Earth, Volume 6, Issue 3
2025 September - 51 articles
Cover Story: Machine learning models are widely used for streamflow prediction due to strong performance, but their data-driven nature hinders interpretation. This study examines Random Forest interpretability for high-streamflow events, comparing feature-importance methods. Mean decrease accuracy, impurity, SHAP, and Tornado highlight similar features, though Tornado differs most. The last observed streamflow demonstrates the highest importance (>20%), despite temporal variability. Results reveal a key catchment region influencing outlet flow. Accumulated local effects and partial dependence plots show infiltration and soil saturation before rainfall impacts streamflow. Short-term precipitation is critical during rising limbs (~72% importance), while prior streamflow dominates near peaks and falling limbs. Models may reasonably represent catchments and offer hydrological insights. View this paper - Issues are regarded as officially published after their release is announced to the table of contents alert mailing list .
- You may sign up for email alerts to receive table of contents of newly released issues.
- PDF is the official format for papers published in both, html and pdf forms. To view the papers in pdf format, click on the "PDF Full-text" link, and use the free Adobe Reader to open them.