This is a timely and useful manuscript on aerosol source apportionment modelling with a coupled regional–urban framework. It addresses an important gap by comparing source-resolved model output against several PMF-based observational datasets across a relatively large set of European sites. The multi-dataset setup is a clear strength, and I think the attempt to look at both the regional model performance and the added value of urban downscaling is particularly worthwhile. Overall, the manuscript is carefully done and generally well written, and I appreciated that the authors are fairly open about the uncertainties involved in comparing model results with PMF-derived factors.

In my view, the paper is publishable after minor revision. My comments are mostly about presentation and interpretation rather than any fundamental problem with the work itself. I think the main results are interesting and worth publishing, but a few points could be stated more clearly.

One issue that runs through the paper is the matching between PMF factors and modelled source categories or species. The authors are aware of this and do flag it in several places, which is good, but I think it deserves a bit more weight in how the results are framed. Some of the conclusions look fairly robust, while others seem to depend more strongly on the exact mapping choices that were made. It would help if that distinction came through more clearly, especially in the results discussion and concluding sections.

The results also suggest that urban downscaling does improve the representation of some local source contributions, especially road traffic and residential heating. At the same time, the picture is not completely straightforward, since in some cases the downscaling appears to magnify existing biases rather than reduce them. My sense is that the manuscript should be a little more careful here and avoid giving the impression that the added urban-scale treatment is consistently beneficial. It clearly helps in some situations, but not in all, and that comes across in the results.

More generally, the paper contains a lot of useful detail, but the main message gets somewhat buried at times because there are several datasets, many factor types, and a number of necessary source-matching decisions to keep track of. I think the final sections would benefit from a somewhat tighter synthesis. It would help to focus to three main points: where the model works well, where the main discrepancies remain, and what these mean for future source-specific applications, especially in urban health and policy.

General comments

This is a timely and well-executed manuscript that evaluates source-resolved aerosol simulations using a coupled regional-urban modeling framework. The use of multiple PMF datasets across a relatively large number of European sites is a clear strength, and the attempt to assess both regional model performance and the added value of urban downscaling is particularly valuable.

Overall, I find the study scientifically relevant and generally well presented. The manuscript addresses an important topic, and the analysis provides useful insights into both model performance and the interpretation of PMF-derived source contributions.

That said, some aspects of the interpretation could benefit from further clarification in order to ensure that the main conclusions are as robust and transparent as possible. In particular, this concerns (i) the mapping between PMF factors and modeled source categories, (ii) the attribution of organic aerosol (especially anthropogenic SOA), and (iii) the interpretation of the urban downscaling results.

These aspects appear closely connected to some of the key conclusions of the manuscript (e.g. regarding the traffic contribution and the added value of downscaling), and a more explicit discussion of associated uncertainties would, in my view, further strengthen the study.

I therefore recommend reconsideration after revision.

Specific comments

1. PMF factor-model mapping

Around lines 214-218 (Section 4.2, introduction to PMF factors) and in connection with Table S3, it may be helpful to provide a clearer and more systematic discussion of how PMF factors are mapped to modeled source categories.

In particular, it would be helpful to clarify which aspects of the results are expected to be relatively robust to reasonable alternative mappings, and which might be more sensitive to these assumptions. This could help the reader better distinguish between discrepancies arising from model performance and those related to structural differences between PMF factors and model variables.

If feasible, a brief qualitative discussion of how alternative plausible mappings might influence the results could further enhance the transparency of the analysis.

2. Allocation of anthropogenic SOA to road traffic

Around lines 240-248, all anthropogenic SOA is assigned to the road traffic factor. While the manuscript provides some justification for this choice, it might be useful to elaborate slightly further on the rationale.

In addition, it would be helpful to comment on how sensitive the results could be to alternative allocations (e.g. partial attribution to other anthropogenic sources such as solvents or residential combustion). Even a short sensitivity discussion, if available, could help to place the traffic-related conclusions in a broader context.

3. Interpretation of the sulfate-rich factor

Around lines 304-309, the inclusion of biogenic SOA and background OA in the sulfate-rich factor is noted. It may be helpful to expand the discussion slightly here.

In particular, this choice may make the interpretation of this factor less directly comparable to a purely secondary inorganic aerosol component. A brief clarification of how this mixed composition should be interpreted when comparing to PMF-derived sulfate-rich contributions would likely improve clarity for the reader.

4. Interpretation of urban downscaling

Around lines 12-15 (Abstract) and 334-340 (Section 4.3), the manuscript highlights improvements from urban downscaling for road traffic and residential heating. At the same time, the results also suggest that downscaling can, in some cases, amplify existing biases (e.g. for biomass burning).

It might therefore be helpful to reflect this nuance more explicitly in the text, so that the overall message does not give the impression that downscaling is uniformly beneficial.

More generally, the discussion of the added value of downscaling could potentially be strengthened by distinguishing more clearly between different aspects of model performance, such as:

• mean concentrations
• temporal variability
• spatial variability across sites

A clearer separation of these aspects could help to better identify under which conditions downscaling leads to improvements.

5. Residential wood burning and condensable emissions

In Section 7.1, the manuscript highlights the sensitivity of residential wood burning to the treatment of condensable emissions, which appears to be an important aspect of the analysis.

It may be helpful if the authors could elaborate more explicitly on how this sensitivity propagates into the source apportionment results. For example, does it primarily affect the absolute magnitude of the biomass burning contribution, its spatial distribution at the urban scale, or both?

Clarifying this point could help to better interpret the role of downscaling relative to uncertainties in emission representation.

6. Synthesis and comparison across factors

In the conclusions (Section 8), the manuscript provides a comprehensive overview of the results. Given the large number of datasets and PMF factors considered, the main messages could perhaps be made more explicit.

It would be helpful to provide a more concise synthesis that highlights:

• where the model shows robust performance
• where the main discrepancies remain
• what factors appear to drive these differences

In addition, a more systematic comparison across factors using consistent statistical metrics (e.g. bias and correlation, as already partly presented in Fig. 2 and Table S4) could help to facilitate comparison between datasets and source categories. A compact summary table might be a useful way to support this.

I believe that addressing these points would further strengthen the clarity and robustness of the manuscript. The comments mainly concern interpretation and presentation rather than fundamental methodological aspects, and I consider that the manuscript has the potential to make a valuable contribution after revision.