy fluxes during1981–2008in the HRB,the Mann–Kendall trend test,the simple linear trend test and variation coefficient of R n,ET p,ET a and H during1981–2008were calculated, respectively.In the mountain regions,it has been shown that R n experienced significant decreases over the past28years(Fig.5a).Only a few increasing trends appeared in the northern mountains.The trends were betweenÀ33MJ mÀ2yrÀ1and5MJ mÀ2yrÀ1(Fig.6a).From the distribution of variation coefficients(Fig.6b),it can be observed that R n is stable in the mountain region.ET a increased significantly in most mountain areas(Fig.5b).The positive trendsFig.3.Linear trends and variation coefficients of(a and b)T,(c and d)P,(e and f)sd,(g and h)rh and(i and j)ws during1981–2008in the HRB.nd use distributions at four different times(1980,1990,2000and2008)at30m resolution in theranged from0mm/yr to10mm/yr in the high mountain(Fig.6c). However,annual precipitation decreased in most mountain areas (Fig.3c).From the perspective of water budget,runoff yield has decreased in the mountain regions.It will threaten water safety in this basin.An opposite trend and variation characteristics of H to ET a are shown in Figs.5c,6e and f.From the perspective of energy budget and structure,the total energy has decreased and the distribution of latent heatflux and sensible heatflux has chan-ged in the mountain regions.In the plain regions,R n also decreased significantly(Fig.5a).The trends were betweenÀ33MJ mÀ2yrÀ1and5MJ mÀ2yrÀ1(Fig.6a), and the variation was larger than that in the mountain regions (Fig.6b).ET a significantly decreased in most plain areas(Fig.5b). The trend values ranged fromÀ26mm/yr to1mm/yr in the pied-mont plain and low plain regions(Fig.6c).The variation of the ET a is shown to be larger around the cities(Fig.6d)due to urban expansion.Decreasing trend of ET a indicates the water consump-tion has a decreasing tendency in the plain regions.However, decreases both in precipitation in the plain regions (Figs.2b and3c)and runoff yield in the mountains may weaken the decreasing trend of water deficit in the plain regions.An oppo-site trend and variation characteristics of H to ET a are shown in Figs.5c,6e and f.The positive trends of H ranged from 0MJ mÀ2yrÀ1to25MJ mÀ2yrÀ1in the piedmont and low plain regions.Especially in the urban regions,H significantly increased by more than22MJ mÀ2yrÀ1,as has occurred in Beijing city.It demonstrates that the energy distribution also changed in the plain regions.3.4.Effects of climate change and human activities on water and energy budgetsThe analysis above shows the characteristics of spatial and tem-poral changes in water and energyfluxes over the past28years in the HRB.However,the factors that are responsible for these changes must be determined.To investigate the effects of climate variability and human activities on land surface water and energy process,the authorsfirst calculated and analyzed the regional mean water and heatfluxes,Bowen ratios,meteorological vari-ables,and their linearfitted trends(Table2)and land use area per-centages(Table1)in the5described topographic regions in the HRB(Fig.1).Then,the authors conducted a correlation analysis between the water/energyfluxes and meteorological variables using a Pearson correlation coefficient,a Spearman correlation coefficient and Kendall’s s rank correlation coefficient(Table3).R n in the5different regions has shown a significant decreasing trend during1981–2008(Table2),whose strength changed from strong to weak across the hilly mountain,piedmont plain,low plain,coastal plain and high mountain regions sequentially.R n is primarily affected by sunshine duration,air temperature,and rela-tive humidity,according to formulas(2–9)in Part1.It is positivelyTable1Area percentages of land use types in5topographic regions at four times(1980,1990,2000and2008)in the HRB.Region Year Cropland(%)Forest(%)Grassland(%)Water body(%)Impervious(%)Unused(%)High mountain198026.752.317.10.9 1.9 1.1 199025.852.817.2 1.0 2.0 1.2200024.551.819.30.8 2.3 1.3200822.753.119.20.9 2.9 1.2Hilly mountain198028.848.015.2 2.4 4.1 1.5 199028.047.215.6 2.8 4.8 1.6200026.745.718.1 2.2 5.4 1.9200825.544.419.5 2.1 6.2 2.3 Piedmont plain198079.2 2.5 2.0 1.912.6 1.85.Mann–Kendall trends(SI:significant increasing,SD:significant decreasing,NSI:nonsignificant increasing,NSD:nonsignificant decreasing;significant level=0.05)n ,(b)ET a and(c)H during1981–2008in the HRB.256Y.Guo,Y.Shen/Journal of Hydrology527(2015)251–261correlated with sunshine duration sd and negatively correlated with air temperature T and relative humidity rh .To describe the trends of R n in different regions,the authors analyzed the trends of meteorological variables that could possibly influence R n .Sunshine duration sd in all regions decreased (Table 2)and thus led to decreasing trends in shortwave radiation R s ;sd thus had a negative effect on R n throughout the basin.The significant increase in air temperature T (Table 2)in all regions resulted in the increase of downward longwave net radiation so that T also had a negative effect on R n ,which is also demonstrated by the negative values of the three correlation coefficients in Table 3.The relatively small decrease in relative humidity rh had no significant effect on R n (Table 2).From the comparison of the correlation coefficients between R n and the related meteorological variables (Table 3),it is found that sd had a significant positive cor-relation with R n in all the areas.Decreasing sd is the primary factor that resulted in a decreasing R n .In the mountain regions,T is also an important factor that decreased R n due to significantly negative correlations between T and R n (Table 3).In the high mountain region,the decreasing trend of R n was smaller than that in other regions because the decreasing trend of sd was significant in all other regions except in the high mountain region,and the decreas-ing trend of rh was larger than that in other regions.Over the past 28years,ET a significantly decreased in the low plain and piedmont plain regions by 2.24mm/yr and 3.53mm/yr,slightly decreased in the coastal plain and hilly mountain regions by À0.07mm/yr and À1.41mm/yr,and significantly increased in the high mountain region by 0.79mm/yr during 1981–2008(Table 2).ET a is primarily controlled by ET p and P according to Equation (19)in Part 1.In irrigated cropland,ET a is also affected by irrigation.ET p is affected by R n ,meteorological variables and vegetation conditions according to Equations (10–12)in Part 1.R n ,T ,ws ,and P are positively correlated with ET a ,and rh is nega-tively correlated with ET a .In the past 28years,R n significantly decreased throughout the basin and decreased much faster in all plain and hilly mountain regions than in the high mountain region (Table 2).R n thus showed significantly positive correlations with ET p in all regions except the high mountain region and contributed to the decreasing of ET p (Table 3).However,significant increases in T generated positive contributions to ET p in all regions.Especially in the coastal plain and high mountain regions,significant positive correlations are shown (Table 3).Therefore,ET p increased significantly in these two regions (Table 2).Furthermore,the increase in ET p in the coastal plain is also related with the decrease in rh due to the sig-nificantly positive linear correlation between ET p and rh showninFig.6.Trends and variation coefficients of (a and b)R n ,(c and d)ET a and (e and f)H during 1981–2008in the HRB.Table 2The temporal trends of net radiation (R n ),actual evapotranspiration (ET a ),sensible heat flux (H ),potential evapotranspiration (ET p ),Bowen ratio (Br ),air temperature (T ),precipitation (P ),sunshine duration (sd ),relative humidity (rh ),wind speed (ws )and the difference between P and ET a (P –ET a )revealed by linear fitted model in different topographical regions of the HRB during 1981–2008.R n(MJ m À2yr À1)ET a(mm/yr)H(MJ m À2yr À1)ET p(mm/yr)Br T(K/yr)P(mm/yr)sd (h/yr)rh ws(m s À1yr À1)P –ET a (mm/yr)Coastal plain À4.49*a À0.07À4.32* 1.05*0.00360.03*À0.44À0.027*À0.014À0.002À0.38Low plainÀ5.61*À2.24*À0.13À0.620.0048*0.04*À0.54À0.028*À0.0250.002 1.69Piedmont plain À5.73*À3.53* 2.92*À0.420.0136*0.04*À0.73À0.030*À0.0390.007* 2.80Hilly mountain À6.50*À1.41*À3.05À0.460.00280.05*À0.42À0.025*À0.046À0.0020.97High mountainÀ3.46*0.79*À5.4*1.36*À0.0118*0.08*0.69À0.006À0.063À0.030*À0.1a Positive values denote increasing trend;negative values denote decreasing trend.*Denotes significant trend (T test,significant level =5%).。