| 摘 要： |
Accurate estimation of soil temperature (T-s) at a national scale under different climatic conditions is important for soil-plant-atmosphere interactions. This study estimated daily T-s at the 0 cm depth for 689 meteorological stations in seven different climate zones of China for the period 1966-2015 with the M5P model tree (M5P), random forests (RF), and the extreme gradient boosting (XGBoost). The results showed that the XGBoost model (averaged coefficient of determination (R-2) = 0.964 and root mean square error (RMSE) = 2.066 degrees C) overall performed better than the RF (averaged R-2 = 0.959 and RMSE = 2.130 degrees C) and M5P (averaged R-2 = 0.954 and RMSE = 2.280 degrees C) models for estimating T-s with higher computational efficiency. With the combination of mean air temperature (T-mean) and global solar radiation (R-s) as inputs, the estimating accuracy of the models was considerably high (averaged R-2 = 0.96-0.97 and RMSE = 1.73-1.99 degrees C). On the basis of T-mean, adding R-s to the model input had a greater degree of influence on model estimating accuracy than adding other climatic factors to the input. Principal component analysis indicated that soil organic matter, soil water content, T-mean, relative humidity (RH), R-s, and wind speed (U-2) are the main factors that cause errors in estimating T-s, and the total error interpretation rate was 97.9%. Overall, XGBoost would be a suitable algorithm for estimating T-s in different climate zones of China, and the combination of T-mean and R-s as model inputs would be more practical than other input combinations. |
在线留言
联系我们
所长信箱