General comments

The manuscript presents results of a multiyear inverse modeling study estimating fossil CO₂ emissions using observations in the Paris region. The ability to estimate annual emissions with errors under 10% is a major advance presented in the paper. The paper is well-written and can be accepted after minor revision. One notable deficiency is lack of an inverse model description, suggest adding a section outlining the method.

Detailed comments

Page 1 Line 30 Suggest giving an uncertainty range to estimated 5.2% trend (seems to be in order of ±6% based on change of 32.6±2.2 MtCO2 in 2020 to 34.3±2.3 MtCO2 in 2021)

P2 L18 Need to check if the references are most recent for Boston (Northeast corridor), and also for Los Angeles. Can mention dense NIST network around Washington DC.

P5 L17 Suggest giving readers more detail about the method used in Lian et al 2022, when reporting the revisions, that would save readers effort and help them understand the full merit of both this and the previous study. Also, need to give somewhere a summary of key points of the inversion approach, like, using station-to-station concentration gradients as “observations”, control state, wind speed filters, horizontal resolution, PBL height filters, etc.

P8 L10, L15 Better give 2% and 3% per year trend numbers with uncertainties like 2±X%

Technical corrections:

P2 L10 Better write “inversion” instead of “atmospheric inversion” when citing Tarantola 2005.

P4 L4 For ODIAC, a more popular reference could be Oda et al ESSD 2018

P9 L12 Remove extra digits in text: diagnostic phenology41

Summary

Lian et al. present a study investigating long-term changes in CO2 emissions in the Greater Paris Area using different emission data products in combination with atmospheric observations. The inventories provided by origin.earth, AirParif and TNO can be the basis for policy decisions and they are validated using a bayesian inversion system which relies on assimilating morning and afternoon observations from a ground-based network. This emission monitoring framework performs well, is able to detect trends and short-term changes in emissions, here due to COVID-lockdowns. Overall, the paper is well-written and clearly structured. The description of the components is concise and a lot of information and illustrations of the actual performance are given in the appendix. The scope of the paper aligns very well with ACP and I can fully recommend publication after some minor changes have been considered.

General comments

1.) Unfortunately, the description of the modelling framework and its performance is very short. A lot of instructive and convincing information (plots) are only found in the supplemental materials. It could be worthwhile considering moving at least one into the main text.

2.) The manuscript does not discuss any other greenhouse gasses. In recent years several mobile surveys have been conducted highlighting significant CH4 emissions in the region. It would be more balanced to mention CH4, N2O as other gases that need to be mitigated (or why they can be ignored for the Plan Climat de Paris).

Specific comments

P3L17: please provide a quantitative measure of the instrument performance, what does ‘high precision’ mean here?

P4L33: formatting issue with “ru le”

P4L36: consider changing “imposed” to “required to be”

P15 L10: The blue boxplots should be added to the legend or the description of the other symbols to the captions. Splitting up the information seems unnecessary.