Pasture biomass and livestock carrying capacity estimation
Satellite-derived vegetation indices and SAR backscatter can estimate above-ground herbaceous biomass across rangelands and managed pastures, converting canopy signals into carrying capacity metrics that inform stocking decisions at paddock scale.
Sensors
- Sentinel-2 MSI: 10 m resolution in visible and near-infrared bands, 20 m in red-edge bands (B5, B6, B7) and shortwave infrared. Five-day revisit at the equator under cloud-free conditions. The red-edge bands are particularly sensitive to chlorophyll concentration and canopy structure at intermediate biomass levels, reducing the saturation problem that affects NDVI in dense swards.
- Planet SuperDove: 3 m resolution with daily revisit across most latitudes. Eight spectral bands including red-edge. The high spatial resolution resolves individual paddocks of a few hectares, which Sentinel-2 can blur across fence lines. Commercial tasking adds cost; cloud cover remains a hard constraint.
- MODIS Terra/Aqua: 250 m to 500 m resolution with daily global coverage. Too coarse for paddock-level management but well suited to regional biomass monitoring across large rangelands. The long archive (2000 to present) supports phenological baselines and multi-year anomaly detection. MODIS-derived products such as MOD13 vegetation indices are operationally used in rangeland early warning systems.
- Sentinel-1 SAR C-band: 10 m resolution, six-day repeat in interferometric wide-swath mode, imaging through cloud and at night. C-band backscatter responds to canopy volume and moisture content. It is useful for detecting green biomass flushes after rain when optical sensors are cloud-obscured, and for separating herbaceous from woody canopy structure when combined with optical data. Sensitivity saturates at relatively low biomass levels in dense swards.
What a canopy reflects, and what that tells you about dry matter
Green plant tissue absorbs red light for photosynthesis and reflects strongly in the near-infrared because of internal leaf scattering. The ratio of these two signals, formalised in indices such as NDVI, correlates with leaf area index and, by extension, with above-ground biomass. The relationship is empirical: it holds reasonably well for herbaceous swards at low to moderate biomass, but it saturates. Once a canopy closes and LAI exceeds roughly three to four, additional dry matter accumulation produces diminishing returns in the reflectance signal. For lush improved pastures in high-rainfall zones, this saturation is a genuine analytical ceiling.
Red-edge bands, available on Sentinel-2 at 20 m and on SuperDove at 3 m, partially relieve the saturation problem. Chlorophyll absorption extends into the red-edge (roughly 700 to 740 nm), so indices built on these bands, such as the Red-Edge Chlorophyll Index or the Normalised Difference Red-Edge index, remain sensitive at higher biomass levels than conventional NDVI. Published comparisons on temperate grasslands suggest red-edge indices can improve biomass estimation accuracy by ten to twenty percentage points relative to NDVI alone, though the exact gain depends heavily on grass species composition and soil background.
The woody encroachment problem
In savannas and semi-arid rangelands, bush encroachment is one of the most persistent confounders in optical biomass estimation. Shrubs and trees reflect in the near-infrared just as grasses do. A pixel with fifty percent woody cover and sparse grass underneath can return the same NDVI as a pixel with dense, palatable herbaceous sward. For a livestock manager, these two situations represent entirely different carrying capacities.
Separating the two signals requires either very high spatial resolution (where individual tree crowns can be delineated and masked), multi-temporal analysis that exploits the different phenological rhythms of woody and herbaceous layers, or SAR data. C-band Sentinel-1 backscatter is sensitive to canopy volume and structure in ways that differ between grasses and woody shrubs, particularly when combined with cross-polarisation ratios (VH/VV). Fusion of Sentinel-1 and Sentinel-2 time series is a published approach for isolating the herbaceous fraction in mixed savannas, though it requires local calibration and does not eliminate ambiguity entirely. Any biomass product delivered for a savanna context should state explicitly how woody cover has been handled.
From biomass to carrying capacity: the translation layer
Carrying capacity is not a satellite product. It is a management calculation that requires satellite-derived biomass as one input alongside agronomic rules about utilisation rates, animal metabolic requirements, and the distinction between total biomass and the digestible fraction that livestock can actually consume. A common convention in rangeland management is that only fifty percent of available herbage should be grazed before resting a paddock, to protect root reserves and soil cover. That utilisation factor, combined with an assumed dry matter intake per animal unit per day, converts a biomass estimate in kilograms per hectare into a stocking rate in animal units per hectare for a given grazing period.
The digestible fraction cannot be read directly from a satellite. Spectral indices measure canopy structure and chlorophyll, not fibre content or metabolisable energy. However, phenological stage is a reasonable proxy: early-season green growth is generally more digestible than mature or senescent material. Time-series analysis that tracks the green-up and dry-down cycle can flag when a paddock has passed peak nutritive value, even if the absolute digestibility figure requires ground-truth data to calibrate.
Spatial resolution and the paddock-scale problem
A typical managed paddock in commercial ranching might cover fifty to several hundred hectares. At that scale, Sentinel-2 at 10 to 20 m is generally adequate for within-paddock biomass mapping, provided the paddock is large enough that boundary pixels do not dominate. MODIS at 250 m is too coarse to resolve individual paddocks reliably; it is a regional tool, not a farm management tool.
Planet SuperDove at 3 m resolves even small paddocks cleanly and can detect localised overgrazing patches that Sentinel-2 would average away. The trade-off is cost and data volume. For a property covering tens of thousands of hectares, a daily SuperDove feed generates substantial storage and processing requirements. The practical approach for most operations is to use Sentinel-2 for routine monitoring and commission high-resolution tasking only when a specific paddock or management question warrants it.
Honest limits of the method
Cloud cover is the most obvious constraint. In tropical and sub-tropical rangelands, the wet season, when biomass is growing fastest and management decisions are most consequential, is also when optical imagery is most frequently obscured. Dense time-series compositing (combining all cloud-free observations over a rolling window) reduces gaps but introduces temporal blurring. SAR provides a partial workaround but requires its own calibration effort.
Soil background effects are significant in sparse, open rangelands where bare soil is visible between grass tufts. Soil-adjusted indices such as SAVI or MSAVI partially correct for this, but they require an assumed or measured soil brightness parameter that varies across a property. Topographic shading affects both optical and SAR signals on hilly terrain. And all satellite-derived biomass estimates carry a calibration uncertainty that typically ranges from fifteen to thirty percent relative error in published studies on semi-arid grasslands, depending on species composition and site conditions. That is a useful planning signal; it is not a precision weighing instrument.
Putting it into operational practice
Satellize runs biomass and carrying capacity analytics on Sentinel-2 and Sentinel-1 open data, with optional Planet tasking added on client licence. The analytical pipeline follows published semi-mechanistic approaches: red-edge index time series calibrated against available ground-truth biomass cuts, a phenology model to separate green and senescent fractions, and a woody-cover mask derived from multi-temporal SAR and optical fusion. Outputs are delivered as paddock-level GIS layers with biomass estimates in kilograms of dry matter per hectare and indicative stocking rate ranges, updated on each cloud-free overpass.
The Tonga crop-estimation programme demonstrated the value of dense time-series analytics on small, fragmented agricultural parcels in a Pacific island context. Rangeland biomass estimation presents a different set of challenges, particularly around species heterogeneity and woody encroachment, but the underlying approach to building calibrated, operationally reliable products from open satellite archives is the same. For a new rangeland client, the first practical step is a calibration campaign: a set of field biomass cuts timed to coincide with satellite overpasses, spread across the range of vegetation conditions on the property. Without that, the uncertainty bounds remain wide.
Typical figures
| Spatial resolution (operational) | 10–20 m (Sentinel-2), 3 m (Planet SuperDove), 10 m (Sentinel-1 SAR), 250–500 m (MODIS regional) |
| Revisit frequency | 5 days (Sentinel-2, equator, cloud-free); daily (MODIS, Planet); 6 days (Sentinel-1 IW mode) |
| Key spectral bands | Red (665 nm), NIR (842 nm), red-edge (705, 740, 783 nm) on Sentinel-2 MSI; C-band 5.4 GHz VV/VH on Sentinel-1 |
| Biomass estimation range | Reliable optical signal roughly 50–3,000 kg DM/ha; saturation above ~3,000 kg DM/ha in dense swards; SAR saturates earlier in C-band |
| Typical calibration uncertainty | 15–30% relative error in semi-arid grasslands (published range); improves with site-specific ground-truth biomass cuts |
| Cloud impact | Optical imagery unusable under cloud; SAR unaffected. Wet-season gaps in tropical rangelands may span weeks for optical sensors |
| Archive depth | Sentinel-2 from 2015; MODIS from 2000; Landsat (compatible indices) from 1984; Sentinel-1 from 2014 |
| Minimum paddock size for reliable mapping | ~5 ha at Sentinel-2 10 m; ~0.5 ha at Planet 3 m (boundary pixel effects become significant below these thresholds) |
| Delivery formats | GeoTIFF raster (biomass kg DM/ha), vector paddock summary (CSV or GeoJSON), time-series chart per paddock |
Analytics Satellize can run
| Above-ground herbaceous biomass map | Red-edge vegetation index regression (NDRE, Red-Edge Chlorophyll Index) calibrated against field biomass cuts; soil-adjusted for sparse canopies | GeoTIFF layer, kg dry matter per hectare, per cloud-free overpass |
| Paddock-level carrying capacity estimate | Biomass estimate combined with configurable utilisation rate and animal unit dry matter intake assumptions | Vector GIS layer with stocking rate range (AU/ha) and recommended grazing days per paddock |
| Seasonal biomass time series and phenology curve | Dense Sentinel-2 time-series compositing with harmonic or asymmetric Gaussian phenology fitting to separate green-up, peak and senescence phases | Per-paddock time-series chart and CSV export; flags when peak biomass and peak nutritive value window are reached |
| Woody cover fraction mask | Multi-temporal Sentinel-1 SAR and Sentinel-2 optical fusion; cross-polarisation ratio analysis to separate woody and herbaceous canopy contributions | Binary or fractional woody cover raster used as input mask for herbaceous biomass product |
| Biomass anomaly alert | Current biomass estimate compared against multi-year MODIS or Sentinel-2 baseline for the same phenological stage; z-score threshold triggering | Email or API alert when a paddock falls below a configurable percentage of its historical median biomass |
| Degradation risk index | Multi-year trend analysis of bare soil fraction (using MODIS or Sentinel-2 time series) combined with rainfall normalisation to separate climate signal from land management signal | Annual GIS layer classifying paddocks by degradation trend severity; suitable for reporting to land stewardship programmes |
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.