Estimating daily minimum grass temperature to quantify frost damage to winter wheat during stem elongation in the central area of Huang-Huai plain in China

Frost damage to winter wheat during stem elongation frequently occurred in the Huang-Huai plain of China, leading to considerable yield losses. Minimum Stevenson screen temperature (ST min ) and minimum grass temperature (GT min ) have long been used to quantify frost damage. Although GT min has higher accuracy than ST min , it is limited in application due to the lack of data. Therefore, this study aimed to select appropriate environmental variables to estimate GT min , as well as to quantify the frost damage. Shangqiu, a frost-prone winter wheat area in the central Huang-Hui plain, was selected as the study area. From the descriptive statistics of ST, air relative humidity (RH), wind speed (WS), cloud fraction (CF), and volumetric soil water content (VWC) during temperature decreasing and increasing, seven variables significantly correlated with GT min were selected, including ST min , maximum reduction of ST (RST), maximum increase of ST (IST), minimum RH during temperature increasing (RH min ), WS at ST min occurrence (WS), minimum VWC during temperature decreasing (VWC min ), and nightly CF. Multiple linear regression (MLR), support vector regression (SVR), random forest (RF), and K-nearest neighbor (KNN) were adopted for estimating GT min based on the various combinations of the variables. Results showed the more variables, the higher the accuracy for the MLR and SVR. However, this pattern was not always true for the KNN and RF. The KNN based on ST min , RST, IST, RH min , and WS achieved the highest accuracy, with R 2 of 0.9992, RMSE of 0.14 ℃, and MAE of 0.076 ℃. The overall classification accuracy for frost damage identified by the estimated GT min reached 97.1% during stem elongation of winter wheat from 2017 to 2021. The integrated frost stress (IFS) index calculated by the estimated and measured GT min maintained high linear fitting accuracy. The KNN with fewer variables demonstrated good applicability at the regional scale.