The paper by Oomen et al presents top-down VOC emissions over Europe derived from TROPOMI over a four-year period. The authors exploit the high performance of TROPOMI to derive the first weekly top-down VOC emissions separating biogenic, biomass burning and anthropogenic emissions over Europe. They propose a dedicated validation of the HCHO columns using FTIR ground-based measurement in Europe to adjust the bias correction to the European domain and use the corrected columns for the inversion. The inversions are performed from May to September each year from 2018 to 2021 and show an increase of biogenic and biomass burning emissions and a decrease of anthropogenic emissions. The authors demonstrate the improvement given by weekly-derived emissions compared to monthly ones. They provide a tentative of evaluation of the results with independent ground-based measurements and stress the need for additional ground-based measurements of HCHO in Europe, especially in the South, to reinforce the evaluation/validation capabilities, which remain poor. They also provide an interesting sensitivity study which underlines all the difficulties to derive accurate VOC emissions from HCHO satellite observations.

The paper is well structured, written and documented. It provides a step forward in inverse modelling of VOC emissions at high temporal and spatial resolution with TROPOMI. The paper is suitable for publication in ACP after some clarifications.

Main comments:

1. HCHO column validation: the authors use FTIR measurements in Europe to validate and propose a bias correction to the HCHO columns to be used in the inversion. Most of the FTIR stations are in the northern part of the domain where emissions and columns are smaller, it may introduce a bias in the bias correction for the large columns, not well represented in the north. The authors stress the lack of ground-based measurement in Southern Europe and its potential impacts in their discussion, but I wonder if they could use FTIR stations in other part of the world, where HCHO columns are higher and similar to the columns in Southern Europe to complement their validation and ensure they do not introduce a bias in the correction.
2. Model evaluation: the comparison with MAX-DOAS measurements and optimized HCHO columns does not show any improvement (except in Greece). The authors state that this might be due to measurements of the MAX-DOAS instruments during cloudy days. Did they try to consider only clear-sly days in their comparison to quantify the impact? It might be interesting.
3. Biogenic emissions: in the inversion setup (lines 256-257), it is written that the emission parameters are applied to both isoprene and monoterpene, but the results are only discussed in terms of isoprene emissions in the manuscript. The authors should add a discussion on the monoterpene emissions if they have effectively inversed them or explain why they do not show the results.

Specific comments:

1. Lines 14-15: the authors should consider giving numbers when they mention the results of their sensitivity study.
2. Line 15 and elsewhere, “variable meteorology”, this expression might be confusing. Meteorology is variable by nature. The authors should clarify what they call “variable meteorology”.
3. P8, section 2.4 : the authors try to use climatological averages based on measurements in the 90s or early 2000s. They mention the difficulties and that the results should be considered with caution. I wonder if something is known or could be derived from climate simulations for example to quantify how far from now these measurements are when considering temperature increase and land-use changes.
4. Line 241: I was confused by the sentence “the inversion is carried out with a time step of 24 hours” whereas the inversion provides weekly emissions. Rephrasing the sentence might be useful to avoid the confusion between the time step of the forward model calculation and the inversion.
5. Inversion setup: It is not completely clear which assimilation window is used. Is it the May-September period?
6. Line 259: On which basis the 10¹⁰molec.cm^-2s^-1 threshold has been chosen? How many pixels are concerned, and where are they located?
7. Line 282: how have the spatial correlation of anthropogenic emission errors been chosen?
8. Line 286 and 570-574: is the filtering out of low columns made on individual observations or after the average?
9. Line 307-308: Is there a reference that compare the different sensitivity? If yes, it might be interesting to refer to or if not, to provide a plot to illustrate it.
10. Line 375-376: are the oceanic data excluded with the filter mentioned line 259 or is it a new filter?

Technical corrections:

1. Line 185: replace ‘wth’ by “with”
2. Line 359: remove the url.