Skip to main content

Earth Observation

Earth observation imagery is the primary input for SpaceCoMP workflows. The simulation supports injecting both synthetic and real sensor data into the pipeline.

Supported​

Thermal IR​

Brightness temperature images in the mid-wave infrared (MWIR, ~3.9 μm) and long-wave infrared (LWIR, ~11 μm) bands. Thermal IR does not depend on sunlight — it measures heat radiated by the surface, so it works at night and through thin cloud. A fire at ~600 K stands out clearly against a ~300 K background. Thermal data is the primary input for wildfire detection workflows.

The full pipeline is implemented end-to-end:

  • Synthetic data — the eosim tool generates brightness temperature rasters with a diurnal background model, spatial variation, sensor noise (NEdT), and configurable fire events (location, onset, peak temperature, spread rate).
  • NOS3 simulator — the thermal camera component serves frames over a virtual SPI bus, detects AOI entry using the satellite's position from 42, and loads the appropriate frame for each pass.
  • cFS app — the wildfire app captures frames via SPI, thresholds brightness temperature, clusters hot pixels, converts to geographic coordinates, and sends alert packets via SRSPP to the ground station.
  • Real data — MODIS and VIIRS thermal bands provide global coverage at 375 m–1 km resolution with multiple daily revisits.

Not Yet Supported​

SAR​

Synthetic Aperture Radar transmits microwave pulses and records the reflected signal. SAR works through clouds, at night, and in all weather. SAR data comes in two forms:

  • Complex (SLC) — preserves both amplitude and phase. Phase differences between repeat passes reveal sub-millimeter surface displacement (InSAR), used for tailings dam displacement monitoring.
  • Backscatter (intensity) — measures how strongly the surface reflects radar. Calm water appears dark (specular reflection away from the sensor), making oil spills and flood extent visible as dark patches. Used for oil spill, flood, and sea ice detection.

Real data is available from Sentinel-1 (free, global C-band SAR, 5×20 m resolution, 6-day revisit). No eosim generator, NOS3 simulator, or cFS processing app exists yet.

Multispectral​

Images captured in multiple wavelength bands (visible, near-infrared, shortwave infrared). The key derived product is NDVI (Normalized Difference Vegetation Index), which measures vegetation health by comparing red and near-infrared reflectance. Healthy vegetation reflects strongly in NIR and absorbs red; a drop in NDVI over time indicates deforestation or crop loss.

Real data is available from Landsat (30 m, 16-day revisit) and Sentinel-2 (10 m, 5-day revisit). No eosim generator, NOS3 simulator, or cFS processing app exists yet.

Optical​

Visible-light imagery. The Arducam (OV5640) camera has a NOS3 simulator and Rust bindings, but it is not integrated into any Earth observation workflow — no eosim generator or cFS processing app uses it for EO purposes. Optical images depend on daylight and clear skies.