This study evaluates the potential for NIMCAM to constrain diffuse methane emissions in Africa. NIMCAM has a unique instrument design: 60 m pixels, no targeting, and only 3 (effective) spectral bands, on and off an absorption feature of methane ~1640 nm. The authors argue that, for typical satellite retrieval quality filtering, NIMCAM offers expanded observational coverage (relative to in particular TROPOMI) via its small footprint not being as impacted by clouds and aerosols. Through OSSEs, the authors further argue that NIMCAM can outperform TROPOMI for flux inversions in select months. The point that higher-resolution observations are valuable for seeing through the clouds in the tropics is well taken, but I do not believe the authors reach the bar of showing that NIMCAM would be useful for constraining diffuse emissions on the continental scale (let alone more useful than TROPOMI). I believe that the paper is not suitable for publication in its current form.

General comments

• As I understand it, the authors use fluxes from Feng et al. (2023) that are at 2 x 2.5 degree resolution, optimized using GOSAT. They regrid these fluxes to 0.25 x 0.3125 degree resolution, run them through GEOS-Chem, and then sample the simulated atmosphere using NIMCAM or TROPOMI observation locations. They then ingest these pseudo observations (with added noise) in an ensemble Kalman filter using the same prior emissions that GOSAT started with and see how well NIMCAM or TROPOMI can recover the true fluxes. I have the following concerns:
  • In general, I was not at all impressed by the OSSE results. Maybe this is related to the errors detailed below. In Figures 7 and 8, comparing “Prior-True” and “Post-True” on the fluxes, it seems that none of the instruments do particularly well, except for NIMCAM correcting some of the overestimated fluxes in the Congo. A lot of the error reductions are in areas where the proximity to the truth actually gets worse. Also, hasn’t this same research group constrained tropical African methane fluxes using TROPOMI (Lunt et al., 2021, https://doi.org/10.1088/1748-9326/abd8fa), for example the Sudd, but now they are saying TROPOMI is useless for this purpose?
  • Is there any proof that NIMCAM will be able to provide an accurate but low precision methane retrieval? There is only a citation to a paper that is not accessible (Woodwark et al., 2026). Have the authors shown that a 3 spectral channel retrieval can avoid systematic biases that would preclude its use in an inversion (perhaps this could be done by coarsening the spectral resolution of existing spectra covering ~1640 nm, such as MethaneSat?)
  • Do you add 15–30 ppb of noise to both TROPOMI and NIMCAM data before ingesting them? Is this correct given the estimated precisions of each instrument? Why is this different than the 2.5e15 and 2.5e17 molecules/cm2 per-pixel precisions given later on? By the way, using 2e25 molecules dry air/cm2, these translate to 0.125 and 12.5 ppb, which are overly optimistic. TROPOMI has a precision ~12.5 ppb (cf. Schneising et al., 2026, https://doi.org/10.5194/amt-19-2407-2026), a factor of 100 difference, so I encourage the authors to clarify how these precisions were determined.
  • Can the authors confirm that the square root should be applied in this manner in Equation 4?
  • Line 169: what do you mean when you say you assume the prior fluxes are the same as Feng et al. (2023)? Can this not be checked?
  • What is the rationale for Equation 2? Why not use the same prior uncertainty as Feng et al. (2023)?
  • How is the number of clear sky retrieval determined (i.e., how do you align spatially the 60 m S2 cloud probabilities with 1 km AOD that has been downscaled to, I believe, 1 km probabilities of 60 m pixels being lower than a threshold). I believe the sentence starting at Line 190 is incomplete?
  • What value is used for the lower limit of uncertainty in Equation 1?
  • I was confused by Line 340 since your OSSE does not vary transport.

Minor comments

• The authors refer to an instrument with 1 km spatial resolution and a 5-day repeat as “nominally GOSAT-2.” I believe GOSAT-2 actually has 10 km spatial resolution with a 6-day repeat (cf. Barr et al., 2025, https://doi.org/10.5194/amt-18-6093-2025).
• The source of the 60 m AOD data used in Appendix B is not stated.

Specific comments

• Line 7: “2–5” should be “3 and 5”
• Line 35: typo around “sensitivity”
• Line 48: remove comma after “region”
• Line 77: “various” to “variability” (?)
• Line 139: 0.315 to 0.3125
• Line 154: add W on 20 degrees
• Line 202: “obs” isn’t subscripted