A new framework called ADAGE has been developed to evaluate how well explanations from Geospatial Artificial Intelligence (GeoAI) models align with established domain knowledge in satellite-based flood mapping. This framework uses the Channel-Group SHAP method to assess the contribution of different spectral bands to flood predictions. Experiments on two flood mapping tasks showed that ADAGE can quantitatively measure this alignment and help domain experts identify explanations that do not match existing remote sensing principles. AI
Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →
IMPACT Introduces a framework to improve trust and applicability of GeoAI models in critical Earth observation tasks.
RANK_REASON This is a research paper introducing a new framework for evaluating GeoAI model explanations.