Dynamic changes in cropland under the cropland balance policy: a case study of Hunan Province

Changes in cropland are critical for ensuring national food security. This study analyzed the dynamics of cropland based on the implementation, adjustment, and enhancement of cropland balance policies in Hunan Province. The results indicated that the deep learning model used for cropland extraction was reliable (overall accuracy of 88% and a kappa coefficient of 0.85). The Geo-Informatic Tupu revealed that the cropland area decreased by 3.15% between 1999 and 2014, with the loss of croplands to forests and grasslands. Cropland increased by 0.96% from 2014 to 2017 due to the transfer of grasslands. From 2017 to 2021, the cropland declined by 0.4%. The regression coefficient of mixed land use and cropland distribution shifted from negative (−0.83 to −0.79) to positive (−0.78 to 0.81) between 1999 and 2017 and reached a high value by 2021. Higher mixed land use correlates with more cropland, while regional differences require tailored policies to protect cropland.