Use machine learning to fill the gaps between satellite passes, producing a continuous daily time series of vegetation index values for your field.
Satellites revisit the same location every 2–7 days, and cloud cover can further reduce usable images. Daily vegetation estimates use ML-based gap filling to interpolate between actual satellite observations, generating an estimated index value for every day in your date range.
The algorithm automatically extends 3 months prior to your start date to gather enough historical observations for accurate interpolation.
Daily estimates are enabled by setting dailyEstimates: true in the processVegetationIndices() mutation:
Note: The dailyEstimates parameter is also available on processPlantHealth() and processWaterStress() mutations.
Daily estimates have specific constraints to ensure the ML model produces reliable results:
| Constraint | Limit | Reason |
|---|---|---|
| Maximum temporal gap | 128 days | If no satellite observation exists within 128 days, the model cannot reliably interpolate. |
| Minimum analysis period | 28 days | The model needs sufficient data points to produce meaningful daily estimates. |
| Maximum polygon area | 50 hectares | Computational cost scales with polygon size. Larger polygons should be split. |
Daily vegetation estimates are a premium service that consumes processing units. Each account includes 50 units. One processing unit generates approximately 250 new daily estimate images for a 4-hectare polygon.
Updating previously generated estimates (e.g., extending the date range) costs 20% of the original processing cost.
Cost awareness: Monitor your processing unit balance with retrieveProcessingUnits(). Overage beyond 50 included units is billed at €0.60 per unit.