NDVI prediction dataset for the Qinghai-Tibet Plateau based on CMIP6 (2025–2100)

All processing and modeling tasks, including simulation and error control, were performed using MATLAB, with masking applied to non-vegetation areas such as glaciers and water bodies. The dataset spans 76 years, covers the entire Qinghai-Tibet Plateau, has a spatial resolution of 1 km, and consists of 228 GeoTIFF raster files with a total volume of 5.65 GB. Model accuracy was evaluated using the coefficient of determination (R²) and root mean square error (RMSE), with all R² values exceeding 0.72 and RMSE values controlled below 0.13, demonstrating high reliability.

This dataset features multi-scenario comparability, long-term temporal continuity, and high spatial detail, making it valuable for studies on ecosystem services simulation, regional carbon sink assessment, and vegetation dynamics under climate change scenarios.