Service Disclosures Summary
OPAL processes satellite analytics dynamically as ESA's Sentinel network updates. Sentinel-2 spectral indices refresh every 5 days under clear sky conditions, while Sentinel-1 radar coordinates refresh every 6–12 days. Cloud cover represents a physical barrier to optical sensors; OPAL handles this transparently using Scene Classification Layer (SCL) pixel masking to reject cloudy records, maintaining an average platform API uptime target of 99.5%.
1. Satellite Overpass Frequencies & Latency
Because OPAL relies on orbiting space constellations, data freshness depends on satellite overpass schedules and ESA ground station processing delays. The expected update intervals are:
| Sensor Type | Satellite Platform | Revisit Frequency | Typical Latency (Acquisition to App) |
|---|---|---|---|
| Optical (MSI) | Sentinel-2A / 2B / 2C | Every 5 Days | 4 to 12 Hours |
| Radar (SAR) | Sentinel-1A / 1C | Every 6 to 12 Days | 6 to 24 Hours |
| Weather Forecasts | Open-Meteo Land Grid | Hourly Updates | Real-Time Ingestion |
2. Cloud Cover and Weather Limitations
Optical satellite passes (Sentinel-2) are blocked by cloud cover, fog, and heavy sandstorms. OPAL manages this with the following quality protocols:
- Automatic Cloud Filtering: During analysis, the system evaluates the Copernicus Scene Classification Layer (SCL). Any pixels classified as cloud (medium/high probability), cirrus, or shadow (`CAST_SHADOW`) are rejected.
- Honest-State Reporting: If a field is 100% cloudy during an overpass, OPAL flags the date as "Insufficient valid pixels" and retains your last known-good clear acquisition, rather than serving estimated or skewed indices.
- Radar Resilience: Sentinel-1 radar backscatter (used to calculate soil moisture) penetrates clouds and dust, ensuring VSM updates continue even during extended periods of overcast weather.
3. Platform Processing Performance & Uptime Target
OPAL aims to maintain a serverless API uptime of 99.5%. Uptime excludes scheduled maintenance windows or upstream outages on Copernicus Data Space Ecosystem (CDSE) servers.
When you enqueue a new field analysis (which pulls a 3-year historical archive), the Cloud Run background worker is designed to complete all backscatter conversions, NDVI calculations, and agrometeorological projections in under 3 minutes for a standard 10-hectare field.