Cropland dynamics and driving mechanisms in Changsha (2000−2020) using google earth engine and an optimal-parameter geographic detector

Cropland is a critical resource for human survival, underpinning food security and social stability. Taking Changsha City as the study area, this research employed the Google Earth Engine (GEE) platform and multi-source remote sensing data from Landsat and Sentinel-2 (2000–2020). An impervious-surface-driven feature system was constructed by prioritizing impervious-surface inputs. The system integrated optimized multi-type features to extract long-term information on cropland distribution. Furthermore, a Human–Ecological Dual-Driver (HEDD) framework was developed by integrating natural and anthropogenic factors. An optimal-parameter geographic detector (OPGD) was introduced to reveal the driving mechanisms of cropland dynamics during rapid urbanization. The results demonstrate high classification reliability, with overall accuracy of 91.67−96.71% and Kappa coefficients of 83.38−93.41%. Cropland area exhibited a slight increase followed by a continuous decline, with an average of 267,800 ha. The dominant drivers of cropland dynamics shifted from Slope to NDVI and became stronger through interactions with several anthropogenic factors.