Cropl

Cropl — Field-level (not county-level) crop yield predictions with per-field confidence intervals and SHAP driver attributions, built on a four-stage pipeline: ingestion of Sentinel-2 and MODIS imagery, weather telemetry, soil composition and historical yield records; derived signal layer of NDVI growth curves, precipitation accumulation, heat-stress indices, soil-moisture proxies and seasonal anomaly detection; ensemble models continuously recalibrated across regions, seasons and crop types against field-level USDA-validated yield ground truth.. Micro-scale operation, unusually clean dataset…

Cropl offers (Alternative, Geospatial, Supply Chain) — Est. millions of acres under analysis — the page claims scalability to millions of acres — translating to roughly 10^5-10^6 field parcels x Sentinel-2 revisit (~5 days) over a ~20-week season x derived indices, plus multi-season yield ground-truth history. Compact tabular layer sits on top of large imagery that is itself open (Copernicus/NASA), so the licensable artifact is the derived field-level series, not the raw imagery..

Early-season crop production estimation for commodity traders and hedgers; agricultural lender collateral and covenant monitoring; crop insurance underwriting and yield-loss indemnification; training satellite-to-yield regression and multi-modal ag foundation models; county-vs-field basis studies for USDA statistical downscaling; input-response and precision-ag recommendation models; food-security and supply forecasting.

The data is with 10 years of history.

Coverage spans US, Japan; Application Software; alternative, geospatial, supply_chain; equities, commodities.