Abstract: 【Objective】The objective of this study is to establishment of inversion models for winter wheat plant water content at different growth stages based on three comprehensive indicators of winter wheat canopy temperature, morphology, and physiology, explore more comprehensive and accurate water deficit monitoring methods to provide theoretical basis for drought resistance work in winter wheat.【Method】The winter wheat was used as the research object, and three water treatments were set up, two wheat varieties: Luomai 22 and Zhoumai 27. Canopy temperature parameters (canopy temperature standard deviation (CTSD) and crop water stress index (CWSI)), morphological indicators (plant height, stem diameter, aboveground biomass, and LAI) and physiological indicators (stomatal conductance, transpiration rate, and photosynthetic rate) of winter wheat were obtained at jointing, booting, and filling stages, respectively. comprehensive temperature parameter indicators (CTPI), comprehensive growth indicators (CGI) and comprehensive physiological indicators (CPI) based on the average weight principle were constructed. the correlation between plant water content (PWC) and comprehensive indicators was analyzed, and multiple linear regression (MLR), partial least squares recurrence (PLSR) and support vector machine (SVM) methods were used to construct the PWC inversion model based on comprehensive indicators according to the growth period.【Result】The canopy temperature parameters, morphology and physiological indexes of winter wheat in different growth stages showed significant differences between water deficit treatment (W1, W2) and control treatment (W3) (P＜0.05). Comprehensive indicators (CTPI, CGI and CPI) at booting and filling stages have a good correlation with PWC, with correlation coefficients (r) of -0.70 (-0.78), 0.84 (0.80) and 0.83 (0.76), respectively. Using MLR, PLSR and SVM methods, the PWC inversion prediction model based on comprehensive indicators (CTPI, CGI and CPI) has high prediction accuracy, among which the PWC model built by SVM is the best, R²_cal (R²_val), RMSE_cal (RMSE_val), and nRMSE_cal (nRMSE_val) were 0.878 (0.815), 0.021 (0.024), and 3.10% (3.33%), respectively.【Conclusion】The SVM-PWC model based on the comprehensive indicators CTPI, CGI and CPI can well predict the water deficit of winter wheat at different growth stages, and provide theoretical basis for drought prevention and drought resistance of winter wheat in the Huang-Huai-Hai region.

Key words: winter wheat, water deficit, comprehensive index, plant water content, support vector machine