Evapotranspiration And Its Components Estimation By Generalized Complementary Model And Linearized Carbon And Water Relationships On The Loess Plateau

The land surface evapotranspiration(denoted as E),which is equivalent to the term“evaporation”,is the link between water balance and energy balance,and plays a key role in connecting the ecological and hydrological processes.It is a difficult issue to estimate evapotranspiration accurately due to the complex land-atmosphere interactions.The generalized complementary relationship for evapotranspiration has attracted much attention because of its rigorous physical boundary conditions and the fact that it can be implemented with conventional meteorological data without the need for the complicated soil moisture or vegetation resistance parameters.In the past few decades,the changes of hydrological processes and sustainable utilization of water resources have caused extensive concern due to the large-scale water conservation engineering measures such as the“Grain for Green”project.In this study,three generalized complementary relationship methods have been verified and compared in the Loess Plateau,and then the performances of annual evapotranspiration estimation have been evaluated by the water-balance-derived evapotranspiration,RS-based,LSMs-based and machine-learning-based evapotranspiration data.Further,the generalized advection aridity model(denoted as GNAA)with the best performance has been applied to estimate the evapotranspiration of the Loess Plateau.In general,the total land surface evapotranspiration includes evaporation(denoted as E_s)of soil and other underlying surface components,and transpiration(denoted as T)of vegetation.In the agricultural ecosystem,quantifying crop transpiration and determining the proportion of each component of evapotranspiration are of great significance for understanding water use efficiency index thoroughly,improving crop yield and managing water resources.Therefore,the ratio of transpiration of winter wheat over evapotranspiration(denoted as T/E)in different time scales were estimated based on regional evapotranspiration estimation on the Loess Plateau.The main results of this paper are summarized as follows:(1)Determining the parameter calibration schemes of GNAA and Sigmoid type generalized complementary function(denoted as SGCF)for accurately estimating the annual evapotranspiration,and clarifying the performance characteristics of GNAA and SGCF models in evapotranspiration estimation.The scatterplots-trend of dependent variable,that is the ratio of actual evapotranspiration over apparent potential evapotranspiration(denoted as y=E/E_pa)versus independent variable,that is the ratio of equilibrium evapotranspiration over apparent potential evapotranspiration(denoted as x=E_e/E_pa)could be captured by the GNAA curve when both?_e and c parameters are calibrated.As long as the calibrated parameter c is greater than 2,the boundary condition of x_min greater than 0 can be achieved.For the SGCF,not only parameters?_e and b^-1 need to be calibrated,but also the upper and lower limits of E_e/E_pa must be constrained.After calibration,both the GNAA and SGCF performed comparably in estimating annual evapotranspiration,and the calibrated?_e in the SGCF is closer to the average value of P-T coefficient of 1.26than that of the GNAA.(2)Detecting functional differences between the basic and extended GNAA forms in estimating evapotranspiration by exploring the potential influence factors and the connotation of parameters.In GNAA,the parameter?_e was correlated with the aridity index(E_pa/P),precipitation(P)and climate seasonality and asynchrony index(SAI),which indicated that?_e is an integrated parameter involving parameter?in P-T equation and other factors.In the Loess Plateau,most of?_e-0 in the basic GNAA are less than 1.0,and it has stronger relationship with E_pa/P.For the extended GNAA,the parameter?_e-c is closer to the P-T coefficient?.However,the basic GNAA form with a semi-empirical function of?_e-0=f(E_pa/P)could accurately simulate evapotranspiration and its inter-annual variation on the catchment scale.The combination of the two forms of GNAA can better estimate evapotranspiration and describe the relationship between E,E_pa and Epo.(3)Elucidating the relationship between the nonlinear generalized rescaling complementary relationship(denoted as S2017)which was proposed by Szilagyi et al.(2017)and the GNAA with parameter?_e-0 function,and comparing the performance of regional evapotranspiration estimation.In addition,applying the GNAA model to estimate the evapotranspiration and analyzing its inter-annual variation,and exploring the hydrological effects associated with the vegetation restoration predominated in ecological construction on the Loess Plateau.The parameter?_e-0 function of the basic GNAA actually rescaled the independent variable(x _B)in the extended GNAA from the different direction.It takes the?_e-0/?_e-c as the scaling factor and satisfies the boundary of“y=0 as?_e-0*E_e/E_pa=0”.The mean evapotranspiration(denoted as E_CR-B)was 391.8 mm/year,and the E_CR-B and P decreased at the rate of 0.33 mm/year and 0.99 mm/year from 1960 to 2011 on the Loess Plateau,respectively.However,annual E_CR-B/P increased significantly at the rate of0.0013.It indicated that the proportion of evapotranspiration to precipitation is increasing at the long time-scale.Evapotranspiration is controlled by precipitation on the Loess Plateau,and the absolute change rate of evapotranspiration is generally less than that of precipitation in different periods although both of them have the same downward trend.Since the large-scale ecological restorations projects have been launched in 1999,the precipitation and evapotranspiration have respectively increased at the rate of 3.26 mm/year and 1.44 mm/year,but the upward rate of evapotranspiration is still lower than that of precipitation.Therefore,the amount of available water resources has not decreased during the vegetation restoration process on the Loess Plateau.When the climate condition does not change significantly,the stability of vegeation supported by available water resources is guaranteed on the Loess Plateau.(4)Using the slope stability of linearized carbon and water relationship,that is the relationship between GPP.VPD^k and E(where GPP,VPD,E and k denoted gross primary productivity,vapor pressure difference,evapotranspiration and coefficient,respectively),the proportion of transpiration of winter wheat to evapotranspiration and its variation in the growth period were estimated by the u WUE(i.e.,u WUE=GPP.VPD^0.5/E)method,against the background of evpotranspiration estimation on Loess Plateau.The linearized carbon and water relationship at different time scales can be expressed as WUE_k=GPP.VPD^k/E.We found that k coefficient decreased with the time scale by comparing the suitable time scale for several water use efficiency indexes.The u WUE index with the best performance on half-hour scale was used to partition E of winter wheat on the Loess Plateau,where the mean value of the transpiration over the evapotranspiration(denoted as T/E)of winter wheat during its growth was 0.57,and the dynamic change in T/E on the five-day time scale showed a bimodal pattern during the growing season.In addition,the higher peak occurred around the booting stage and exceeded 0.80.The diurnal variations in the winter wheat T/E had an“M-shaped”trend.However,different WUE indexes have their own application conditions,and the traditional WUE index is irreplaceable in charactering the relationship between water consumption and photosynthetic production in the monthly and annual scales.