Patterns and drivers of terrace abandonment in China: Monitoring based on multi-source remote sensing data

As urbanization and industrialization surge in China, the problem of land abandonment intensifies. However, the situation of terrace abandonment in China remains unclear. We conducted the first-ever remote sensing monitoring of terrace abandonment in China with full space coverage by combining high-precision terrace data with land use datasets to reveal the abandonment pattern. By fully considering natural and socio-economic factors, the XGBoost-SHAP framework was used to investigate the driving factors of terrace abandonment. The results show that approximately 2.42 % of terraces were abandoned from 2019 to 2021, mainly distributed in the Southwest and Loess Plateau regions. Agricultural regions with more terraces exhibited higher abandonment rates. The ratio of the population with pension insurance, cropland quality, slope, and land parcel size were prime drivers of terrace abandonment. There were significant spatial differences in the contribution of each factor. It is noteworthy that there was a significant deceleration in terrace abandonment trends in 2021, potentially ascribed to the impact of the COVID-19 pandemic leading to a substantial decrease in non-agricultural employment opportunities, thereby slowing down rural-to-urban emigration and even prompting a return migration of migrant workers. Grasping this critical post-pandemic period is crucial and should support returning migrant workers in engaging in agricultural activities by establishing diverse new agricultural entities and providing agricultural technical guidance.