Hysteresis in flag leaf temperature based on meteorological factors during the reproductive growth stage of wheat and the design of a predictive model

Abstract

The physiological state of functional leaves in crops plays a vital role in yield formation. Over two consecutive winter wheat growing seasons, we continuously monitored the flag leaf temperature (Tf) during the reproductive growth stage and collected key meteorological indicators, including air temperature (Ta), relative humidity (Ha), soil temperature (Ts), and photosynthetically active radiation (PAR). Pearson correlation analysis, stepwise regression analysis, and path analysis revealed that Ta, PAR, Ts, and Ha are the main environmental factors influencing Tf. These variables were identified as key for further analysis. Notably, Tf exhibited a positive time lag correlation with PAR, while Ta and Ts lag showed positive lag correlation with Tf, and Ha demonstrated a negative lag correlation with Tf. Among the analyzed meteorological factors, soil temperature displayed the smallest lag effect relative to Tf, consistently trailing behind it. PAR showed a pronounced lag effect, shifting an hour earlier than Tf, while Ta exhibited a significant hour-long delay after Tf. Ha primarily functioned as a cooling influence, lagging approximately one hour behind Tf. Moreover, the intensity of the time delay effect will vary depending on the developmental stage. Integrating these time-lag relationships significantly enhanced the accuracy of Tf simulations. Support Vector Regression (SVR) demonstrated robust predictive performance (R² = 0.937, RMSE = 2.048 °C), indicating its potential for accurate prediction of Tf in wheat production. This study highlights the time-delay effects between Tf and meteorological factors during the reproductive growth stage of wheat, offering a predictive model that provides a foundation for monitoring crop physiological conditions in real time.