The Related Driving Factors of Water Use Efficiency and Its Prediction Model Construction in Winter Wheat
[Objective]The water use efficiency can comprehensively reflect the growth suitability and energy conversion efficiency of winter wheat.The driving factors of winter wheat in response to standardized water use efficiency(WP*)at different growth stages were screened and explored,and the WP* prediction model of related driving factors was constructed,which was of great significance for the monitoring of water use efficiency and efficient use of water resources in winter wheat in the Huang-Huai-Hai Plain.[Method]Three water treatments were set up,including water deficit treatments(W1:35 mm,and W2:48 mm)and control treatment(W3:68 mm),and the canopy temperature parameters,physiological indexes and standardized WP* of winter wheat at the jointing,booting and filling stages were measured.Stepwise regression and pathway analysis were used to screen the main driving factors in response to WP* changes at each growth stage,the relationship between WP* and related drivers was explored,and finally the partial least squares regression(PLSR)and support vector machine(SVM)methods were used to construct a driver-based WP* prediction model in each growth stage.[Result]Compared with W3,the canopy temperature parameters,physiological indexes and WP* of winter wheat under the water deficit treatments showed significant differences.Based on the stepwise regression method,the main driving factors in response to WP* at each growth stage were screened,and the sensitivity of each driving factor in response to WP* was ranked by pathway analysis,that is,maximum temperature difference(MTD),stomatal conductance(Gs),leaf water content(LWC)and POD were selected at the jointing stage;canopy relative temperature difference(CRTD),equivalent water thickness(EWT),soluble sugar content(SSC)and crop water stress index(CWSI)were selected at the booting stage;SSC,standard deviation of canopy temperature(CTSD),LWC and Gs were selected at the filling stage.Finally,the driver-based WP* prediction model for each growth stage was construct by using PLSR and SVM.Among them,the prediction model of WP* at booting stage constructed by SVM had the best accuracy,with R2cal(R2val),RMSEcal(RMSEval)and nRMSEcal(nRMSEval)of 0.945(0.926),0.533 g·m-2(0.580 g·m-2)and 2.844%(3.075%),respectively.[Conclusion]By screening the relevant driving factors of WP* at each growth stage of winter wheat and constructing a prediction model of winter wheat water use efficiency,this paper provided a theoretical basis for accurate monitoring and management of winter wheat moisture in the Huang-Huai-Hai Plain.
winter wheatstandardized water use efficiency(WP*)driving factorpathway analysissupport vector machine
国家科技期刊平台
NSTL
万方数据
