Estimation of Hourly Air Temperature in China Based on LightGBM and Himawari-8

Air temperature (Ta) is one of the most fundamental and important variables in ecological and environmental science. In this work, we fuse the brightness temperature data of Himawari-8 and auxiliary data such as terrain, vegetation, time, and meteorological elements to develop a model based on the Light Gradient Boosting Machine (LightGBM). The hourly high-resolution Ta is estimated and the spatial distributions are mapped in China. The evaluation results show that the model performs well in the study area. Based on the five-fold cross-validation of all samples, the R
²
is 0.986 and the RMSE is 1.598 K. The spatial distribution patterns can well reflect the Ta changes in different regions of China.