Methane detection above produced-water evaporation ponds in oil fields
Produced-water evaporation ponds degas dissolved methane continuously yet rarely appear in facility emission inventories. GHGSat high-resolution SWIR retrievals and TROPOMI column enhancements can identify active ponds, though co-located tank and compressor sources complicate attribution.
Sensors
- GHGSat-C series (Claire, Luca, Oliver and successors): Shortwave infrared Fabry-Pérot spectrometer centred near 1.65 µm. Nominal ground pixel roughly 25 x 25 m, scene footprint approximately 12 x 12 km. Minimum detectable enhancement in a single overpass is typically cited by GHGSat as around 100 kg/h for a point source under good atmospheric conditions, though pond-distributed sources are harder than point sources at equivalent total flux. Tasked on demand; not a continuous global imager.
- Sentinel-5P TROPOMI: Methane column (XCH4) at 7 x 5.5 km pixel (reprocessed to 5.5 x 3.5 km from August 2019 onward), daily global coverage. Detects basin-scale enhancements of tens of ppb above background. Too coarse to isolate individual ponds, but useful for identifying anomalous field-level signals that justify targeted high-resolution tasking.
- Sentinel-2 MSI: 10 m visible and near-infrared bands for pond delineation: surface area, turbidity proxy, and seasonal fill-level change. No methane sensitivity, but pond geometry and surface-area time series feed the flux model denominator. 5-day revisit at mid-latitudes with both satellites.
- Landsat 8/9 OLI and TIRS: 30 m multispectral for pond boundary mapping and 100 m thermal for surface temperature, which is a secondary input to evaporation and degassing rate estimates. 16-day revisit per satellite; combined 8-day with both. Free archive back to 2013 (Landsat 8) supports historical pond inventory construction.
Why ponds are missing from most emission inventories
Produced water, the brine co-extracted with oil and gas, carries dissolved methane at concentrations that depend on reservoir pressure, temperature, and gas-oil ratio. When that water reaches a surface pond at atmospheric pressure, dissolved gas degasses passively and continuously. The process is not combustion, carries no flare stack, and generates no visible plume. Regulatory frameworks in most jurisdictions require operators to report combustion emissions and vented gas from identifiable equipment, but pond degassing sits in a poorly defined category. Studies published in peer-reviewed remote sensing literature have found that produced-water ponds in basins such as the Permian and Uinta contribute measurably to basin-wide methane budgets, yet the ponds themselves rarely appear as discrete entries in facility-level inventories.
The practical consequence is a systematic undercount. A regulator looking at a facility's reported figures sees the compressors, the storage tanks, and the flares. The pond, which may cover several hectares and receive thousands of barrels of produced water per day, is either absent or lumped into a residual fugitive category with no spatial or temporal resolution.
What a floating roof gives away, and what a pond does not
Point sources such as storage tank vents or compressor seals concentrate methane in a relatively small atmospheric volume, making them tractable targets for GHGSat's narrow-footprint retrievals. A pond is an area source: the same total flux spread across hundreds to thousands of square metres produces a lower column enhancement per pixel than an equivalent point source would. GHGSat's published detection floor of roughly 100 kg/h applies to compact point sources. For distributed pond sources, the effective detection threshold is higher unless the pond is large and the degassing rate substantial. This is not a reason to dismiss the sensor; it is a reason to be precise about what a non-detection means.
TROPOMI's 5.5 x 3.5 km pixel resolves nothing at the pond level, but an anomalous XCH4 enhancement persistent across multiple overpasses over a field sector can flag a basin for targeted GHGSat tasking. The two sensors are complementary rather than redundant, operating on timescales and spatial scales that bracket the problem from opposite ends.
The attribution problem: ponds inside a crowd of sources
Oil field facilities are dense. A produced-water pond typically sits within a few hundred metres of storage tanks, separator vessels, compressor stations, and active wellheads. Each of those is a potential methane source. When a GHGSat overpass detects an enhancement plume downwind of a facility cluster, attributing the fraction that originated from the pond versus the tanks versus the compressors requires more than a single image.
Wind-integration flux estimation, sometimes called the cross-sectional or mass-balance method, uses the retrieved methane column along a transect perpendicular to the mean wind vector, multiplied by wind speed, to estimate total facility flux. Separating pond contribution from the rest demands either repeated overpasses under different wind directions (so the plume geometry changes relative to fixed source locations), or simultaneous high-resolution thermal and optical data to identify which surface features are actively emitting. Neither approach is clean. Honest practice is to report a facility-level flux with a stated uncertainty, note that pond contribution cannot be isolated without additional constraints, and flag the pond as a candidate source requiring ground-truth or repeated tasking.
Meteorological inputs matter enormously. Flux estimates derived from satellite column data carry uncertainty that scales with wind speed error. A 20% error in the boundary-layer wind speed propagates directly into a 20% error in the derived flux. Analysts should use reanalysis products (ERA5 is the standard) and, where available, local anemometer data, and they should report the resulting flux as a range rather than a point estimate.
Pond delineation as the necessary first step
Before any methane retrieval is meaningful, the analyst needs a current, accurate map of pond locations, surface areas, and fill levels. Sentinel-2 MSI at 10 m resolution is the practical tool: water bodies are readily separable from bare soil and vegetation using the normalised difference water index or simple band ratios of green and shortwave infrared. Landsat 8/9 extends the archive to 2013 and adds thermal context.
Surface area matters because degassing flux scales with the air-water interface. A pond at 30% capacity emits differently from the same pond at capacity, and seasonal variation in produced-water disposal rates can be substantial. A time series of pond area from Sentinel-2, combined with operator disposal records where available, provides the denominator for any per-unit-area flux estimate. This is unglamorous preprocessing work, but skipping it means the methane retrieval has no geometric anchor.
Flux estimation: what the wind-integration method can and cannot deliver
The cross-sectional mass-balance approach integrates the methane column enhancement across a downwind transect and multiplies by effective wind speed and boundary-layer height. Applied to GHGSat data, it can yield facility-level flux estimates in the range of tens to hundreds of kg/h, with published validation studies suggesting uncertainties of roughly 30 to 50% for individual overpasses under moderate wind conditions. That uncertainty is large by the standards of ground-based measurement, but it is far more informative than no measurement at all, particularly for sources that have never appeared in any inventory.
For pond-specific attribution within a facility, the method requires either multiple overpasses under varied wind directions or a priori knowledge that the pond is the dominant source. In practice, regulators and operators are often satisfied with a facility-level bound that places the pond inside a credible emission range. That is enough to trigger an inspection or a ground-based measurement campaign, which is the appropriate next step rather than satellite-only closure.
Satellize applies these retrieval and flux-estimation methods as part of its analytics work on open and commercial constellations. The workflow follows the same principles used in published basin-scale methane studies, adapted to the specific geometry of oil field pond clusters.
Practical limits a buyer should understand before commissioning work
Cloud cover blocks SWIR retrievals entirely. GHGSat overpasses are tasked, not continuous, so a cloudy day is a lost opportunity with no automatic reschedule unless the tasking plan accounts for it. In persistently cloudy regions or seasons, detection probability per month can be low. TROPOMI has daily coverage but its coarse pixel means it will miss all but the largest or most clustered pond signals.
Aerosol loading, common in dusty oil field environments, degrades SWIR column retrievals by introducing scattering uncertainty. High-aerosol scenes are either flagged and discarded or carry larger retrieval uncertainty; the analyst should report which applies. Finally, the absence of a detectable enhancement in a single GHGSat overpass is not evidence that a pond is not emitting. It may mean the pond flux was below the detection threshold on that day, or that wind conditions dispersed the plume outside the footprint. A programme of repeated tasking over a season is more defensible than a single-pass non-detection used to clear a facility.
Typical figures
| GHGSat-C spatial resolution | Approximately 25 x 25 m per pixel, 12 x 12 km scene footprint |
| TROPOMI XCH4 pixel size | 5.5 x 3.5 km (post-August 2019 reprocessing) |
| Sentinel-2 MSI pond delineation resolution | 10 m (visible/NIR bands), 20 m (SWIR bands) |
| TROPOMI revisit | Daily global coverage, approximately 14 orbits per day |
| GHGSat revisit | On-demand tasking; no fixed revisit. Multiple overpasses per week possible with constellation scheduling |
| Spectral bands used | GHGSat: ~1.65 µm SWIR (CH4 absorption); TROPOMI: 2305-2385 nm (SWIR CH4); Sentinel-2: 490-2190 nm (pond delineation); Landsat TIRS: 10.6-12.5 µm (surface temperature) |
| Minimum detectable flux (GHGSat, point source) | ~100 kg/h under good atmospheric conditions; higher effective threshold for distributed area sources such as ponds |
| Flux estimation uncertainty (cross-sectional method) | Typically 30-50% for single overpasses under moderate wind; improves with repeated overpasses and local wind data |
| Archive depth for pond delineation | Sentinel-2: 2015 to present; Landsat 8: 2013 to present; TROPOMI: late 2017 to present |
| Latency (TROPOMI operational product) | Near-real-time product typically available within 3 hours of overpass; offline reprocessed product within days |
Analytics Satellize can run
| Pond inventory layer | Normalised difference water index and SWIR band ratio applied to Sentinel-2 and Landsat time series | GIS polygon layer of pond extents, surface areas, and seasonal fill-level time series; updated quarterly or on request |
| TROPOMI XCH4 anomaly screening | Background-subtracted column enhancement analysis over defined facility bounding boxes, multi-overpass compositing to reduce noise | Monthly anomaly report flagging field sectors with persistent XCH4 elevation above regional background, ranked by enhancement magnitude |
| GHGSat facility-level flux estimate | Cross-sectional mass-balance retrieval using GHGSat column data, ERA5 boundary-layer wind, and scene-specific aerosol quality flags | Per-overpass flux estimate with stated uncertainty range (kg/h), facility attribution note, and detection/non-detection confidence classification |
| Source attribution assessment | Multi-overpass wind-direction analysis comparing plume geometry against mapped source locations (ponds, tanks, compressors) | Written attribution assessment stating which sources can and cannot be separated given available data, with recommended follow-on tasking schedule |
| Seasonal degassing trend | Correlation of pond surface-area time series (Sentinel-2) with TROPOMI XCH4 anomaly time series over the same facility sector | Time-series chart and summary table linking pond fill level to column enhancement, suitable for regulatory submission |
| Tasking optimisation schedule | Prevailing wind climatology analysis (ERA5 reanalysis) to identify overpass windows with highest expected plume-in-footprint probability | Monthly tasking recommendation specifying preferred overpass times and wind-direction acceptance criteria for GHGSat scheduling |
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.