农作物受灾, 时空格局, 影响因素, Catboost模型, Tobit回归模型

In order to deeply understand and reveal the relevant laws and influencing factors of crop disasters, the spatiotemporal evolution characteristics of crop disasters in the mainland of China were comprehensively discussed by using kernel density estimation and Dagum Gini coefficient. On this basis, the influencing factors of crop disasters were analyzed by using Catboost model and Tobit model. The results are concluded that the overall level of crop damage in the mainland of China shows a belt-like distribution characteristics, and it can be found that the level of crop damage in the Loess Plateau region has been relatively high for a long time. On the whole, the degree of crop damage in the north of the Loess Plateau is stronger than that in the south of the Loess Plateau, while the overall crop damage level in the Central China Plain is relatively high in the south of the Loess Plateau. The Gini coefficients of the eastern and western regions are always higher than those of the central and northeast regions. With the passage of time, the spatial distribution difference in the eastern part of the country gradually decreases, but in the western region it has a tendency to expand, and the difference in the western region is mainly reflected in the fact that the overall level of crop damage in the northwest is higher than that in the southwest. Water conservancy construction and soil erosion control are the leading influencing factors of crop disasters in the mainland of China.

crop damage, spatiotemporal pattern, influencing factors, Catboost model, Tobit regression model