A multi-source eco-hydrological dataset for Northern, Southern, and Eastern Xinjiang (2000–2023)

This dataset provides a spatially explicit eco-hydrological data collection for Xinjiang, China, covering the years 2000–2023. It integrates long-term remote sensing and climate records to support studies of vegetation–water interactions, land cover dynamics, and environmental attribution across arid landscapes. All data are resampled to a unified 825 m resolution and georeferenced in WGS84 coordinate system. **1. Original Remote Sensing and Climate Data** This section includes year-by-year observations from publicly available Earth observation sources:- MODIS-derived indicators: Gross Primary Productivity (GPP), Evapotranspiration (ET), NDVI, and LAI - TerraClimate variables: Precipitation (PR), Soil Moisture (SOIL), and Average Temperature (TEMP) - Land cover: Annual maps from the China Land Cover Dataset (CLCD) - Anthropogenic proxy: Night-time light intensity (DMSP-OLS) - Shapefiles: Ecohydrological zoning of Northern, Southern, and Eastern Xinjiang **2. EcoIndex and Coupling Indicators** This section contains composite indicators and diagnostic layers derived from the original variables:- EcoIndex: A dimension-reduced composite index integrating water use efficiency (WUE = GPP/ET) and NDVI, using a maximum variance projection method - ESI (Ecohydrological Synchrony Index): Cosine similarity metric between NDVI and WUE, reflecting vegetation–water coordination strength - Quadrant classification maps: Annual categorical layers representing four vegetation–water functional regimes (co-enhancement, co-degradation, stress release, emerging stress) **3. Attribution of Climatic and Anthropogenic Drivers** This section includes pixel-wise attribution products generated from machine learning models:- Temporal trend layers: Sen’s slope and Mann–Kendall trend significance for EcoIndex and ESI - Driver importance maps: Random forest–based variable importance for climatic (PR, TEMP, SOIL) and anthropogenic (night-time lights, land cover) factors - Tabular summaries: - `trend_statistics.csv`: Regional statistics of upward/downward/stable trends and average slope magnitudes - `Regional Statistics of Driver Importance Means and Medians in Xinjiang.csv`: Regional averages and medians of driver importance scores All layers are stored in GeoTIFF, CSV, or ESRI Shapefile formats and organized into subfolders by category and year. The dataset supports eco-hydrological research, spatiotemporal modeling, and environmental attribution analyses across dryland regions.