Nitrous oxide flux mapping above fertilised agricultural fields
Soil nitrification and denitrification after synthetic nitrogen application produce N2O pulses detectable in TROPOMI column retrievals. Regional flux anomalies can be correlated with fertiliser calendars derived from Sentinel-2 crop-stage mapping, though field-level attribution remains beyond current instrument precision.
Sensors
- Sentinel-5P TROPOMI: Primary N2O column sensor. Nadir pixel footprint of 5.5 × 3.5 km (after 2019 upgrade from 7 × 3.5 km). Daily global coverage. Single-sounding N2O precision approximately 0.6 ppb (1-sigma); multi-day averaging over large districts reduces noise to detectable flux anomaly levels. Retrieval exploits shortwave-infrared absorption near 4.5 µm.
- IASI (MetOp-A/B/C): Thermal infrared sounder covering the 7.8 µm N2O absorption band. Footprint roughly 12 km diameter at nadir. Two overpasses per day per satellite (three satellites in orbit). Sensitivity peaks in the free troposphere rather than the boundary layer, so IASI complements TROPOMI rather than replacing it for surface-flux work.
- GOSAT-2: JAXA sun-synchronous sounder with a 10.5 km diameter footprint and roughly three-day revisit at mid-latitudes. Retrieves N2O in the 3.9 µm and 7.8 µm bands with high spectral resolution. Sparse spatial sampling limits its use to validation and cross-calibration of TROPOMI retrievals rather than continuous district monitoring.
- Sentinel-2 MSI: 10 m resolution optical imagery in 13 spectral bands, five-day revisit at the equator (2–3 days at mid-latitudes with both satellites). Used here not for gas detection but for crop-stage classification: green-up timing, bare-soil exposure and NDVI trajectories define the fertiliser application window that constrains flux attribution.
Why the first two weeks after spreading are the signal
Synthetic nitrogen fertiliser applied to soil is converted by nitrifying and denitrifying bacteria into N2O on a timescale of days. Peak emission flux typically occurs within 48 to 96 hours of application under warm, moist soil conditions, then decays over one to three weeks. This temporal structure is the analytical opportunity: if you know when and where fertiliser was spread, you can look for a corresponding column anomaly in the days that follow.
TROPOMI measures the total atmospheric column of N2O in molecules per square centimetre, expressed as a mixing ratio in parts per billion. The global background is approximately 332 ppb and rising at roughly 1 ppb per year. The agricultural pulse you are trying to detect sits on top of that background. At district scales covering tens of thousands of hectares, published flux inversion studies have demonstrated that TROPOMI multi-day composites can resolve anomalies of a few tenths of a ppb above regional background, which corresponds to meaningful emission events. A single overpass cannot.
Crop-stage mapping as a fertiliser calendar proxy
No satellite directly observes a farmer spreading urea. What Sentinel-2 does observe is the crop canopy state before and after the event. Bare or sparsely vegetated soil with low NDVI in early spring, followed by rapid green-up within one to two weeks, is a reliable phenological marker for the pre-emergence nitrogen application window in cereal systems. Combining this with land-cover classification and regional agronomic calendars produces a probabilistic fertiliser application map: a spatial and temporal prior that tells the flux inversion where to look.
The practical workflow is sequential. Sentinel-2 scenes are classified weekly to track NDVI trajectories across agricultural parcels. Districts entering the expected application window are flagged. TROPOMI daily L2 N2O retrievals over those districts are then composited across five to fifteen days to average down instrument noise. A simple flux anomaly is computed by subtracting a baseline column derived from the same districts in non-fertilisation periods or from upwind reference pixels. The result is not a field-level measurement. It is a district-level signal that can be compared against reported application rates.
What the physics allows and what it does not
Honesty matters here. TROPOMI's 0.6 ppb single-sounding precision means that individual overpasses over a single field produce noise that swamps the signal entirely. The instrument was designed for global budget studies, not farm-scale enforcement. Published work by the TROPOMI science team and independent groups has demonstrated regional N2O anomaly detection over large agricultural districts in Europe and China, but the spatial unit of attribution is typically 50 to 200 km across, not a parish or a farm.
Cloud cover is a hard blocker. TROPOMI requires clear-sky or thin-cloud conditions for valid retrievals; heavy cloud in a wet spring can eliminate useful observations for days at a time, precisely when soil moisture conditions are most favourable for N2O emission. IASI's thermal infrared retrieval is less sensitive to cloud but peaks in sensitivity above 3 km altitude, which means boundary-layer enhancements near the surface are partially diluted before the instrument sees them. Neither sensor can currently attribute emissions to a specific farmer or even a specific field. What they can do is flag districts where column N2O is elevated relative to the expected background during the application season, providing a basis for targeted ground inspection.
Flux inversion: turning column measurements into surface rates
A column measurement in ppb is not the same as a surface emission rate in kilograms of N2O per hectare per day. Converting one to the other requires an atmospheric transport model. The standard approach is a Bayesian flux inversion: a prior emission estimate (from fertiliser application statistics and emission factor tables such as those in IPCC guidelines) is updated using the observed column anomaly and a transport model that describes how surface emissions mix upward and advect downwind.
The inversion is sensitive to meteorological inputs. Wind speed, boundary-layer height and atmospheric stability all affect how a surface flux maps onto a column anomaly. Errors in reanalysis wind fields propagate directly into flux uncertainty. Published inversion studies typically report posterior flux uncertainties of 20 to 50 percent at the district scale, which is useful for national inventory verification but too imprecise for individual compliance decisions. The method is best understood as an independent cross-check on reported fertiliser use, not a replacement for ground-based measurement.
Where this fits in national inventory verification
Agricultural N2O is the single largest source category in the national greenhouse gas inventories of several major grain-producing countries, yet it is estimated almost entirely from activity data and emission factors rather than direct measurement. TROPOMI-based flux mapping offers a top-down cross-check on those bottom-up estimates. Discrepancies between reported application volumes and satellite-inferred flux anomalies can flag districts where emission factors may be systematically underestimated, or where application reporting is incomplete.
This is the context in which Satellize runs district-level N2O anomaly analysis: as a verification layer for clients who need independent evidence on agricultural emission reporting, not as a substitute for field measurement campaigns. The approach is analogous in spirit to the crop-estimation work Satellize runs for the Kingdom of Tonga, where remote sensing provides an independent data stream that sits alongside, rather than replaces, ground-collected information. Regulators and commodity buyers are the natural audience: both need to know whether reported nitrogen use in a supply shed is plausible.
Reading the output honestly
A district-level N2O anomaly map delivered as a seasonal composite has genuine value for policy and procurement, provided the consumer understands what it represents. It shows where atmospheric N2O was elevated relative to background during the fertilisation window. It does not show which specific fields were over-applied, whether the excess came from synthetic or organic sources, or whether mitigation measures such as nitrification inhibitors were used.
Combining the anomaly map with Sentinel-2 crop-type classification, soil-type data and precipitation records from ERA5 or similar reanalysis products narrows the attribution considerably. Districts with sandy, well-drained soils show different emission dynamics from heavy clay soils; wet springs amplify denitrification. Building these covariates into the inversion prior improves posterior precision. The honest ceiling, given current satellite capability, is a district-scale flux estimate with roughly 20 to 50 percent uncertainty, updated seasonally, with spatial resolution no finer than approximately 25 km for reliable anomaly detection.
Typical figures
| TROPOMI N2O pixel footprint | 5.5 × 3.5 km (post-August 2019 upgrade) |
| TROPOMI N2O single-sounding precision | ~0.6 ppb (1-sigma); multi-day district composites reduce effective noise substantially |
| TROPOMI revisit | Daily global coverage; cloud-free observations variable by season and latitude |
| IASI footprint and revisit | ~12 km diameter at nadir; twice daily per satellite, three MetOp satellites operational |
| Sentinel-2 MSI resolution (crop staging) | 10 m (visible/NIR bands); 2–5 day revisit at mid-latitudes with both satellites |
| Spectral bands used | TROPOMI/GOSAT-2: SWIR ~4.5 µm; IASI: TIR ~7.8 µm; Sentinel-2: VNIR/SWIR for NDVI |
| Minimum detectable district anomaly | Approximately 0.2–0.5 ppb above regional background over districts >50 km across, using 10-day composites |
| Flux inversion uncertainty | Typically 20–50% at district scale in published studies |
| TROPOMI archive depth | May 2018 to present (Sentinel-5P launch April 2017, science data from May 2018) |
| Deliverable spatial unit | Agricultural district polygons (typically 25–200 km across); not field-level |
Analytics Satellize can run
| Seasonal N2O column anomaly map | TROPOMI L2 N2O retrievals composited over 10–30 day windows; background subtraction using upwind reference pixels or pre-season baseline | GeoTIFF and GIS polygon layer showing anomaly magnitude (ppb above baseline) per district, delivered post-season or at monthly intervals |
| Fertiliser application window classifier | Sentinel-2 NDVI time-series analysis and bare-soil detection to identify pre-emergence application periods by crop type and district | Weekly GIS layer flagging districts in active application window, with crop-type attribution |
| District-scale N2O flux estimate | Bayesian flux inversion using TROPOMI column anomaly, ERA5 meteorological reanalysis and IPCC emission factor priors | Tabular report of posterior flux estimates (kg N2O-N ha⁻¹ season⁻¹) with uncertainty ranges per district polygon |
| Bottom-up vs top-down inventory comparison | Comparison of satellite-inferred flux against reported fertiliser application volumes from national statistics, using IPCC Tier 1 emission factors as the conversion bridge | Discrepancy report identifying districts where satellite signal and reported activity data diverge by more than one posterior standard deviation |
| Multi-year trend analysis | Year-on-year comparison of seasonal N2O anomalies using TROPOMI archive from 2018 onward, controlling for meteorological variability | Time-series chart and summary report showing whether district-level emissions are tracking changes in reported nitrogen application rates |
| Cloud-gap-filled composite | Temporal interpolation and IASI cross-calibration to fill TROPOMI cloud-blocked days, following published multi-sensor blending approaches | Continuous seasonal composite layer with data-gap flags and confidence scores per pixel |
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.