Oil and gas flaring and venting verification for methane offset project baselines
Methane offset projects need a credible, independent baseline before any credit can be issued. VIIRS Nightfire flare detection, Sentinel-5P TROPOMI methane columns and published point-source inversion methods provide that audit trail, with honest limits stated.
Sensors
- VIIRS Nightfire (Suomi-NPP / NOAA-20): 375 m pixel resolution; detects sub-pixel thermal anomalies in shortwave infrared bands M7, M8, M10, M12 and M13. Provides nightly global coverage with latency of roughly 6 hours from overpass. Derives flare temperature, area and radiant heat, from which combustion volumes can be estimated via the Elvidge et al. regression method published in the peer-reviewed literature.
- Sentinel-5P TROPOMI: 7 x 5.5 km nadir pixel (resampled from original 3.5 x 5.5 km after August 2019 upgrade); daily global coverage. Measures column-averaged dry-air mole fractions of methane (XCH4) in the shortwave infrared at 2305-2385 nm. Minimum detectable enhancement from a point source is approximately 1 tonne per hour under favourable, low-wind conditions; sensitivity degrades under cloud cover, high aerosol loading or wind speeds above roughly 10 m/s.
- GHGSat-D / C1 / C2: Approximately 25 m spatial resolution; targets individual facilities on tasking. Quantifies methane flux at the facility level using the integrated mass enhancement method. Detection floor is approximately 100 kg per hour per source under clear-sky conditions. Commercially tasked; not freely available, but adds the point-source precision that TROPOMI cannot achieve alone.
- ECMWF ERA5 reanalysis wind fields: 31 km horizontal resolution, hourly temporal resolution. Used as the dispersion model input for Gaussian plume or more sophisticated Lagrangian inversion schemes that back-calculate source emission rates from observed atmospheric columns. ERA5 is free to access via the Copernicus Climate Data Store.
What a flare actually reveals, and what it hides
A flare stack burning gas is a thermal anomaly visible to VIIRS at night from 828 km altitude. The sensor's shortwave infrared channels detect sub-pixel combustion events, and the Colorado School of Mines VIIRS Nightfire algorithm converts radiance in those channels into a flare temperature and a radiant heat output. From radiant heat, a regression model estimates combusted gas volume. That chain of inference is publicly documented and has been applied to global flaring inventories since at least 2016.
The limit is combustion efficiency. A well-operated flare burning at 98 percent efficiency produces almost no methane emission; the carbon leaves as CO2. Venting, by contrast, releases raw methane with no thermal signature at all. VIIRS sees the flare but is blind to the vent pipe. This is not a minor caveat: in many mature oil fields, venting contributes a substantial fraction of total methane loss, and operators have an incentive to classify unburned releases as 'flaring' in self-reported data. An honest baseline methodology must combine thermal detection with atmospheric column measurement.
TROPOMI columns: what the atmosphere remembers
Sentinel-5P TROPOMI measures the total column of methane above a 7 x 5.5 km footprint once per day. When a facility is venting or operating a malfunctioning flare, methane accumulates downwind in a detectable plume. Pair that column observation with ERA5 wind speed and direction at the time of overpass, run a Gaussian plume inversion, and you recover an estimated source emission rate. Studies using this approach on known industrial sources have shown agreement with independent measurements within roughly 20 to 40 percent under favourable conditions.
Favourable conditions matter enormously. Cloud cover blocks the shortwave infrared retrieval entirely; TROPOMI's global daily coverage does not mean daily cloud-free coverage over any specific facility. In persistently cloudy regions, months can pass without a usable overpass. Wind speeds above about 10 m/s dilute the plume below the detection threshold of approximately 1 tonne per hour. Buyers of offset credits should understand that a three-month cloud gap in the baseline record is not evidence of zero emissions; it is a gap. A credible methodology quantifies that uncertainty rather than papering over it.
Building the baseline: what the public record requires
A defensible offset baseline needs at minimum two years of pre-project satellite data, processed consistently. For flaring, that means a VIIRS Nightfire time series extracted for each identified flare site within the project boundary, with radiant heat converted to estimated combustion volume using the published Elvidge coefficients. Gaps at night due to cloud are identifiable from the VIIRS quality flags and should be interpolated conservatively, not optimistically.
For venting and fugitive methane, TROPOMI XCH4 anomalies are composited over the baseline period. Individual overpasses with cloud fraction above 0.3 are typically excluded from retrievals. The remaining clear-sky observations are inverted against ERA5 winds to produce a probability distribution of source emission rates, not a single point estimate. That distribution is what an independent verifier should see: a range with stated confidence, not a suspiciously precise number. Where TROPOMI's detection floor is too high to resolve individual well pads, GHGSat tasking can fill the gap, though at commercial cost and with the access constraints that implies.
Post-project monitoring: detecting backsliding before credits are retired
The same sensor stack that establishes the baseline also serves as the ongoing monitoring layer. VIIRS Nightfire runs nightly; a flare that was supposed to be eliminated reappearing at a known site is detectable within 24 hours of the overpass, subject to cloud. TROPOMI provides a weekly to monthly check on whether atmospheric methane over the project area has returned toward baseline levels.
Backsliding detection is where the combination of sensors earns its keep. An operator might cap a flare stack while quietly increasing venting elsewhere on the same lease. VIIRS would show the flare gone; TROPOMI would show the column methane unchanged or rising. That divergence is the signal. It is not conclusive on its own, but it is exactly the kind of anomaly that triggers a GHGSat tasking request or a ground inspection. The satellite record creates accountability without requiring a permanent inspector on site.
One practical complication: oil and gas infrastructure changes. Wells are drilled, fields are partially abandoned, compressors are added. The project boundary defined at baseline must be checked against current operational maps, ideally using commercial synthetic aperture radar or optical imagery to confirm which infrastructure is active. Crediting an emission reduction at a well that was already being decommissioned is a known failure mode in self-reported programmes.
Where the method is honest about its limits
TROPOMI cannot resolve sources separated by less than roughly 7 km. In dense production basins, multiple facilities sit within a single pixel, and the inversion cannot attribute the column enhancement to any one of them without additional information. VIIRS at 375 m can distinguish individual large flares but misses small or intermittent ones. GHGSat's 25 m resolution resolves individual facilities but is commercially tasked and therefore not available as a continuous archive.
ERA5 wind fields introduce their own uncertainty. Reanalysis winds at 31 km resolution smooth out local effects: terrain channelling, sea breezes, nocturnal jets. In complex terrain, inversion errors of 50 percent or more are plausible. Published studies using aircraft or ground-based lidar as reference have documented this range honestly. A baseline methodology that claims better than 20 percent accuracy without independent validation is overstating what the physics and the data allow.
Satellize processes VIIRS Nightfire and TROPOMI data on open-access archives and can add commercial GHGSat tasking under client licence. The workflow is the same one used in the Tonga crop-estimation programme in the sense that it rests on open constellations augmented by commercial layers, though the sensors and the inversion methods are entirely different. The relevant next step for a project developer is to define the facility list and boundary, then run a retrospective baseline extraction before committing to a methodology.
Typical figures
| VIIRS Nightfire spatial resolution | 375 m per pixel; sub-pixel flare detection via radiance anomaly |
| VIIRS revisit and latency | Nightly global coverage; approximately 6 hours post-overpass to processed product |
| TROPOMI pixel size | 7 x 5.5 km (since August 2019 reprocessing) |
| TROPOMI revisit | Daily global; cloud-free revisit site-dependent, often 3 to 10 days in cloudy regions |
| TROPOMI methane detection floor | Approximately 1 tonne CH4 per hour per point source under low cloud, wind below ~10 m/s |
| GHGSat spatial resolution | Approximately 25 m; minimum detectable flux approximately 100 kg CH4 per hour |
| ERA5 wind field resolution | 31 km horizontal, 1-hour temporal; free via Copernicus Climate Data Store |
| VIIRS Nightfire archive depth | Suomi-NPP from 2012; NOAA-20 from 2018 |
| TROPOMI archive depth | May 2018 to present; reprocessed v02 available from ESA |
| Inversion accuracy (published range) | 20 to 50 percent depending on wind field quality, cloud fraction and source isolation |
Analytics Satellize can run
| Baseline flare radiance time series | VIIRS Nightfire algorithm (Elvidge et al.); radiant heat to combustion volume regression | Site-level CSV and GeoTIFF stack covering user-defined pre-project period, with quality flags and gap log |
| TROPOMI XCH4 anomaly composite | Clear-sky pixel selection (cloud fraction < 0.3), background subtraction using upwind reference pixels, temporal compositing | Monthly raster composites of methane column enhancement over project boundary, with uncertainty band per composite |
| Point-source emission rate estimate | Gaussian plume inversion against ERA5 hourly wind fields; posterior distribution of source strength | Per-facility emission rate probability distribution (5th, 50th, 95th percentile) in tonnes CH4 per hour, delivered as structured report |
| Backsliding alert feed | Threshold exceedance on VIIRS radiance and TROPOMI column anomaly relative to post-intervention baseline; divergence flagging | Near-real-time alert (within 48 hours of overpass) when flare reappears or column methane exceeds agreed threshold; JSON feed or email notification |
| Facility activity change detection | Multi-temporal SAR or optical imagery cross-referenced against VIIRS flare presence to confirm infrastructure status | Annotated facility map updated quarterly, flagging newly active or decommissioned sites within project boundary |
| Registry-ready evidence package | Aggregation of VIIRS, TROPOMI and inversion outputs into a documented provenance chain with sensor metadata, processing versions and uncertainty statements | PDF and machine-readable annex formatted for submission to carbon registry MRV requirements |
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.