In the paper, ‘DRIVE v1.0: A data-driven framework to estimate road transport emissions and temporal profiles’, the authors present a data-driven framework to calculate road transport emissions using a multi-modal macroscopic traffic model. Road transportation is one of the largest sources of greenhouse gas and air pollution emissions. Therefore, modeling this sector is critical not only to develop an understanding of the spatial/temporal dynamics but also to monitor, manage, and ultimately mitigate emissions from this sector. The authors used this model to compute emissions for Munich, Germany, and evaluate their computed magnitudes in space and time.

The following would be my overall ratings using the review criteria selected by the journal:

• Scientific significance: Excellent (1)
• Scientific quality: Good (2)
• Scientific reproducibility: Poor (4)
• Presentation quality: Good (2)

As my ratings show, reproducibility or traceability is a major problem with this paper. The reasons (explained in detail later) range from improper or insufficient citation, use of non-English data sources, and proprietary datasets. All these reasons make it difficult for other scientists to recreate this data product for different parts of the globe. As a result, I will recommend ‘major revisions’ so as to allow the authors to modify the manuscript and/or the model accordingly. Moreover, the authors’ approach to computing road transportation emissions is not novel. Similar methods can be found in the DARTE and Vulcan models (Gately, Hutyra and Sue Wing, 2015; Gurney et al., 2020). In their revision, I would encourage the authors to refer to previous attempts at modeling road transportation emissions and highlight how DRIVE is novel or different.

Issues of reproducibility:

• The authors calculate emissions for Munich, Germany, using a lot of European and German data. However, they do not state whether the model they present is limited to Germany or Europe.
• The authors refer to HBEFA emission factors in multiple parts of the paper, but there is no citation or reference to a table in the paper.
• The authors use a macroscopic traffic demand model maintained by the City of Munich. Unfortunately, this model is not cited, and the source is not mentioned. It is also unclear whether the model is open source or publicly/freely available.
• The authors use traffic counting data from the city administration and BASt. Citations and sources are missing. From the description provided by the authors, it appears that the data sources might be in German. If this is true, it would be an additional hindrance to reproducibility.
• It was not clear how the correction factor k_i,vcwas calculated.
• Table B2 (Appendix B) describes scaling factors for Passenger Car Units. However, a description of how or why such scaling factors were used is missing.
• All German citations lead to reproducibility issues.
• In section 3.2, a citation for ‘emission regulation’ is missing.

Issues of quality:

• The authors have often used qualitative terms like “a substantial portion”. I would encourage them to quantify these.
• DRIVE is built for Munich, a city in Germany. The authors do not explain why they focused on this city and did not develop a model for Europe or even a global one.
• Related to the above point, the authors do not elaborate on whether this model can be extrapolated to the rest of Europe or the globe.
• The HBEFA emission factors are national-scale factors, but DRIVE is for a specific city. The authors need to address the variation of factors within a country and/or whether the national-level statistics are representative of the city of Munich.
• While introducing the macroscopic traffic demand model for Munich, the authors do not describe its spatial resolution.
• In section 2.1.3, the authors need to elaborate on what “manual data curation” and “automatic preprocessing” mean and/or involve.
• Section 2 of the manuscript is the ‘Methodology’ section. However, the authors include some results in this section. These should be moved to the proper section.
• The universal and/or SI system uses a dot (“.”) as a decimal point and a comma (“,”) as a separator. The authors should use these standards consistently in the manuscript and not use the German system.
• Section 3.1 starts talking about ‘LOS’ without introducing or explaining the acronym previously.
• In the same section, the authors also do not explain ‘Stop & Go’ and ‘Stop & Go 2’.
• While describing ‘well-to-tank’ and ‘tank-to-wheel’ emissions, the authors need to explain how they avoid double-counting (e.g., oil tanker trucks can feature in both).
• To scale weekday traffic volume to the weekend, the authors multiply by 0.8. The source (or citation) of this number needs to be explained.
• Figure 6: Different vehicular types have very different magnitudes, so I recommend using different color palettes for each map to demonstrate distinct spatial patterns better. Magnitudes can be contrasted using a simple bar plot or a pie chart.
• Figure 7: The top row figures lack a color bar, and the bottom row figures need a quantitative color bar rather than a qualitative one. Moreover, I recommend plotting a percentage relative difference map to show spatial patterns of differences.

References:

Gately, C.K., Hutyra, L.R. and Sue Wing, I. (2015) ‘Cities, traffic, and CO2: A multidecadal assessment of trends, drivers, and scaling relationships’, Proceedings of the National Academy of Sciences, 112(16), pp. 4999–5004. Available at: https://doi.org/10.1073/pnas.1421723112.

Gurney, K.R. et al. (2020) ‘The Vulcan Version 3.0 High-Resolution Fossil Fuel CO2 Emissions for the United States’, Journal of Geophysical Research: Atmospheres, 125(19), p. e2020JD032974. Available at: https://doi.org/10.1029/2020JD032974.

The paper presents DRIVE v1.0, a road-transport emissions inventory framework that fuses a static macroscopic traffic model with city traffic counters and applies HBEFA 4.2 to produce link- and hour-resolved emissions with associated uncertainty. The workflow and data sources are clearly described and the results are policy-relevant for city-scale air-quality and GHG applications. The manuscript would benefit from clarifying several key assumptions and from moving some information that is currently in the supplement into the main text to support reuse. 

Major comments

In Section 2.3 (p. 9), LOS thresholds are tuned so that VKT shares match national FCD data within 1%. This calibration is critical for emission factor assignment, yet the actual threshold values per road class and the pre/post VKT distribution are not presented. Please report these in the main text. In addition, discuss whether using national FCD distributions for an urban network dominated by signalised intersections introduces systematic bias, and quantify the sensitivity of NOx and CO to a ±10% shift in all thresholds.

Section 2.5 fixes a 1.5 km allocation radius based on an assumed travel time at 60 km/h. This assumption may not hold across all road types and congestion states. Please provide a sensitivity analysis (e.g., 0.8 km, 2.0 km) to show the impact on spatial allocation of cold-start emissions. Also, all temperature binning uses a single urban station. Given the size of the domain, is this representative? Finally, Figure 8 and p. 16 note negative NOx cold-start factors above 25°C. Clarify whether such negative factors can lead to negative hourly or link-level totals and whether you impose a non-negativity constraint.

The discussion on p. 15 attributes much of the CO difference with UBA and TNO to the assumed uniform 120 km/h motorway limit. While I understand that counter-based speed data have limitations, “unreliable” (Section 2.1.2) is too vague—please quantify coverage or bias. Even partial speed information could help construct a more realistic speed distribution. Consider adding a scenario with higher or unrestricted motorway speeds to estimate the impact on CO, and report CO contributions by road class.

The mapping from 8+1 counter classes to HBEFA categories is in Appendix B2 but is central to your method. This should be moved into the main paper or SI. The spatial correction factor κ is derived from weekday averages; please comment on whether this remains valid for weekends/holidays. If possible, validate κ-corrected class shares at the 64 independent stations, not only total volumes.

Equation 6 combines activity error with emission factor uncertainty assuming independence, yet EF depends on LOS which is derived from activity. Please test a scenario with positively correlated perturbations (for example, correlation coefficient 0.3–0.5) to illustrate the possible underestimation of total uncertainty.

Section 2.2 uses the same temporal factor for all minor roads due to lack of counters. This is a practical assumption but potentially introduces bias. Please provide an upper bound estimate of the VKT/NOx error this could cause. Also report the proportion of days filled by imputation and its effect on annual totals.

Figure 9 shows systematic overestimation at high volumes. Please include a breakdown of errors by road class and LOS to help identify whether this bias is linked to particular conditions. Also, specify how many stations were excluded from validation and their spatial distribution.

The comparison in Table 4 is useful but could be more diagnostic. Splitting differences by road class or simple urban/rural zones would help disentangle whether mismatches are driven by spatial allocation or by speed/EF assumptions.