Grassland Resilience Assessment Using Explainable Machine Learning in Arid Ecosystems of Northwest China (2001–2023)

📝 Dataset Description

This dataset supports the study “A Study of Explainable Machine Learning Method to Explore Grassland Resilience in the Arid Ecosystem”. It focuses on grassland ecosystems in northwest China, covering the period 2001–2023. The project integrates multiple data sources and explainable machine learning techniques to analyze the resilience of vegetation under climate variability.

The dataset includes:

Raw climate and vegetation data: Including weather station records, ERA5 reanalysis data, MODIS NDVI, and FVC datasets at 1 km resolution.

grassland classification maps: For spatial filtering and labeling of grassland types.

Processed environmental indicators: Such as long-term vegetation trends (TAC), climate variability metrics (CV, AC1), and predicted climate variables.

Machine learning results: From random forest models and SHAP-based explainability analyses.

Visualization scripts and boxplots: Showing grassland type responses and feature importance.

All data and code are organized under a reproducible structure for climate interpolation, ecological modeling, and explainable ML analysis.