Seasonal Pasture Distribution Maps of Qinghai-Tibet Plateau from 1990 to 2020

This dataset provides detailed seasonal pasture distribution maps of the Qinghai-Tibet Plateau for the period 1990–2020, with a 30-meter spatial resolution and five-year intervals. The dataset includes three categories: warm-season pastures (3), cold-season pastures (4), and non-pasture areas (5). The data is generated through an innovative technical framework that integrates machine learning algorithms and Landsat imagery. The framework uses time-series spectral analysis to determine the optimal time window for differentiating cold- and warm-season pastures. Based on this window, training samples are automatically generated and purified, which are then used to train a Random Forest classifier. The classifier is applied to produce the seasonal pasture maps. Validation against independently interpreted reference samples demonstrates that the overall classification accuracy for the seasonal pasture maps across all time periods exceeds 93.10%.