- The paper presents a multi-sensor retrieval framework (HyGAS) integrating Tanager-1 with EnMAP and PRISMA for methane plume detection and flux estimation.
- It introduces a column-wise matched filter (CW-CMF) approach to mitigate cross-track radiometric artifacts, ensuring robust, artifact-aware retrieval.
- Operational results at a Buenos Aires landfill demonstrate reproducible emission mapping and highlight the importance of sensor-specific calibration strategies.
Multi-Sensor Methane Mapping in a Unified Framework: Integration of Tanager-1 and Comparison with EnMAP and PRISMA
Introduction
The paper addresses the critical challenge of differentiating and quantifying methane emissions from large point sources by leveraging advances in high-resolution spaceborne imaging spectroscopy. The methodological backbone is a multi-sensor framework, HyGAS, built for rigorous, consistent processing of hyperspectral satellites such as Tanager-1, EnMAP, and PRISMA. The approach emphasizes matched-filter-based plume detection and flux estimation, with a technical focus on quantifying and mitigating instrument-specific radiometric artifacts—especially the cross-track column dependencies ubiquitous in pushbroom sensors.
Methodological Framework
The core retrieval algorithm is a Clutter Matched Filter (CMF) operating on Level-1 radiances. In this treatment, each pixel spectrum is modeled as the sum of spatially variable background and a linear additive CH₄ absorption component. The retrieval is anchored in a physically motivated, linearized enhancement in concentration-path-length units (ppm·m) under a near-nadir, moderate enhancement regime. Sensor-specific CH₄ spectral templates are resampled using each instrument’s spectral response function, ensuring comparability across sensors.
A central innovation is the systematic examination of covariance estimation strategies tailored for pushbroom sensor radiometry. Three CMF variants are tested: scene-wide, cluster-tuned (CTMF), and column-wise (CW-CMF). The latter explicitly models column-dependent non-uniformity intrinsic to pushbroom systems, absorbing structured background artifacts (striping) that otherwise manifest as false positives and impact plume quantification.
Uncertainty is rigorously propagated via the same per-pixel spectral statistics as the retrieval, accounting for both radiance noise and background variability. Plume segmentation utilizes a scale-aware framework, and subsequent integrated mass enhancement (IME) and flux (Q) estimates propagate both spectral and wind uncertainties.
Cross-Sensor Radiometric and Artifact Diagnostics
A series of homogeneous, high-reflectance calibration scenes are used to benchmark the effective signal-to-noise ratio (SNR) and striping amplitude for Tanager-1, EnMAP, and PRISMA in the SWIR methane windows. By radiance-normalizing all SNR calculations to a common reference scene, the analysis isolates sensor-intrinsic noise performance from surface reflectance effects.
The comparative results show:
- EnMAP achieves the highest radiance-normalized SNR in the 2.3 µm methane window, with PRISMA lower and Tanager-1 demonstrating further improvement, affirming the expected design hierarchy for modern sensors.
- Despite favorable SNR, Tanager-1 exhibits the strongest cross-track (column-dependent) radiometric variability, as revealed by ratio-based striping metrics. EnMAP features minimal striping, and PRISMA is intermediate.
This artifact profile has direct retrieval implications: even with superior SNR, unmodeled column structure in Tanager-1 results in structured false positives unless explicitly accounted for in the background model—hence the operational need for CW-CMF. These findings validate the necessity of tailored artifact mitigation for each sensor, as naïve or scene-wide models are insufficient.
Case Study: Buenos Aires Landfill Super-Emitter
Operational testing is conducted on the Complejo Ambiental Norte III landfill, using non-simultaneous Tanager-1 and EnMAP acquisitions. Application of the CW-CMF variant produces stable enhancement maps and robust plume segmentation over a heterogeneous urban background, where traditional approaches suffer from artifact-induced instability.
Reported quantitative results are:
- Tanager-1: IME = 30,590.94 ± 101.29 kg, Q = 31.53 ± 8.21 t/hr
- EnMAP: IME = 98,858.52 ± 481.67 kg, Q = 74.02 ± 16.29 t/hr
These numbers are not interpreted as direct inter-sensor bias estimates due to the non-simultaneity of acquisitions and lack of ground truth. Instead, the focus is on the reproducibility and uncertainty stabilization of flux estimates across challenging backgrounds, with demonstrated suppression of structured false positives via column-wise statistical treatment. The results support the assertion that artifact-aware retrieval is mandatory for operational monitoring with modern pushbroom imaging spectrometers.
Implications and Future Directions
The study provides a rigorous template for harmonized, physically consistent methane monitoring across the next generation of hyperspectral satellites, overcoming key challenges imposed by pushbroom artifacts. Practically, it points the way toward routine, operational use of multi-sensor fusion for global monitoring of facility-scale methane emissions, leveraging open-source, modular pipelines.
From a theoretical standpoint, the findings highlight the limits of increasing radiometric SNR without concomitant treatment of structured instrument variability, reinforcing the need for adaptive retrieval frameworks that account for both noise and systematic background artifacts at all stages of the processing chain.
The open-sourcing of the HyGAS framework is scheduled upon journal publication, which will further support reproducibility, cross-mission validation, and accelerate the adoption of these methods by the atmospheric remote sensing and GHG monitoring communities.
A key avenue for future work is the analysis of near-simultaneous multi-sensor acquisitions, which would allow for direct inter-sensor bias quantification and cross-validation, as well as systematic comparison to emerging methane retrieval products from Planet and other commercial providers.
Conclusion
This paper delivers a technically detailed assessment and integration of Tanager-1 for facility-scale methane detection, in explicit comparison with EnMAP and PRISMA, within a unified matched-filter-based processing architecture. The analysis underscores the interplay between sensor SNR and structured artifact suppression, and demonstrates the operational necessity of artifact-aware retrieval (CW-CMF) for pushbroom imagers. These advances enable robust, uncertainty-quantified emission mapping in complex, heterogeneous terrains, setting a methodological benchmark for future multi-sensor methane monitoring efforts.