Integrated drought monitoring and analysis: A novel framework based on multi-source remote sensing data and ensemble...

Area of Science:

• Environmental Science
• Remote Sensing
• Machine Learning

Background:

• Drought risks are escalating globally due to climate change, impacting agricultural regions.
• Existing drought monitoring methods lack accuracy, spatial representativeness, and explanatory power.
• Effective drought monitoring tools are crucial for agricultural resilience.

Purpose of the Study:

• To develop and validate a novel framework for drought monitoring using integrated multi-source remote sensing data and an ensemble machine learning model.
• To enhance the accuracy and interpretability of drought prediction and driver analysis.
• To provide a scalable framework for data-driven drought risk management.

Main Methods:

• Integration of multi-source remote sensing data.
• Application of a Bayesian-weighted ensemble machine learning model.
• Validation using the Beijing-Tianjin-Hebei-Shandong-Henan region, China.
• Utilized SHAP (SHapley Additive exPlanations) for driver interpretability.

Main Results:

• The ensemble model achieved high accuracy in predicting the Standardized Precipitation Evapotranspiration Index (SPEI) (R2: 0.71-0.74) across multiple time scales.
• Accurate classification of drought severity (over 78%) and extreme drought detection (98%).
• Identified precipitation anomalies and potential evapotranspiration as key short-term drought drivers, and soil moisture as critical for long-term drought.

Conclusions:

• The proposed framework offers an effective and interpretable tool for regional drought monitoring and analysis.
• The study highlights the importance of integrating diverse data sources and advanced ML techniques for drought management.
• The findings support data-driven drought risk management strategies in vulnerable agricultural areas.