Please find the attached file. 

The study by Thongsame uses a chemistry model and new geostationary satellite data to estimate nitrogen oxide emissions in Thailand. Overall, it is an interesting study using established methods to estimate these emissions. Secondary comparisons with TROPOMI and surface sites provide additional validation of the work. The discussion on the discrepancies between TROPOMI and GEMS are an interesting and valuable addition to the literature. The analysis is detailed, which suites the scope of ACP and suitable of publication subject to some minor corrections:

• In the abstract, GEMS does not “simulate” NO2 (Line 29).
• The authors state that GEO has higher spatial resolution than LEO. If oversampling is used, the LEO datasets can be higher resolutions (i.e. < 1 km) (Line 90).
• The discussion on the averaging kernels is unclear. Please add the equations used to apply the AKs and clear description. It would also be good to add a figure (e.g. in supporting information) comparing the GEMS and TropOMI AKs.
• Why is 230 hPa used for the tropopause (Line 167)? It will also shift in the day, which can be important. Please justify the reasoning behind this.
• Please expand on how the RMSE is calculated as the error term for GEMS data (Line 226)? Would the RMSE be based on the comparison of GEMS with a secondary data set?
• In Section 2.7, how are the metrics (NMB) calculated? Is this done using the observational error? I ask as the MB is used for the surface sites and no error term is available, so I assume the satellite metrics are normalised via the error term. Please make this clear what is used.
• In Section 3.1 and discussion of Figure 5, please add some more quantitative discussion as scientifically weaker as it is.
• Figure 6, would you not expect more iterations before the IFDMB converges. Iterations of 2-4 seems low or very quick to converge on a solution?
• A more detailed discussion on the differences between TropOMI and GEMS. Why can you not directly compare the two products? Would that not provide a systematic bias between the two instruments which other studies could use.
• For Section 3.5, could you not replace e.g. the GEMS apriori with that of TropOMI and recalculate the trop col NO2 around the TropOMI overpass time and GEMS hourly measurements. That would allow more direct and useful comparisons.