Context-Aware Concept Distillation for Trustworthy Flood Prediction

2026-07-25

AI Summary

This study addresses the limited interpretability of existing deep learning models for flood forecasting, which undermines their suitability for emergency decision-making that demands transparency and trustworthiness. To bridge this gap, the authors propose a Context-Aware Concept Distillation (CACD) framework that unsupervisedly discovers a “hydrological language” and integrates it with a residual hypernetwork to distill a black-box LSTM into an interpretable surrogate model grounded in hydrological semantics. This approach enables dynamic, catchment-characteristic-driven modeling. Evaluated across 5,203 global catchments, the method achieves a median Nash–Sutcliffe Efficiency (NSE) of 0.70, significantly outperforming black-box baselines such as MLPs, while delivering operationally verifiable, causal explanations at the decision-relevant level.

Abstract

Effective flood risk management relies on accurate forecasting, yet the "black box" nature of stateof-the-art Deep Learning models creates a barrier to trust and accountability in high-stakes public safety decisions. While existing Explainable AI (XAI) methods offer local attributions, they fail to provide the verifiable, operationally meaningful causal narratives required by disaster response authorities. To address this societal challenge, we propose Context-Aware Concept Distillation (CACD), a framework developed in collaboration with domain experts to distill opaque LSTMs into interpretable, hydrology-aware surrogate models. We introduce an unsupervised pipeline to discover a "Hydrological Language" and a Residual Hypernetwork that dynamically modulates these concepts based on static basin characteristics. Evaluated on 5,203 basins globally, our model achieves high fidelity (Median NSE 0.70), significantly outperforming black-box baselines (e.g., Multi Layer Perceptrons) on unseen future data. By demonstrating that human-interpretable concepts are sufficient to reconstruct flood dynamics, this work balances AI accuracy with the transparency required for responsible environmental decision-making.

Problem

• Trustworthy AI
• Flood Prediction
• Explainable AI
• Black-box Models
• Causal Interpretability

Innovation

• Context-Aware Concept Distillation
• Hydrological Language
• Residual Hypernetwork
• Explainable AI
• Interpretable Surrogate Models