CrisiSense-RAG: Crisis sensing multimodal retrieval-augmented generation for rapid disaster impact assessment

Timely and spatially resolved disaster impact assessment is essential for effective emergency response. However, automated methods typically struggle with temporal asynchrony. Real-time human reports capture peak hazard conditions, while high-resolution satellite imagery is frequently acquired after peak conditions. This often reflects flood recession rather than the maximum extent. Naive fusion of these misaligned streams can yield dangerous underestimates when post-event imagery overrides documented peak flooding. We present CrisiSense-RAG, a multimodal retrieval-augmented generation framework that reframes impact assessment as evidence synthesis over heterogeneous data sources without disaster-specific fine-tuning. The system employs hybrid dense-sparse retrieval for text sources and CLIP-based retrieval for aerial imagery. A split-pipeline architecture feeds into asynchronous fusion logic that prioritizes real-time social evidence for peak flood extent while treating imagery as persistent evidence of structural damage. Evaluated on Hurricane Harvey across 110 ZIP-code queries with multimodal coverage, the full multimodal configuration achieves a flood extent MAE of 8.86% to 19.64% and damage severity MAE of 10.14% to 14.47% across four model backends. Full multimodal RAG reduces flood extent MAE by up to 8.62 percentage points relative to text-only inference. Prompt-level alignment proves critical for quantitative validity: without explicit metric grounding, models interpret damage severity as qualitative intensity rather than the statistical average fraction of building value destroyed. These results demonstrate a practical and deployable approach to rapid resilience intelligence under real-world data constraints.