This manuscript describes a high-resolution road traffic emission inventory developed for the megacity of Delhi using advanced and detailed traffic data and speed based EFs. The strength of the estimation methodology presented in the paper is in its usage of very detailed and advanced input datasets, which allows obtaining high-resolution spatio-temporal emission maps and disaggregate the emission results according to several categories, including vehicle types, road classes or hours of the day. The resulting dataset is therefore relevant for policy makers, but also for air quality who wants to use it as input in the chemical transport models. The paper is well written and structured, and its quality is very good, which makes it a good contribution to ESSD. However, there are some aspects related to the methodology proposed that should be better clarified before the manuscript is accepted for publication.

1. Hourly congestion data from TomTom is used to estimate traffic flow information per road link following equation 3, which is presented in section 2.1.1. According to this equation, if congestion is 0, the resulting traffic flow will also be 0. Nevertheless, null congestion does not imply having no cars circulating. Can you clarify how this issue is corrected in the model?
2. The resulting traffic flow information is validated by comparing estimated and reported annual VKT information. However, results from this comparison are not provided. Please add them. Also, could you provide a comparison between estimated and measured hourly traffic flow for those locations in which you have observations? This will give to the reader a better feeling of how robust this approach is.
3. According to the authors, “the emissions are further adjusted with a factor of 1.2 to account for real-time driving behaviour (frequent braking, acceleration, deceleration) as per the study by Lejri et al., (2018)”. This assumption seems to me a bit arbitrary a not well justified. Is this factor applied to all hours of the day (including those when congestion is low)? Is this factor applied equally to all pollutants? Why?
4. As mentioned by the authors, emissions are estimated using COPERT 5, which is a European emission model and has not been calibrated for Indian conditions. Can the authors elaborate a bit more on the potential uncertainty for key vehicle categories such as two-wheeler motor bikes? Perhaps the EFs reported by COPERT could be compared against results reported by local studies such as Adak et al., 2016, https://doi.org/10.1016/j.scitotenv.2015.11.099.
5. The vehicular classification is done making use of shares provided by different local sources of information. Are these shares based on information of registered vehicles or actual circulating vehicles? (i.e., old vehicles may appear as registered by they are barely used in reality)
6. According to the authors, “annual emissions have been calculated by summing the hourly emissions to get daily emissions and then multiplying with 365”. By doing this, authors are assuming that for all days of the week (Monday to Sunday) and all months of the year (January to December) traffic activity and emissions present the same intensity. However, traffic activity and associated emissions typically present a drop during weekends when compared to weekdays, and they can also present drops/increases during certain months of the year. Is this not the case for Delhi? Can the authors provide some information that support they hypothesis (i.e. emissions are constant throughout the year).

Other comments:

• Particulate matter emissions are usually expressed as PM (regardless of the fact that they include or not non-exhaust emissions). I would recommend to change the acronym from PME to PM - In the text the authors already specify that PM emissions only include exhaust - and also specify if this PM equals PM10 and/or PM2.5.
• “In this study we have shown a data driven approach where the quality of input data is likely to improve the emission estimates.” I believe this is a too strong conclusion. Emission results from this work differs significantly from previous estimates as it makes use of more updated and refined information, but it cannot be concluded that the estimates have been improved. In order to say that, an evaluation of the emission dataset should be performed by, for instance, performing an air quality modelling study and comparing the results against observations.

This study presents a multi-pollutant hourly gridded vehicle emission inventory over Delhi, with a bottom-up methodology. Hourly congestion data from TomTom were used to account for hourly changes in the speed. The traffic flow was derived from speed based on fitting formula established by the previous research. The percentage share of vehicle technical parameters (vehicle type, fuel used, emission standards, etc.) was provided by survey reports or previous study in Delhi. Emission factors were calculated by COPRET-5. This paper is well-written and presents results that would be interesting to the air quality modeling community or policy makers. However, I have several concerns that the authors should consider when revising the manuscript, as listed below. I recommend this work to be published after the following comments are adequately addressed.

1. Section 2.1.1., Values in Table S2 were different from those given in Malik et al., 2021. This would lead to huge deviations in subsequent calculations. What are the reasons for this revision?
2. Line 2018-220, How to correct the speed and traffic volume, all roads or some specific roads?
3. Whether the vehicular share (%) was constant for the specific road throughout the day? It was not reasonable. And, this directly determined the result of vehicular volume and emission share (section3.1 and section 3.6).
4. Table S4, all 3W vehicles were Euro 4 ?
5. Section 3.3 and 3.4, The authors seem to assume that as long as the road types are the same, the relationship between speed and vehicle flow were the same too. The resulting spatial distribution may have large errors. Authors should consider making some corrections.