This geometry-only dataset provides agricultural field boundary polygons across the USA for the calendar year 2025, derived from Sentinel-2 satellite imagery and machine learning. It includes high-resolution vector footprints at roughly 5x5 meter granularity, with per-feature unique identifiers, bounding-box identifiers, and area in hectares. Polygons report delineation accuracy up to 0.96 IoU. The layer excludes crop type, yield, ownership, and farm-level metadata. Clean, topologically consistent geometries enable mapping and geographic visualization, spatial queries and proximity analysis, single-year change detection against other references, automated field onboarding, and field-level exposure or risk assessments. Delivered as a geometry-only vector layer optimized for spatial joins and indexing.

Key attributes include:

AgTech developers, insurers, lenders, and public sector planners can use precise 2024 field geometries to accelerate onboarding, spatial integration, and field-level risk or exposure analysis without requiring ground surveys.