Methods & Transparency
Our open ecological intelligence pipeline: from raw satellite bytes to actionable probability layers.
01. Data Ingestion
We pull directly from NASA's FIRMS API and ingest VIIRS (S-NPP and NOAA-20) and MODIS (Terra and Aqua) data.
Thermal Detections
375m spatial resolution for VIIRS, updated every 6-12 hours depending on satellite pass.
Weather Models
GFS and ERA5 data for wind, humidity, and temperature.
02. Denoising & Calibration
Raw thermal hotspots are prone to false positives from gas flares, volcanoes, and urban heat islands. We apply geospatial masking and temporal filtering.
Raw Hotspots
Temporal Filtering
Active Fire Layer
03. Forecast Model (Probabilistic)
Our spread model uses a modified Huygens principle. We run 1,000 simulations per fire detection, varying wind direction and speed within forecast uncertainty bounds.
- Fuel Load: Integrated from ESA WorldCover and CGLS-LC100.
- Topography: SRTM digital elevation model to calculate slope-aided spread.