Fire fuel load and fire-risk mapping
Satellite-derived vegetation indices and spaceborne lidar can map canopy fuel loads and live fuel moisture proxies at landscape scale, giving fire managers a quantitative basis for suppression planning before ignition occurs.
Sensors
- Sentinel-2 MSI: 10 m resolution in visible and near-infrared bands, 5-day revisit at the equator with both satellites. Bands 8A and 11 (shortwave infrared) are the primary inputs for NBR and moisture-sensitive indices. Cloud cover is the main operational constraint; persistent smoke can also degrade retrievals.
- Landsat 8/9 OLI: 30 m resolution, 16-day revisit per satellite (8-day combined). The longer archive (Landsat 5 back to 1984) enables multi-decadal fuel accumulation trends. OLI bands 5 and 7 support NBR calculation consistent with Sentinel-2 outputs.
- GEDI (Global Ecosystem Dynamics Investigation): Spaceborne lidar operating from the ISS, producing canopy height and canopy bulk density profiles at approximately 25 m footprint diameter. Coverage is limited to latitudes between roughly 51.6° N and 51.6° S. Vertical structure data is the closest satellite analogue to field-measured canopy bulk density used in fire behaviour models such as FlamMap.
- MODIS Terra/Aqua: 250 m to 500 m resolution, daily revisit. Too coarse for fine fuel mapping at operational scales, but MODIS-derived NDVI and EVI time series spanning more than two decades provide the baseline against which current-season greenness anomalies are measured. Useful for regional drought-stress signals.
Why pre-fire mapping is harder than it looks
Fire behaviour models such as Rothermel and its descendants require four fuel inputs: loading (mass per unit area), moisture content, surface-area-to-volume ratio, and bulk density. Satellites can contribute meaningfully to two of those. They struggle with the other two.
Live fuel moisture content (LFMC) is the most operationally useful satellite-derived variable. Shortwave infrared reflectance, particularly around 1.6 µm and 2.2 µm, is sensitive to liquid water in leaf tissue. The Normalised Burn Ratio (NBR = (NIR - SWIR) / (NIR + SWIR)) and indices such as the Normalised Difference Water Index exploit this physics. Published studies using Sentinel-2 and Landsat have shown reasonable correlations between SWIR-based indices and field-measured LFMC in shrublands and grasslands, though the relationship is species-specific and degrades under heavy canopy shadow.
What a floating roof gives away about canopy fuel
Canopy bulk density, the mass of available canopy fuel per unit volume of crown space, is the key input for crown fire initiation and spread. GEDI provides the most direct satellite estimate of this quantity. Its full-waveform lidar resolves vertical foliage distribution within the canopy, from which bulk density can be modelled. At a 25 m footprint and non-contiguous sampling (GEDI does not produce wall-to-wall coverage in a single pass), the practical approach is to train a spatial model using GEDI shots as reference points and Sentinel-2 or Landsat spectral and textural predictors as wall-to-wall covariates.
The result is a continuous canopy bulk density surface at 10 to 30 m resolution. It is an estimate, not a measurement. Uncertainty is higher in structurally complex mixed forests than in relatively uniform plantations, and GEDI data quality degrades in persistent cloud regions because the lidar requires clear sky.
The surface fuel problem: what satellites cannot see
Surface fuels, dead litter, duff, fallen logs, and cured grass, drive the initial rate of spread in most wildland fires. They are also the hardest fuel class to map from orbit. Optical sensors observe the top of the canopy or, in open areas, the top of the grass layer. They cannot measure litter depth, decomposition state, or the moisture content of dead fuel.
SAR (synthetic aperture radar) adds some information in open shrublands, where backscatter is sensitive to soil moisture and coarse woody debris, but it does not reliably resolve fine surface fuel loads beneath closed canopy. The honest position is that satellite-derived surface fuel estimates are proxies at best. Grass curing percentage in open grasslands is a partial exception: optical sensors can track the transition from green to cured grass reasonably well using NDVI and EVI time series, and several national fire agencies use this operationally. For litter and duff beneath forest canopy, field sampling or airborne lidar remains the higher-quality source.
Building a fire-risk layer that a planner can actually use
A practical fuel-risk product combines several inputs. A current-season LFMC anomaly map, derived from Sentinel-2 SWIR indices relative to a multi-year MODIS baseline, identifies where vegetation is drier than normal for the time of year. A GEDI-informed canopy bulk density layer identifies where crown fire potential is structurally high. A slope and aspect layer from a digital elevation model adds topographic fire behaviour context. Together these feed a composite risk index that can be updated at each Sentinel-2 overpass, roughly every five days in cloud-free conditions.
Revisit frequency is the practical ceiling on how responsive the product can be. A drought stress signal that develops over days will be captured; a single hot, dry wind event that desiccates surface fuels overnight will not. Fire managers should treat satellite-derived risk layers as strategic planning tools, updated weekly to monthly, rather than as tactical real-time guidance.
Satellize runs exactly this kind of multi-source compositing on open constellation data. The methodology is the same class of index fusion used in the Tonga crop-estimation programme, adapted from vegetation condition monitoring to fuel condition monitoring.
Calibration, validation, and the limits of any index
No vegetation index is a universal LFMC proxy. The NBR-LFMC relationship published for California chaparral does not transfer unchanged to Australian mallee or Mediterranean maquis. Any operational system should be validated against local field measurements before being used in suppression planning decisions. The MODIS archive, which extends back to 2000, is useful for establishing what 'normal' looks like for a given region and season, but the coarse spatial resolution means that local fuel heterogeneity is smoothed away.
Cloud cover is the other persistent constraint. In tropical and subtropical regions, the wet season that grows the fuel load is also the season with the highest cloud frequency. Sentinel-2's 5-day revisit in cloud-free conditions can become a 30-day or longer effective revisit in persistently overcast areas. Compositing multiple observations reduces this problem but introduces temporal blurring.
From map to decision: what fire agencies actually need
A fuel map sitting in a GIS is not a decision. The useful outputs are specific: a ranked list of high-risk landscape units by fire management zone, a time series showing how risk has evolved through the current dry season, and a flag when a zone crosses a defined LFMC threshold that triggers a prescribed burn review.
Prescribed burn planning is the most direct application. Identifying where canopy bulk density and surface dryness are both elevated lets managers prioritise where fuel reduction burns will have the greatest effect on potential fire behaviour. The satellite layer does not replace on-the-ground fuel assessment for individual burn prescriptions, but it makes the landscape-scale prioritisation tractable without requiring thousands of field plots.
Typical figures
| Spatial resolution (optical) | 10 m (Sentinel-2 NIR/SWIR); 30 m (Landsat 8/9 OLI) |
| Spatial resolution (lidar) | ~25 m footprint diameter (GEDI); non-contiguous sampling, not wall-to-wall |
| Revisit frequency | 5 days (Sentinel-2, both satellites combined); 8 days (Landsat 8+9 combined); daily (MODIS) |
| GEDI latitude coverage | Approximately 51.6° N to 51.6° S; no polar coverage |
| Key spectral bands for fuel moisture | SWIR at ~1.6 µm (Sentinel-2 Band 11, Landsat Band 6) and ~2.2 µm (Sentinel-2 Band 12, Landsat Band 7) |
| Effective cloud-free revisit (tropics) | Can extend to 30+ days during persistent overcast seasons; compositing partially mitigates this |
| MODIS archive depth | Terra from 2000; Aqua from 2002; suitable for multi-decadal anomaly baselines |
| Landsat archive depth | Landsat 5 from 1984; continuous record enables multi-decadal fuel accumulation analysis |
| Minimum detectable fuel signal | LFMC anomalies of roughly 10-15 percentage points detectable in open shrublands; performance degrades under closed canopy |
| Delivery formats | GeoTIFF raster layers; zone-level summary tables (CSV); optional WMS/WMTS tile service |
Analytics Satellize can run
| Live fuel moisture content anomaly map | SWIR-based index (NBR, NDWI) compared against multi-year MODIS baseline; per-pixel z-score relative to historical seasonal mean | GeoTIFF updated at each cloud-free Sentinel-2 overpass; zone-level risk summary table |
| Canopy bulk density surface | GEDI full-waveform lidar shots used as training reference; Sentinel-2 spectral and textural predictors extrapolated wall-to-wall via random forest regression | 30 m GeoTIFF canopy bulk density layer; uncertainty band raster |
| Grass curing percentage map | NDVI and EVI time series from Sentinel-2 and MODIS tracking green-to-cured transition in open grassland pixels | Weekly GeoTIFF; threshold-breach alert when curing exceeds defined percentage in a fire management zone |
| Composite fire-risk index | Weighted combination of LFMC anomaly, canopy bulk density, topographic slope/aspect, and drought stress indicator (VHI from MODIS) | Ranked risk layer by fire management zone; PDF priority report for suppression planning review |
| Seasonal fuel accumulation trend | Multi-year Landsat NDVI and NBR time series decomposed into trend and seasonal components; identifies zones with above-average biomass accumulation since last disturbance | Annual GeoTIFF trend layer; tabular summary by management unit |
| Prescribed burn prioritisation layer | Overlay of high canopy bulk density, elevated dryness anomaly, and time-since-fire from Landsat burn scar archive; ranked output by combined score | GIS polygon layer with priority scores; exported to standard fire agency GIS formats |
Who does the work
We can get this done for you. Satellize runs its own analyst desk and a strong science team. You do not buy a data feed and work out what it means; our people source the imagery, run the analysis described on this page, and hand you the answer with its confidence limits stated. Discuss this requirement.