Cropland Abandonment in China Dataset

This dataset provides high-resolution spatial and statistical data on cropland abandonment across China from 1990 to 2023.The dataset consists of three main components:Annual and total abandoned cropland from 1990-2023 (GeoTIFF): These files were generated based on long-term land-use datasets.Prefectural-level Attribute Data (Excel): Prefectural-level attribute dataset is used for analyzing the potential factors of cropland abandonment. Variables include socio-economic, demographic, topographic, climatic, landscape and other indicators. Each record corresponds to a prefecture-level administrative unit in China.Python and Jupyter Code Files: The files of .py are designed to identify annual and total abandoned cropland based on multi-year land-use data. The .ipynb file for training and testing six machine learning models (LightGBM, XGBoost, Random Forest, Gradient Boosting, CatBoost, AdaBoost) is to analyze factors at the prefectural scale.Together, these datasets and codes enable the reproduction of the spatial identification and factor analysis of cropland abandonment in China and support future research on sustainable land-use transitions.