基于连续变化检测和分类算法的重庆市长寿区耕地非农化监测OA

The promotion of urbanization has led to the reduction of cultivated land area,which impacts grain production and food security.Therefore,research on monitoring the conversion of arable land to non-agricultural uses holds significant importance.The Changshou district of Chongqing was taken as the study area.Initially,the continuous change detection classification(CCDC)algorithm and Landsat images was used to detect the time of land cover type change points from 2008 to 2023 on the google earth engine(GEE)cloud platform.Then,the shapley additive explanation(SHAP)interpretable framework and Bayes-optimized categorical boosting(CatBoost)model were utilized to produce land cover classification maps in 2008,2013,2018,2023,respectively,and the quantitative analysis of cultivated land non-agricultural was carried out.The results showed that the optimized CatBoost model could classify land cover with the highest accuracy of 88.50％and Kappa coefficient of 0.843 0.The non-agricultural of cultivated land showed a trend of first increasing and then decreasing from 2008 to 2023.The non-agricultural area was the least from 2008 to 2013,and the non-agricultural area increased from 2014 to 2018,which was due to the policy of multiple rounds of returning farmland to forest and grassland and the acceleration of urbanization process.The non-agricultural area of cultivated land decreased significantly from 2019 to 2023,which was due to the policy of prohibiting non-agricultural cultivation of cultivated land.Above results provided scientific support for cultivated land protection and food security.