The Spatiotemporal Variability Of Catchment Evapotranspiration And Its Attribution Analysis In The Loess Plateau Under The Budyko Framework

The catchment hydrological processes change with climate variability and surface condition alterations.In particular,as the linkage between hydrology and ecology,evapotranspiration has been paid great attentions.The Budyko framework,connecting climatic conditions to runoff and actual evapotranspiration in a catchment,has solid theoretical basis and clear physical meaning,and is thus a powerful tool to attribute the changes in hydrological processes in the world.The ecoenvironment of the Loess Plateau(LP)has been changing over the past 50 years due to the extensive and intensive human activities,including the Grain to Green Project as well as the constructions of energy bases and new towns.Under the above background,further investigation should focus on how and why the evapotranspiration have changed.Therefore,we first analyzed the spatiotemporal variations in water and heat conditions and conducted the quantitative attribution analysis for the changes in potential evapotranspiration(ET0)using the total differential method.Then,we explored the relationships between the catchment water-heat coupled controlling parameter and vegetation dynamics(M)as well as climate seasonality(S),and further discussed the correlations among the factors impacting controlling parameter and the time-scale effect of these correlations.Finally,the contributions of the change in M and S to ET variation were assessed.The main conclusions are as follows:First,the annual temperature averaged across LP increased significantly(p<005),while ET0 from the FAO 56 Penman-Monteith equation showed a downward trend(p>0.05),indicating the ‘evaporation paradox' phenomenon in LP.The intra-annual variations and fluctuation of P was greater than those of ET0,and the climate seasonality index S exhibited a downward trend,which means the seasonal fluctuation of the difference between P and ET0 tend to decrease.Second,the attribution analysis of the spatiotemporal changes in ET0 during 1960-2012 using the total differential method indicated that ET0 was most sensitive to actual vapour pressure(ea),followed by solar radiation(Rs),mean temperature(Tmean)and wind speed(U2).Meanwhile,ET0 was impacted by the actual change in each climatic variable.As a result,the major factor influencing the temporal variation of ET0 was the change in Tmean,followed by U2 and Rs,and the contribution of U2 was the least.Spatially,the dominant factors decreasing ET0 from northwest to east and to south were the decreasing Rs and increasing ea,and decreasing U2 induced a downward trend in ET0 from northwest to southwest.Third,after analyzing the relationships between water-heat coupled controlling parameter ? and M as well as S for 13 catchments,we found that: parameter ? presented an upward trend,especially after the 1980 s.The parameter ? generally had a positive correlation with M,while negatively correlated with S.Considering the above relationships and the limiting conditions of M and S,the semi-empirical formula of parameter ? was developed.The results of cross validation showed that the semi-empirical formula was superior for estimating the annual ET in LP.Fourth,the correlations between the factors impacting controlling parameter on different time scales were investigated.On long-term scale,M and S were highly related;however,the stepwise regression showed that multicollinearity existed between M and S.Therefore,as the key factor,vegetation information could fit the controlling parameter individually.On annual scale,the correlation between M and S was not significant in most catchments;therefore,M and S could be considered into the controlling parameter on annual scale.The relief amplitude of catchment land surface was better than topographic wetness index and slope to explain the effects of topography on catchment water balance under the Budyko framework.With the implement of the Grain for Green Project,the vegetation coverage is increasing with the relief amplitude.Finally,after comparing the four attribution methods for ET change,i.e.the total differential method,extrapolation method,complementary method and decomposition method,we found that the complementary method was the best.From 1980 to 2012,ET was most sensitive to P,followed by parameter ? and ET0.During the past 30 years,ET became less sensitive to ? and ET0 but more sensitive to P.The improved vegetation conditions positively contributed to ET changes,while the increasing climate seasonality had a negative effect on ET.The contribution of landscape condition to ET would be estimated with a large error if the effects of climate seasonality are ignored.Based on the Budyko hypothesis,this study analyzed the spatiotemporal variations of evapotranspiration in LP and the attribution from the changes in climate and surface condition.This study deepened and extended the understanding of the factors that impacted the water-heat coupled controlling parameter.Our research provides important information for water resources management,agricultural production improvement and ecoenvironment construction.