This study developed two merged VOD datasets, VODCA CXKu and VODCA L (VODCA V2). To achieved this goal, a weighted merging methodology was used to combine multi-sensors with high-frequency bands into VODCA CXKu indicative of upper canopy characteristics and to merge SMOS and SMAP into VODCA L which is more sensitive to AGB. I appreciate the improved methodology in merging and evaluation processes employed in VODCA V2 than v1. However, I have a few comments before the publication:

Major comment:

1.    Given the potential for extensive use in long-term ecological applications due to its 35-year timespan (1987-2021), ensuring consistency throughout the VODCA CXKu & L data is paramount. To this end, I strongly recommend including yearly time series of VODCA CXKu / L at global, continental, or landcover scales.

Minor comments:

1.    Line 15: does the canopy include trunks? 

2.    Line 125 “Ku-band (19.4 GHz)” and “Ku-band (18.7 GHz)”, microwave at 19.4 GHz and 18.7 GHz belongs to K-band (18-27 Ghz), why “Ku-band” was used in this ms?

The manuscript submitted by Zotta et al. aimed at the development of long-term vegetation optical depth products called VODCA V2 from multi-sensor, multi-frequency microwave sensors. Compared with VODCA V1, VODCA V2 has made some new changes, such as adopting a new weighted fusion scheme, combining high-frequency bands together, and providing L-band VOD. Overall, the manuscript reads well and the research topic fits within the scope of Earth System Science Data (ESSD). However, the current version of the manuscript is still far from ESSD publication requirements unless the authors address the following major and detailed comments.

Major Comments:

1. The rationale provided for combining high-frequency VODs (i.e., C, X, and K-band) is currently inadequate. The authors should include detailed scientific evidence, such as assessments of the spatial and temporal correlations among these three frequency VODs.
2. Despite the work done by the authors in evaluating VOD products, however, I did not see interesting information considering the low correlation values between VOD and the vegetation proxies selected, for example with the ASCAT slope. Could the authors clearly indicate what the ASCAT slopes represent? Is it vegetation water content or?
3. As a long-term product fused from multiple satellite sources, it is essential for the authors to dedicate a section to evaluating the continuity of the developed VOD, particularly over time. While the temporal correlation between VOD and vegetation proxies was calculated, this alone does not adequately assess the continuity of VOD.
4. Why were all sensors fused together instead of selecting the highest quality VODs during overlapping observation periods? For example, considering SMAP's superior quality over SMOS, which is more prone to radio frequency interference (RFI), was an evaluation done to determine if fusing SMOS and SMAP data during their overlapping periods might compromise data quality compared to using SMAP alone? Notably, L-VOD can provide global coverage with a 10-day temporal resolution, regardless of whether it uses only SMAP or both sensors.
5. What advanced features do the fusion products offered by the authors have compared to existing L-band fused VOD products? Furthermore, comparisons with the Xu’ AGB cannot highlight any advantages of VODCA L. For example, in Fig. 14, obvious inconsistencies can be found in the Amazonian and North American boreal forests. The authors might consider comparing this with existing L-band products to better illustrate its benefits.
6. In line 45 of the introduction, the authors overlooked the significant contribution of some French teams to the application of VOD in biomass studies. A more thorough review of existing research is needed.

Specific comments:

Abstract: The Spearman correlations between the vegetation proxies used by the authors and VOD were all below 0.6. Given this, how can one assert that there is 'good agreement' or that they 'agree well'?

Lines 79-80: As mentioned above, the authors need to provide a substantial basis for their claims.

Lines 93-95: It is critical to assess the spatial and temporal continuity of VODCA V2.

Table 1: It appears that not all sensors provide global coverage, which the authors should clarify. Additionally, which observation angle was used for SMOS in this study?

Line 46: Bin center 2.5-62.5°?

Line 47: Yes, but what strategies did the authors use to rigorously raise the issue of RFI in SMOS?

Line 191: Duplicated.

Methods: Flowcharts need to be used to describe the fusion process.

Line 295: Why?

Line 321: Further detailed results, such as plotting the time series, are needed to demonstrate the improvements achieved by using SSMI F17 as a reference.

Lines 338-339: For pixels that do not meet the specified condition, how is the AC (1) value determined, and how is the weighting assigned?

Lines 369-370: How did the authors conclude this? The effective scattering albedo seems to have a greater effect on VOD values than the surface roughness parameter.

Did Figure 4 convey any useful information?

Figure 5: Typos. Additionally, the discussion of AC(1) is overly technical for the results section and would be more appropriately detailed in the methodology.

Figure 6: The units don't seem to make sense.

Figure 7: There is no colorbar in this figure.

Why there is a negative correlation between VODCA CXKu and ASCAT slope in Fig. 8? The authors' explanation in line 457 lacks persuasiveness. Could the authors elaborate, particularly concerning the types of savannas in the Brazilian Amazon as depicted in Fig. 8c?

Line 471: need a reference, e.g., 10.1038/s41561-022-01087-x

Line 478: It is necessary to show RFI map to make this clear.

Figure 14: It's clearly inconsistent.

Figure A14: Wrong caption.

I will be happy to provide comments on the revised version.