Biochar project feedstock land-use and additionality verification

Satellite time-series can test whether biochar project feedstock genuinely originates from agricultural residues and whether claimed avoided-burning additionality is supported by observed fire history, though pyrolysis yield and biochar stability remain field-verified quantities.

Sensors

Suomi-NPP / NOAA-20 VIIRS: VIIRS I-band active-fire detection at 375 m pixel resolution with twice-daily global overpass. The VIIRS Nightfire product additionally resolves sub-pixel combustion temperatures, helping distinguish low-intensity residue burns from larger vegetation fires. Near-real-time fire radiative power data available within hours of acquisition.
Sentinel-2 MSI: 10 m optical imagery in visible and near-infrared bands, 20 m in shortwave infrared (SWIR). Five-day revisit at the equator (two-to-three days at mid-latitudes with both satellites). Used for land-cover classification, burned-area mapping via the Normalised Burn Ratio, and cropland boundary delineation. Cloud cover is the principal constraint; tropical and monsoonal regions may lose weeks of usable imagery during growing seasons.
Landsat 8 / 9 OLI-2: 30 m multispectral imagery with a 16-day single-satellite revisit, reduced to roughly eight days with both platforms combined. SWIR bands 6 and 7 are sensitive to char and soil exposure after burning. The 40-plus-year Landsat archive enables land-use history analysis extending well before most carbon registries began requiring baseline periods.
MODIS Terra / Aqua: The MCD64A1 burned-area product provides monthly global burned-area mapping at 500 m resolution, with a continuous record from 2000. Coarse resolution limits utility for small field-scale burns but the archive depth makes it the standard tool for establishing multi-year fire-history baselines. Daily active-fire products (MOD14/MYD14) complement VIIRS for temporal cross-checking.

What a biochar project is actually claiming

A biochar carbon project converts agricultural or forestry residues into charcoal through oxygen-limited pyrolysis, then applies or buries the char to sequester carbon in a stable form. The carbon credit rests on two linked assertions: first, that the feedstock is genuinely waste biomass that would otherwise have decomposed or been burned openly; second, that the project's intervention caused a reduction in field burning that would not have happened anyway. Both assertions have a geographic and temporal signature that satellite data can interrogate.

The additionality claim is where most scrutiny concentrates. If a project asserts it is preventing rice-straw burning in a specific district, there should be a credible prior record of burning in that district, and a measurable decline coincident with project activity. If fire history is sparse before the project start date, or if burning continues at similar rates in the project area after crediting begins, the additionality claim weakens. Satellite fire products make this testable.

Reading fire history from the archive

MODIS MCD64A1 burned-area data runs from 2000, giving auditors more than two decades of annual fire maps at 500 m resolution. For a biochar project claiming avoided burning of crop residues, the standard approach is to extract burned-area statistics within the declared project boundary and a matched control zone for each year of the pre-project baseline period. Persistent annual burning in the same field parcels, timed to post-harvest windows, is consistent with residue-burning practice. Its absence suggests the feedstock was never burned at scale to begin with, which undermines additionality.

VIIRS active-fire detections at 375 m add temporal precision. Because VIIRS passes twice daily, it can resolve whether fire events cluster in the days immediately after harvest, which is the pattern expected for intentional residue burning rather than accidental or land-clearing fires. Fire Radiative Power values from VIIRS give a rough indication of combustion intensity; low-intensity, short-duration events spread across many small parcels are more consistent with straw burning than with forest clearance. The distinction matters because land-clearing fires would signal a very different feedstock origin problem.

Land-cover classification and feedstock origin

The second verification question is whether the land supplying feedstock is genuinely agricultural. A project claiming rice-straw or maize-stover feedstock should be drawing from parcels that satellite imagery confirms are under continuous arable cultivation, not from areas that were recently forest, woodland or intact grassland converted for the purpose. Sentinel-2's 10 m resolution is sufficient to map individual field boundaries in most agricultural landscapes, and its SWIR bands distinguish bare soil, green canopy, dry stubble and char with reasonable reliability.

Multi-temporal classification across several growing seasons strengthens the argument. A parcel showing repeated cycles of green crop, harvest-stage spectral response and post-harvest bare or burned soil is almost certainly in active arable use. A parcel showing forest or dense shrub cover two years before the project start date and arable use after is a red flag. Landsat's archive depth is valuable here: it allows classification going back to the early 1980s if needed, well beyond what any carbon registry baseline period requires.

One honest limit: optical classification cannot reliably distinguish crop species at the sub-field level without ground truth or high-resolution commercial imagery. Knowing a parcel is arable is straightforward; confirming it grew the specific crop whose residue the project claims requires ancillary data, typically national agricultural statistics or field survey records.

What satellite data cannot tell you

Satellite verification of biochar projects hits a hard ceiling at the pyrolysis stage. Once feedstock enters a kiln or retort, no current spaceborne sensor can measure conversion efficiency, pyrolysis temperature, or the resulting biochar's stability class. These are the quantities that determine how much carbon is actually sequestered and for how long. They are governed by the H:C ratio and O:C ratio of the product, which require laboratory analysis. No spectral index substitutes for that.

Similarly, satellite data cannot verify the quantity of feedstock collected. It can confirm that a field was harvested and that residues were not burned, but the mass of straw removed and converted is a field-measured figure. Auditors should treat satellite evidence as a necessary constraint on geographic and temporal plausibility, not as a substitute for mass-balance records or third-party site inspection. The two methods are complementary, not interchangeable.

Putting the evidence package together

A credible satellite evidence package for a biochar project combines three layers. The first is a land-cover time series covering at least five years before the project start date, classifying each parcel in the declared feedstock catchment as arable, forest, grassland or other. The second is an annual fire-history record for the same parcels, drawn from MODIS burned-area and VIIRS active-fire products, establishing that burning was a real and recurring practice. The third is a post-project fire-detection record showing a statistically significant decline in burning within the project zone relative to a control area where no intervention occurred.

Satellize builds these packages as GIS-deliverable time-series products, drawing on open Sentinel, Landsat, VIIRS and MODIS archives. The Tonga crop-estimation programme demonstrated the value of combining multi-sensor optical and thermal data for agricultural land characterisation in island and small-parcel contexts, a methodology that transfers directly to feedstock-origin audits. For carbon registries or independent auditors who need evidence that meets emerging MRV standards, the starting point is agreeing on the project boundary geometry and the baseline period dates. Everything else follows from the archive.

Typical figures

Optical spatial resolution (land cover)	10 m (Sentinel-2 visible/NIR), 20 m (Sentinel-2 SWIR), 30 m (Landsat 8/9 OLI-2)
Active-fire detection resolution	375 m (VIIRS I-band), 1 km (MODIS MOD14/MYD14)
Burned-area mapping resolution	500 m monthly (MODIS MCD64A1); 20 m event-level (Sentinel-2 NBR)
Revisit frequency	Twice daily (VIIRS/MODIS active fire); 2-3 days (Sentinel-2, dual satellite); ~8 days (Landsat 8+9 combined)
Active-fire latency	VIIRS near-real-time products typically available within 3-6 hours of overpass via FIRMS
Archive depth	MODIS burned area from 2000; Landsat from 1972 (TM from 1982); Sentinel-2 from 2015
Key spectral bands for char and residue detection	SWIR 1 (~1.6 µm) and SWIR 2 (~2.2 µm) on Sentinel-2 and Landsat OLI-2; NIR for NBR calculation
Minimum detectable burned area	Sub-hectare events detectable with Sentinel-2 NBR; MODIS and VIIRS less reliable below ~50-100 ha aggregate
Coverage	Global; all sensors provide open-access data with no geographic restriction
Delivery format	GeoTIFF time-series stacks, vector parcel-level attribution tables, PDF audit summary report

Analytics Satellize can run

Feedstock parcel land-cover classification	Multi-temporal supervised classification (Random Forest or similar) on Sentinel-2 and Landsat spectral-temporal features; validated against publicly available crop-mask products such as ESA WorldCover	Annual land-cover GIS layer per parcel for baseline period, flagging any conversion from non-arable cover
Pre-project fire-history baseline	MODIS MCD64A1 burned-area extraction and VIIRS active-fire point aggregation within project boundary and matched control zone, 2000 to project start date	Annual burned-area statistics table and map series; statistical test of burning frequency against control
Post-project burning trend analysis	Year-on-year comparison of VIIRS active-fire density and Sentinel-2 NBR burned-area within project zone versus control, with Mann-Kendall trend test	Additionality credibility report with trend charts and spatial fire maps by project year
Harvest-timing fire attribution	Cross-correlation of VIIRS fire event dates with crop-calendar windows derived from NDVI phenology on Sentinel-2; distinguishes post-harvest residue burns from land-clearing or other fire types by timing and spatial pattern	Fire-type attribution table per event cluster; flagged anomalies for auditor review
Land-use change alert for feedstock catchment	Bi-annual Sentinel-2 change detection within declared catchment boundary; flags new clearance or land-use transition events	GIS alert layer with change polygons, acquisition dates and spectral evidence; updated each six-month crediting period
Registry-grade evidence package	Compilation of above outputs into a structured time-series archive with methodology notes, sensor metadata and uncertainty statements consistent with Verra VCS or Gold Standard documentation requirements	PDF report plus zipped GIS data package suitable for submission to a third-party auditor

Sources

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.