This paper addresses the 2025 fires at the Iberian Peninsula using TROPOMI CO and NO₂ data.

The main development presented in the paper is the “synergy” method to quantify CO and NO₂ emissions from individual fires. These top-down estimates are compared to bottom-up estimates from GFAS, GFED, and FINN.

A main finding is that the CO emissions from the top-down method are much higher (3x) than the bottom-up estimates. Concerning the NOx/CO mole density ratio (MDR: NOx emissions relative to CO emissions: a metric for combustion efficiency), the paper finds variability in time and space, which is sometimes consistent with bottom-up estimates.

Overall, the paper presents a new “satellite data only” methodology that holds some promise to independently quantify wildfire emissions in complex fire scenes.

The paper is well written and clearly presents the results. I have a number of small comments, which I add as annotated manuscript and supplement. I have some major remarks that could improve the quality of the manuscript further. These points are listed below.

1. Satellite data quality

TROPOMI columns are derived from fingerprints in the absorbed spectrum (UV/vis for NO₂, Short-wave IR for CO). Specifically, for NO₂, the sensitivity to near-surface NO₂ depends on the amount of scattering. Specifically, in the quoted Wang et al. (2026) paper it is written: “ However, due to the lack of explicit fire-related a priori information in current satellite NO₂ retrieval algorithms, the resulting data products exhibit large uncertainties under fire conditions.”, and “Retrieved tropospheric NO₂ VCDs increase by up to 100 % at locations greatly impacted by fires and by about 80 % in surrounding areas. “. I am a bit surpised that this source is error is not discussed at all.  These errors will both affect the plume mask and the MDR. For CO TROPOMI, errors related to fire scenes are likely smaller, because of the vertically flat averaging kernel and limited effects of aerosol on the CO retrieval.

2. Plume masking algorithm

The main issue in plume masking is to determine the CO and NOx added to the background. Thus, a background is proposed (section 4.1.1). However, in the case of overlapping plumes, the background might be influenced by upwind plumes (which is OK, because you want to analyse the amount that is added by the new plume). Although this becomes clear from the paper (e.g. figure 5) and the supplement, it would be worthwhile to explain this better in the manuscript. Moreover, in case of an upwind plume, the size of the background (0.8 degree by 0.6 degree) will encapsulate likely only part of the upwind plume and hence lead to uncertainty. Since the “synergy” methods is central to the paper, it would be helpful to present a more elaborate error analysis for background selection (now figure 3 presents a plume-free background, while figure 5 presents a more complicated case). Concerning figure 5 and potential biases, it remains a bit unclear how the “total” result from the “standard threshold” method differs from the newly developed “synergy” method. Mentioning these numbers would be valuable information.

In the Supplement additional information is provided. However, the “standard”, “threshold”, and “synergy” methods are not very clearly explained. It would be good to start with the “standard threshold” method (e.g. used for NO2) and then explain how the synergy method applies the NO2 mask to CO satellite data, i.e. define the “standard threshold” method as standard.

3. Plume chemistry

CO is relatively long lived, and emissions are estimated using the IME method. For NOx the application of the IME method is questionable. Apart from the NO (emitted) to NO2 conversion, also the lifetime of NOx (~ hours) might bias the IME method. In line 536 “plume chemistry” is explicitly mentioned, so it is important to quantify potential errors due to (lack of) chemistry.

The manuscript quantifies CO emission rates and NOx/CO mole density ratios (MDRs) for 46 individual TROPOMI measurement points spanning 15 major wildfire events during the August 2025 Iberian fire season, and compares these top-down estimates with three fire emission inventories: GFAS, GFED, and FINN. The authors developed a synergy plume-masking approach that uses TROPOMI NO₂ plumes to spatially constrain the CO plume mask in complex, multi-fire scenes. The results show that inventories underestimate CO emissions by approximately a factor of three and that TROPOMI-derived NOx/CO ratios are systematically lower than inventory-based ratios. The approach and findings are of interest; however, several methodological choices require clearer justification or more detailed description before the reported “factor of three” difference and MDR results can be fully assessed. I provide detailed comments below.

Major comments

1. TROPOMI observes plumes at a much finer spatial resolution than GFAS and GFED. Although the manuscript sums inventory emissions within a fixed area around each fire, it does not discuss whether differences in spatial resolution contribute to the scatter in the results. Coarse inventories can mix emissions from nearby fires within the same grid cell, whereas the TROPOMI plume-masking method is designed to separate individual plumes. This may partly explain why TROPOMI–inventory correlations are weaker than inventory–inventory correlations. A sensitivity test, or at least a discussion of this issue, would strengthen the analysis
2. Related to the previous comment, the manuscript does not clearly explain that the quantities being compared are fundamentally different. TROPOMI provides an instantaneous emission estimate at the satellite overpass time (~13:30), whereas GFAS, GFED, and FINN provide daily average emissions derived from bottom-up methods. Part of the reported factor-of-three difference between TROPOMI and the inventories may therefore be due to differences in temporal aggregation. A short explanation of these differences and their potential impact on the comparison is recommended.
3. The choice of the 0.8° × 0.6° upwind background region is not fully justified. It is unclear why this size was selected, whether it was applied to all cases regardless of plume conditions, and whether sensitivity tests were performed. In complex fire scenes, the background region could contain smoke from nearby fires, potentially biasing the results. A sensitivity analysis and discussion of this uncertainty would be valuable. The authors should also explain why they chose the upwind method instead of the alternative approaches cited in the manuscript.
4. The use of ERA5 winds at 0.25° resolution may introduce uncertainty because these winds may not accurately represent plume transport, especially in mountainous terrain. Since wind speed is already identified as the largest source of uncertainty, the manuscript should discuss wind representativeness and plume injection height more explicitly.
5. The choice of qa > 0.5 for TROPOMI data is not fully justified. Although the manuscript explains that this threshold increases the number of observations near fires, it does not explain why 0.5 was selected instead of the more commonly used qa > 0.75 threshold.
6. The plume-masking method is the main contribution of the paper, but the description in the main text remains largely qualitative. Important details, such as threshold values and the procedure used to match NO₂ plume pixels with CO pixels, are only described in the Supplement. A more complete description in the main text is recommended.

Minor comments

1. The study uses a fixed conversion factor of 1.3–1.5 to estimate NOx from NO₂. Since this factor is important for interpreting the MDR results, a sensitivity analysis using a wider range of values would help demonstrate the robustness of the conclusions.
2. Table 1 provides only a single latitude and longitude for each fire. For reproducibility, it would be helpful to provide the actual spatial extent used in the analysis, such as bounding boxes or fire polygons.
3. The Abstract refers to NOx/CO MDR, although TROPOMI measures NO₂ rather than NOx directly. For clarity, consider using a term such as “NO₂-derived NOx/CO MDR.”