In this paper, Jacobs et al. evaluate the impact of different digital elevation models (DEMs) on XCO2 retrievals from the OCO-2 Instrument. By comparing different DEMs, including OCODEM used in the ACOS V10 algorithm, the NASADEM+ used in ACOS V11, and the Copernicus DEM used in ACOS V11.1, the authors conclude that the Copernicus DEM has better overall continuity and accuracy. The authors demonstrate that the differences in DEM have a significant impact on the bias-corrected XCO2 retrievals. With the use of the Copernicus DEM and updated quality control filtering, the ACOS V11.1 XCO2 retrievals show generally similar or improved accuracy and spatial continuity as compared with V10 and V11 retrievals. In addition, V11.1 also has increased the data volume that passes through the data quality filtering. This paper discusses an important input dataset for satellite trace gas retrievals. The results and recommendation from this study should be useful for both algorithm developers and data users. The paper is well organized, and the figures are mostly clear. However, I feel that in its current form, the paper is more tailored towards readers who are already familiar with (or have working knowledge in) the ACOS retrieval algorithm. It would be fine as a chapter within the ATBD or product readme file, as in those documents, the other chapters would have provided the necessary context. As a stand-alone paper, I found it rather difficult to follow. Some terms and parameters are casually introduced with little explanation. Some important conclusions are drawn without the necessary supporting evidence. I would, therefore, recommend major revisions before this paper can be accepted for publication in AMT.

Specific comments:

Page 2, Line 6: can you provide a reference that the 14-year GOSAT data record is long enough to inform multi-decadal climate variations?

Page 2, Line 25: can the authors elaborate more on the tests that fix surface pressures? What are the other deleterious effects and why? This would be interesting, given that there are GHG retrieval algorithms that use assimilated surface pressures (rather than retrieving them).

Page 5, Line 5: this part repeats what has already been introduced in page 2.

Section 2.1: can the authors clarify on the status of OCO-2 XCO2 products? Will V11 be released or only V11.1 be released?

Page 5, line 25: can you briefly explain the footprint bias correction (it is also in equation 6)?

Section 2.1.1, could you please rewrite this section to help those who are less familiar with ACOS to better understand it?

Equation 6: where does “0.016” come from? One can guess from the figure, but it would be helpful if the authors can give some explanation.

Page 6, Line 28: can the authors give some quantitative results on how much more robust the new filtering is?

Page 8, Line 1: is the aggregation done to all DEMs or just OCODEM?

Page 9, Line 13: GMTED2010 is not defined or used elsewhere in the paper.

Section 3.1: it would be good to run V11 full physics algorithm for a small subset of OCO-2 data to confirm that you will get the same bias-corrected XCO2 as V11.1.

Section 4.5: it could be useful to produce maps for Lauder, Pasadena, and Eureka that are similar to Figure 11 (even if just place them in the supplemental information).

Figure 1: fitting at low TCWV range appears to be of worse quality - how does this affect the corrected h2o_ratio and the overall results, given that significant improvement is seen over low TCWV?

Figure 5: consider putting mean and standard deviation of delta-altitude in different panels so that the differences in the mean between NASADEM+ and Copernicus DEM are more obvious.