Quality Assessment of PROBA-V Surface Albedo V1 for the Continuity of the Copernicus Climate Change Service

Round 1

Reviewer 1 Report

This manuscript provides an extensive assessment to the quality of a remote sensing product, i.e., the PROBA-V surface albedo product, which is part of Essential Climate Variables in the Copernipus Climate Change Service portfolio. The assessment covers many aspects of the product, including completeness, precision, spatio-temporal consistency with similar products, and accuracy. And the authors also provide some comments to the status of current surface albedo product as Climate Data Records, which are valuable to all concerned users. This manuscript is worthy to be introduced to readers. However, some of the following parts can still be improved.

(1) The abstract is too long.

(2) In the Introduction part, parallel researches on other surface albedo products or their validation are barely mentioned.

(3) Section 2.4 is different from section 2.1~2.3. I think it should serve well as part of section 2.1.

(4) In page 7, line 267~269, this statement about black-sky albedo and its agreement with tower measurement is totally confusing.

(5) In page 11, line 387~388, in determining the intra-annual precision, did the authors consider the different temporal interval of the 3 kinds of products, i.e., 10 days for PROBA-V and SPOT/VGT, 1 day for MCD43. And the spatial resolution is also different.

(6) In page 13, line 465~466, in deriving the distribution of temporal length of missing values, how to process the high latitude areas where polar night results in extensive missing values.

(7) In Section 5.3, is it possible to provide some results about comparing the satellite albedo products with GBOV LP, instead of GBOV RM ? So that the reader will have an idea of how much is the role of scale discrepancy in the total uncertainty of direct validation.

(8) In page 22, line 693~696, the authors mentioned the different spectral region of broadband albedo, but did not provide the details of these spectral regions. In my opinion, I don't think the small difference in the spectral range of broadband albedo will result in any essential uncertainty.

(9) In page 22, line 697~699, the authors mentioned the temporal noise in NIR domain. But I cannot quite understand the logic of this sentence. Can the authors explain it more, or put it in other words?

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

This study evaluates the quality of the PROBA-V surface albedo product as the continuity product from AVHRR and SPOT VGT sensors. It follows the CEOS WGCV good practices, which also provides quantitative benchmark criteria for product assessment. Overall the study is well written, comprehensive, and follows an established assessment methodology. The study presented is a subset of a larger body of work evaluating the PROBA-V albedo product.

I have some questions about the methods relating to in-situ calibration sites and a few minor points for the authors to address, outlined below. I suggest care be paid to addressing these points before publication.

 

General comments

1. GBOV sites selection method inconsistency

The first site filtering criteria applied was based on a bias threshold of 20% between reference measurements and the land products [L285]. First, it was not entirely clear which specific land products were being assessed. Secondly, the threshold was applied to discard heterogeneous sites [L286]. Why then does a detailed analysis come after this threshold was applied specifically looking at spatial representativeness [s3.2]? There would be many other factors involved in the bias % than just heterogeneity.

The second threshold for selecting homogeneous sites based on the semivariogram and used a ST and RAW score (>2). It was not clear then if sites were included where leaf-off were lower than the threshold but leaf-on conditions were above the threshold, and vice versa. Or was it only the time periods where the thresholds were > 2? Also, if the aim is to filter based on heterogeneity I recommend to use a single metric, i.e. the semivariogram, and not the bias %. Otherwise a clearer justification is needed.

There was also no mention about uncertainty of the in-situ measurements apart from spatial scaling. Some instrument uncertainty specifications would be helpful to add, including absolute accuracy and drift as provided by the manufacturer.

I highly recommend adding a figure of 2-3 contrasting sites with different homogeneity including their scores and a map of the sites so that the reader can get a better appreciation of how the method works.

2. Dedicated cal/val in-situ measurements not discussed

There have been many efforts to measure spectral reflectance with highly calibrated spectrometers at dedicated cal/val sites, often in sync with satellite overpasses, yet the study contains none of this type of data. This was also not discussed anywhere, which is important as calibrated spectrometers are likely a better once-off method for validation than a tower mounted net radiometer. I recommend some comments are added addressing this point.

3. Study recommendations lacking

The main finding was the albedo products still fall well short of the ambitious WMO/CEOS accuracy criteria yet there was a distinct absence of recommendations to improve this. Is it likely the current products can improve with re-processing (i.e. those mentioned in the full Copernicus report)? What is recommended so that future satellite products may meet these requirements? These would be very helpful and insightful for readers and I suggest adding these.

 

Minor comments

L63: First time ECV mentioned in the body of the manuscript, I recommend expanding the acronym.

L364: A key missing point of the LANDVAL sites is they were selected for their high temporal stability. Please add this. It is also worth being explicit that no in-situ measurements were used at these sites for validation purposes, and they were purely used for product intercomparison.

L434: reference error to address ‘!Error!...’

L461: please clarify the meaning of ‘similar evolution’

L462: Recommend removing the word ‘Noteworthy’

L487: Change ‘les’ to ‘less’

L492: Change ‘slight large’ to ‘slightly larger’

L495: Suggested: ‘A seasonality effect was observed with the sign of the bias between…’

L516: suggested: ‘with slight’ to ‘with a small’

L629: The overestimation of PROBA albedo for some forests was never discussed. It would be insightful to know the reason why.

L660: Suggest to replace the qualitative ‘much more gaps’ with the quantitative values

L736: Suggest adding a quantitative value for ‘a representative amount of ground data’. I.e. how many sites, biomes, years, growing seasons etc. ‘Representative’ is broad and vague.

 

Well done on the quality piece of work!

Author Response

Please see the attachment

Author Response File: Author Response.pdf