This manuscript presents an excellent study that estimates aerosol emissions in the global aerosol climate model ECHAM-HAM using assimilation from POLDER-3/PARASOL observations for the year 2006. There is extensive evaluation with independent observational datasets (AERONET as well as MODIS Dark Target). Furthermore the results of this study are nicely compared with AEROCOM I, AEROCOM III and CMIP5 as well as some of the previous assimilation studies. Last but not it is also discussed (using sensitivity experiments) the sensitivity of the emissions to relative humidity due to aerosol water uptake, especially important for the case of soluble species. I think this study is an important contribution on the topic and I strongly suggest acceptance of the manuscript for publication but I have a few minor and technical comments to be considered.

Comments

What I missed is some information for the meteorological set up of the simulations with ECHAM-HAM climate model (e.g. if the simulations are nudged, spin-up time of the simulations, if is there is a specific reason for the selection of the year 2006).

Maybe they authors could think of revising the title inserting also the term of assimilation.   

Caption of Figure S1: I think it is wrongly written (f) sulphur dioxide (SO2). It should rather be WAT. Furthermore the acronym WAT should be also defined in the caption and could be introduced  in the text of the manuscript when the water uptake of soluble species (resulting to high AOD values) is discussed (e.g. lines 575-584).  

Caption of Figure S8: For consistency with the text it should be noted as "CTLERA5" than simply "ERA5".

line 20:  I would suggest " over isolated island sites at the ocean" instead of "over isolated island sites over the ocean".

line 24: Define at some place the acronyms such as GFAS

line 48 and line 50: You may delete "note that" in both sentences.

line 50: It should read "(from 1m to several km)"  instead of (about 1m to several km).

line 52: "Emissions from biomass burning" instead of "Emissions from biomass burning emissions"

Lines 54-56: Please define at some place the acronyms such as GFED4, FINN1.5, QFED2.4, FEER1.0 and GFAS.

Line 62: Is the term "diversity" used throughout the manuscript, the proper or the common word to express the ratio of the standard deviation to the mean. In many studies the word "range" is commonly used.

line 70: I would suggest "at least" instead of "at best".

line 72: I would suggest the plural "precursors" instead of "precursor".

line 74: Maybe "assessed" or "found" instead of "used".

line 79: I suggest " ...similar information and methods which they are not ..." instead of "... similar information and methods and are not ..".

lines 119-121: The sentence is not clear and it needs some rephrasing.

line 273: It is not clear what is the setup of the DASERA5 experiment? Is it the data assimilation experiment in ECHAM-HAM using relative humidity for ERA5? Please clarify in the text.

line 298: "along with the MAE" instead of "along the MAE".

line 298: " 3-hourly differences between the Experiments – POLDER" . It is better to be more specific. e.g. 3-h differences of CTLECHAM-POLDER and DASECHAM-POLDER.

lines 531-533: Considering the complexity of the loss and production processes that control the SO2 and SO4 fate in the atmosphere mentioned in this sentence, it could be nice to have a link to section 4.4.2 that you discuss these processes (e.g. as discussed in Section 4.4.2).

line 544: It should be "Figure 10" instead of "Figure 9".

 line 579: " ...matches the underestimation of RH by ECHAM-HAM (Figure 10c) while ..." I am rather confused here with the underestimation. Do you mean the small underestimation over ocean (Figure 10 c) below 500 m? Above this level there is clear overestimation of RH.

Review comments on “Aerosol emissions estimation with POLDER” by Tsikerdekis et al. submitted to Atmospheric Chemistry and Physics.

This study by Tsikerdekis et al. applied a Local Ensemble Transform Kalman Smoother (LETKS) method to derive aerosol emissions from POLDER SRON products of AOD (550 nm), AE (550-865) and SSA (550 nm) based on the ECHAM-HAM climate model. The methodology has been developed and tested in their preceding studies (Schutgens et al., 2012; Tsikerdekis et al., 2021, 2022). The present study is focus on the discussions about the obtained (top-down) global aerosol emissions and intercomparisons with reported values from model ensembles (AEROCOM and CMIP) as well as other observational constrained studies. Generally, the global aerosol emissions of all species with POLDER aerosol data assimilation increase with respect to the bottom-up inventories used in the ECHAM-HAM model and the values are consistent with other top-down studies. Additionally, the effects of relative humidity (aerosol water update) on the retrieval of emissions are also discussed and experimented. Overall, the manuscript is clearly structured and I would support for publication in ACP after some minor corrections.

1. I appreciate the authors discuss explicitly the values of annual emissions of global aerosol species and intercompare with other studies. However, what is missing but important is the daily or monthly variations of emissions which is one of the superiorities of top-down technique. I would suggest some discussions in the present study or more explicit study in the future to investigate the capability to capture the daily or monthly emission variations and to intercompare the variations from top-down emission datasets, for example, some datasets used in Elguindi et al. (2020, 1029/2020EF001520).
2. Although English is not my first language, I find some typos and mistakes in the manuscript. I would suggest the authors need to correct and revise carefully in terms of writing.

Specific comments:

Abstract L21: Why do not include AEROCOM II?

L52: Emissions from biomass burning or Biomass burning emissions

L52: … are estimated based on satellite measurements …

L72: -> differences … are …

L97: Zhang et al. (2015, 10.5194/acp-15-10281-2015) assimilate OMI AAOD to estimate BC emission over East Asia.

L155: The POLDER observations… Actually, you are using POLDER aerosol products derived from POLDER observations.

L170: Why do you use AE550-865, instead of AE440-865 which should be more proper over desert region?

L199: … based on Petters and Kreidenweis (2007).

L213: aerosol emission or aerosol emission fluxes? Please keep one consistently throughout the paper.

L231: have you ever tried to understand the effects on derived emissions once you change the assimilation time window?

L257: R -> bolded R

Equations 1,2: Please use consistent font, italic, bold as in the main text for matrix and vector.

The subtitles of 3.1, 3.2 and 4.1 are the same, please correct them.

The subtitles of 4.2 and 4.3 are the same, please correct.

Figure S1 (f): what do you mean of WAT AOD550 which accounts for 62% of total global AOD?

Figure A1. POLDER uncertainty for AOD… It’s better to indicate POLDER SRON product uncertainty.

Figure A1. In my opinion, the relative uncertainty for AE is a bit meaningless, because it may be dominant by the coarse mode cases where the AE is small and the relative uncertainty is high.

L285: how do you deal with dust bins and emission fraction of each dust bins?

Figure S2: could you explain how can you obtain POLDER uncertainty of AE? why is it ‘no value’ over Sahara region?

L308-309: what cause high AAOD over Australia coastline?

Figure S3: the global mean value does not appear at the left bottom corner.

L320-321: in this paragraph, you talk about AAOD, why you report MAE for SSA at the end, instead of MAE for AAOD?

L355-357: it’s interesting to know the effect on absorption is negative if only AOD is assimilated. Do you have any explanations?

L361: the latest C6 and C6.1 products should be better than C5 for comparison. In addition, why do you use DT only, you may find more insights from DT+DB combined dataset over bright surface?

L390: do you have high confidence of POLDER AE over ocean where the aerosol load is generally low?

L407: AEROCOM phase II?

L409: The latest study by Chen et al. (2022, 10.1038/s41467-022-35147-y) report global values from 2006-2011, and the emission for reference year 2010 is reported in Chen’s 2019 paper (10.5194/acp-19-14585-2019).

 L419: permitted -> emitted?

L426: dust emission is highly relying on the particle size range that you are modeling, could you explicit your setup of dust bins?

L441: some of the factors … are …

L611: -> global SS emissions

L710: is there any comment on the treatment of absorbing organic carbon?