EarthRecord: From Pixels to Land Change with Multimodal Agents over Four Decades of Satellite Imagery

EarthRecord is a benchmark built on the raw record of 56 sites featuring four decades of Landsat, MODIS, VIIRS, and Sentinel-2 data. It poses 238 land-change questions across ten analysis tasks. Evaluations of twelve multimodal agents reveal that leading agents complete most analyses end to end but falter when an answer requires mapping from imagery.