Improving Evapotranspiration Model Performance By Energy Closure And Parameter Optimization

As an important process of energy exchange and material exchange between land and air,evapotranspiration(ET)is closely related to the water energy cycle.Accurate simulation of ET through ET models is of great significance for efficient utilization of water resources and climate change.At present,Shuttleworth-Wallace dual-source evapotranspiration model(SW model for short)has been widely used in the simulation of evapotranspiration and has been applied well in many regions.However,the flux measured by the eddy system is not closed in energy,which has a certain influence on the calibration of model parameters and the prediction of the model.In addition,evapotranspiration model is only studied through latent heat flux at present,and evapotranspiration process is a complex process,and the change of latent heat flux is directly accompanied by the change of sensible heat flux.The parameter calibration of single target data leads to the uncertainty of model parameters and prediction.In view of this,based on eddy data and meteorological data of maize farmland ecosystem at Daman Station and alpine meadow ecosystem at Arou Station in the Heihe River Basin,this paper considered multiple target data(latent heat flux and sensible heat flux),introduced energy closure factor,combined with Bayesian method for parameter optimization,and put forward three optimization schemes:Scheme 1 is a single target data,that is,only the latent heat flux data are considered in the observation data.Scheme2 is multi-objective data,namely,the observation data takes into account both latent heat flux and sensible heat flux data.In scheme 3,energy closure factor is introduced on the basis of scheme 2.According to each evaluation index,the simulation effect of latent heat flux and sensible heat flux simulated by the original SW model and three optimization schemes during the calibration period and verification period was comprehensively evaluated.The main conclusions are as follows:(1)According to the energy of the heihe river basin different sites and SW model evaluation findings closed,The Heihe River basin vortex system there are two site energy is not closed,the big full standing corn farmland ecosystem energy closure rate is 0.83,energy is not closing rate of 0.17,the soft station of alpine meadow ecosystem energy closure rate is 0.78,The energy non-closure ratio is 0.22.When the parameters of the original SW model are not optimized,the original SW model can only simulate the approximate changes of latent heat flux and sensible heat flux,but the simulated value of latent heat flux in each time period is lower than the measured value,and the simulated value can only reach half of the measured value.Because the sensible heat flux is directly affected by the simulation of latent heat flux,the simulated value of sensible heat flux is higher than the measured value,and the simulated value is more than twice of the measured value.And the simulation effect of sensible heat flux is poor compared with that of latent heat flux.(2)Based on energy closure and parameter optimization,three optimization schemes are proposed,and DREAM algorithm is used to optimize and calibrate the parameters of the model for the two stations respectively.The parameter uncertainties of the three optimization schemes are all reduced.Compared with the observation data considering both latent heat and sensible heat,the confidence interval of the parameters is shortened,indicating that the uncertainty of the parameters is reduced.The introduction of energy closure factor in the model can directly affect the distribution of K_a and K_q parameters in the model,and the frequency of other parameters is significantly increased,indicating that the introduction of energy closure factor in the model can affect the model parameters,improve the probability of model parameters,reduce the uncertainty of parameters,and thus improve the reliability of model parameters.(3)The combination of qualitative evaluation,quantitative evaluation index evaluation method,and the ratio between the two sites during the calibration period and validation periods,latent heat flux in simulation,three optimization simulation performance than the original model,and the observation data at the same time considering the latent and sensible heat flux data and only considering the latent heat flux data of simulation of promotion effect is almost the same,The scheme introducing energy closure factor in the model has the best simulation lifting effect.In terms of sensible heat flux,the three optimization schemes have been improved to different degrees.Considering both latent heat flux and sensible heat flux data,the simulation performance improvement effect is better than that of only considering latent heat flux data.The simulation effect of introducing energy closure factor into the model is the best,and the simulation performance improvement effect is the largest.In addition,the application of energy closure factor in simulating sensible heat flux in the alpine meadow ecosystem is better than that in the maize farmland ecosystem.