Model Overview
OYPT-CA uses a proprietary statistical model to predict almond yield from satellite remote sensing data and orchard characteristics. The model uses previous year's satellite observations to predict the current year's harvest — meaning predictions are available before the growing season begins.
The modeling framework accounts for orchard-level variability, learning each orchard's baseline productivity while identifying how satellite-derived features drive year-to-year yield changes.
Input Features
The model combines satellite-derived vegetation and moisture indices with orchard maturity information. Each variety uses a tailored set of features selected through extensive model comparison and validation.
Model Validation
The model was trained and validated on California Central Valley almond orchards using multi-year field data. Simple, interpretable models were selected over complex alternatives based on out-of-sample prediction accuracy.
Confidence Intervals
The prediction interval reflects model uncertainty. For orchards not in the training data, the interval is wider to account for unknown orchard-specific factors.
Data Sources
- Sentinel-2 (Copernicus) — satellite imagery
- Field validation data — California Central Valley orchards
Limitations
- Trained on Central Valley almonds only.
- Cloud cover may cause missing satellite data in some years.
- Prediction accuracy varies by variety.
- Use the confidence interval alongside your own field knowledge.
Model version: March 2026. Coefficients updated annually.