United StatesPublic sector2024 – ongoingU.S. Department of Agriculture
Random-forest crop classification for the Cropland Data Layer
USDA used random forests to produce its 10-meter crop map
Problem
NASS needed crop-specific maps to supplement acreage estimates. Satellite classifications required ground-reference data to distinguish crops.
Approach
For the 2024 Cropland Data Layer, NASS replaced its previous decision-tree classifier with a random forest in Google Earth Engine. It combined satellite imagery with agricultural training and validation data. The completed product supported internal acreage estimation and public mapping.
Results
Takeaway
Internal acreage estimates combined the mapped classifications with survey data through regression. Counting classified pixels alone did not reproduce those estimates.
Categories
- Size
- Large enterprise
- Industry
- Public sector
- Business function
- Production and quality
- Innovation type
- Process innovation
- AI technology
- Predictive analytics and ML
- Adoption
- Off-the-shelf tool or platform
- Goal
- Quality and risk
- Evidence
- Company's own claim
Sources
Source summaries
NASS documented the completed 10-meter product and its random-forest classifier. It described the training data and the layer's role in acreage estimation.
Full details are in the original sources.