Performance of GLASS and MODIS Satellite Albedo Products in Diagnosing Albedo Variations during Different Time Scales and Special Weather Conditions in the Tibetan Plateau

Round 1

Reviewer 1 Report

The authors spent a lot of efforts to improve their manuscript. I have some minor comments, which should be addressed before publishing the paper.

1. Title: exchange “MCD43B3 and MCD43A3” by “MODIS”; suggestion: “Performance of GLASS and MODIS satellite albedo products in diagnosing albedo variations during different time scales and special weather conditions in the Tibetan Plateau”

Abstract:

2. L23: “… in the eastern TP”; add: “at two site, Maduo and Maqu”
3. L24: mention MODIS here
4. L26: “with the observation”; specify “ground-based observations”

Introduction:

5. L49: explain the acronym “MODIS” here not in line 59/60
6. The motivation of this paper could be stressed a little more by giving the required accuracy of surface albedo estimates for proper climate modeling and other applications. What do we aim for? à check Henderson-Sellers and Wilson, 1983; Jacob and Olioso, 2005; Sellers et al., 1995
7. L65: “.. better correlated with”; give numbers
8. L80/81: “…while GLASS and MODIS underestimated.”; underestimated by what?
9. L95: “weather process” à “ weather situation”
10. L95: “It was based that the albedo changes…”; change wording
11. L98: start new sentence after “albedo changes”
12. L98: “then we will try to induced..”; wording, further it suggests, that the authors want to actively perform modeling on atmospheric circulation; please rewrite
13. L98: “into the modeling”; Which kind of modeling?
14. Introduce also the surface measurements in your outline

Data and Methods:

15. L115: “Angular Bin (AB)”; don’t use capital letters, other you use them everywhere when explaining acronyms
16. L122: use “VIS” and “NIR” instead of “vis” and “nir”
17. Give also the spatial resolution for MCD43A3 in Sec. 2.1.2
18. L142: “To make the observations …”; simplify the wording, something like that: “To select representative measurement sites ..”
19. L148: à footprint of satellite data
20. Figure caption: it’s not really a land cover map, moreover b) and c) show photographs of the surrounding
21. L166: CNR1 is a net radiometer for longwave and shortwave radiation measurements. Checking the website of Kipp & Zonen, I think the instrument has actually 4 sensors, two pyranometers and two pyrgeometers, while CM3 sensors are pyranometers. However, both instruments give the albedo. How do they compare? What is the measurement uncertainty? Probably it is in the range of your final bias between satellite and ground-based observation. Comment on that. Give also the range of uncertainty in Table 2.
22. L173/174: “Deviations of more ..”; I don’t understand the procedure here.
23. Table 2: Remove the last two lines / collector and manufacturer; It’s the data logger, which you list here. The reader doesn’t care about that. Adjust CNR1 parameters, since it has two separate sensors included (shortwave and longwave)
24. L180: “Its vegetation ranging from 0.05 to 1.10 m …” change wording
25. L186/187: Same with CNR1, it is composed of a pyranometer and a pyrgeometer. Give instrumental uncertainty.
26. L218: “proportion of diffuse illumination… ”; add “r” here
27. (1): Where does this equation come from?

Results:

28. Discussing the results in this section, please always specify which albedo you mean (BSA, WSA, blue-sky). Don’t just write “GLASS albedo” as in L240/241.
29. L233/234: “The annual change ..is small..0.16-0.67..”; I wouldn’t call It small.
30. L286: “mutations”; change wording
31. 5,6: Specify axis labels: “Surface Albedo” à Satellite-based, “Observation” à ground-based
32. L319: “The MCD43B3 albedo has errors when the SZA exceeds …”; It has also errors for SZA < 70° à change wording
33. L323: “mean albedo values”; which albedo (BSA, WSA, ?)
34. L326/327: “.. at the low value of the diagonal”; change wording
35. L337: “.. few uncertainties and defects”; change wording
36. L338: “beating larger”; change wording
37. Fig 9.,10.: maybe use the name of the month for the x-labels (1-1 à Jan 1)
38. L392: Title à snow is also a kind of precipitation but not meant here; adjust the title
39. L404: “The cloud on the TP changes …”; change wording

 

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper described validation of the applicability of albedo products of GLASS and MODIS at two observation stations in TP. The manuscript is much improved compared to the manuscript submitted before. Therefore, this manuscript has a merit for the publication after miner revisions.

L365 (Figure 9) Caption was wrong.

L298: Regarding references of 52-53 for example, authors should refer certain refereed journal/book if possible as follows:

Wiscombe, W. J. and S. G. Warren, A model of spectral albedo of snow I: Pure snow, Journal of the Atmospheric Science, 37, 2712-2737, 1980

Warren, S. G., Optical properties of snow, Review of Geophysics and Space Physics, 20, 67-89, 1982

L382: "Snow albedo on the TP is closely related to snow depth, and the snow is thin." --> "Snow albedo on the TP is closely related to snow depth because the snow is thin." The albedo of a thin snowpack obviously depends on the albedo of the underlying surface because optical thickness (depth) of snow is not enough. Note that the optical thickness of snow depends on both snow depth and snow density, implying that snow albedo depends on the both snow depth and snow density. Authors should modify the explanation correctly.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

The authors used two sites in the Tibetan Plateau to validate GLASS, MCD43A3, and MCD43B3 albedo products.  I cannot recommend publication.  The paper is poorly organized, especially the methods, missing important information needed to judge the validity of the results, replete with figures whose relevance barely discussed, omits fundamental measures of agreement (e.g., Mean Absolute Deviation; MAD) and misinterprets results.  Some examples are discussed below.

1) Grammar and syntax need substantial improvement.  The paper needs a professional English-language editor.

L142 – The 3-steps site process was not explained sufficiently; a) how were remotely sensed data, camera data, and the DEM used to identify uniform sites; b) what was the goal of data variance and consistency tests and exactly how were the implemented; 3) how was representativeness defined and what was the role of the footprint in the definition? Moreover, the fundamental characteristic for site selection, homogeneity of soil and vegetation cover was not discussed including the scale (e.g.,  2-x-2 km at which it was measured

L180 – The reader never learns the area of the field of view of the pyranometers.  The variation in vegetation cover seemed quite high at the Maduo site and therefore one is left one wondering if homogeneity was realized.

L193 – Subsection title does not mention GLASS

L195 – What was the rationale for averaging the two pixels and how was the direction of the neighboring pixel determined?

L206 – Rationale for temporal alignment of GLASS and MCD43B3 was unclear (not discussed) since they were not being compared to each other

L213 – The paragraph discusses why estimatation of BSA was not done (missing data on aerosol optical depths), but the next section discusses estimation of BSA for MCD43A3.  This was confusing.  It is possible to estimate BSA for MCD43A3 but not MCD43B3 and GLASS?

L233 – This basic descripting result is not stated correctly.  The range is not annual an average because it covers the entire period.  Further, one may rightly argue that the range is actually large since it is 50% of the maximum theoretical range.

L243 – Taylor plots are complicated, take some time to decipher, and will not be familiar to all readers.  However, there is only passing reference to them.  I would say they are not needed if they’re not going to be used to interpret the results.  Despite the wealth of information they contain, they do not provide any information on bias (e.g., table 2).  They seem to be included for the “wow” effect.

L261 – Maduo description of results is not consistent with the same for Maqu.

Figures 5 & 7 – Why is there more than one symbol per plot? How should one interpret a blue plus sign inside a black circle?  These plots should be satellite albedo on the x-axis and ground observed albedo on the y-axis (i.e., how well does satellite estimates predict ground observations) with one symbol.  Axes should be clearly and unambiguously label (i.e., both satellite and ground are observations).  Axes should be clearly labeled.  Figure 5 caption does not indicate that it is an 8-day period.  It should.

L422/L509 – The conclusions are counter to the results presented.  There is no relationship between soil moisture and albedo in Figure 11A. There were no goodness-of-fit or other model results for the regression equations (presumably) in the figure.  A significant slope in 11A would, if it exists would be attributable to a large number of observations.  The scatter in 11B suggests that many other factors contribute to albedo variance at Maduo.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

Review on: Title: Verification and evaluation of the applicability of albedo products of GLASS, MODIS, and GlobAlbedo under the alpine meadow over the Tibetan Plateau by An et al.

The authors compared satellite albedo products with ground-based observations on two sites for different temporal regimes. Several statistical quantities were given to evaluate the different products. In general, the manuscript is really limited to the comparison of data without deeper discussion.  The conclusion, that satellite products have shortcomings for short-term and local changes of the surface albedo (e.g., due to snow events), is well known already. Moreover, the choice of only two sites limits the significance of this study. In the following, I will give some more specific concerns to publish the manuscript in its present shape.  Since I suggest to reject the manuscript for publication, I will give only the major points to think about.

1. The authors already published a similar study at Plateau Meteorology (An et al., 2019). Since this publication is written in Chinese, I can only judge from the English abstract. Is there any new content of the current manuscript compared to the other publication?
2. The authors motivate their study by the fact, that multi-year comparisons of albedo products are necessary, which I fully agree with. However, the choice of the satellite data sets is kind of weird. Tab. 1 and Fig. 2 show the temporal restriction of the GlobAlbedo product compared to the other data sets. Why did the authors use this product at all, since it comprises a significantly shorter period than the others? All temporal averaged measures, as the seasonal and monthly means (Figs. 3, 7, 8, 9, ..) should be cleaned for data which don’t comprise the same temporal period. For a strict comparison only data matching in time and space should be used.
3. Talking about comparability the authors have to show, that the spatial inhomogeneity in the FOV of the satellite sensors pointing on the two sites can be neglected. Otherwise the comparison is kind of unfair. As an example, Wu et al. (2018) also compared satellite and ground-based observations for 13 sites but discussed the spatial heterogeneity first.
4. From Fig. 2 we learn, that the ground-based observations in Maqu are higher than the MODIS observation, while in Maduo the data sets match quiet well. What is the reason behind? I miss a discussion on the regional differences. Figure 1 doesn’t help to understand the regional differences. A land cover map showing spatial structures would be more helpful. There are high-resolution images (e.g., 30 m Landsat) available, I think. From the coordinates I estimated, that Maduo is in close vicinity of a lake (several 100 m), which suggests a higher spatial inhomogeneity within the satellite footprint (see comment 3).
5. There are too many figures and too less discussion on the results. Simply, listing numbers is not worthwhile for a publication. For example, Fig. 3 and Fig. 4 could be comprised by giving an additional set of columns for the mean RMSE and bias in Fig. 4. Fig. 6 is not needed. Fig. 10 shows similar numbers as the two figures before.
6. Comparing BSA and WSA with the surface albedo from the observations is the easy way for an evaluation. However, several former publications determined the blue-sky-albedo for a fairer comparison. I know, there are some more efforts to calculate the blue-sky-albedo, but since the data set is restricted to just two sites, it should be done.
7. The in-situ measurement were not introduced adequately. Which instruments were used? What are their specifications including measurement uncertainty? How is the quality assurance of the data set realized?
8. I’m wondering if a more extensive comparison of surface albedo products would give a more significant conclusion on the individual uncertainties in the Tibetan Plateau. Specifically, I think of a large scale comparison of BSA and WSA (instead of comparing with only two sites). Also other products should be taken into account. How about using also combined product as for example MuSyQ (Wen et al., 2017)? Such an evaluation will certainly reveal more general shortcomings of the individual products on a regional base than the current study does.

Cited literature:  

AN Yingying, MENG Xianhong, ZHAO Lin, LI Zhaoguo, LÜ Shihua, MA Yutang. Evaluation the Applicability of Albedo Products of GLASS, MODIS and GlobAlbedo under the Alpine Meadow over the Qinghai-Tibetan Plateau. Plateau Meteorology, 2019, 38(1): 88-100.

2017. Wen et al., "Forward a Small-Timescale BRDF/Albedo by Multisensor Combined BRDF Inversion Model," in IEEE Transactions on Geoscience and Remote Sensing, vol. 55, no. 2, pp. 683-697, Feb. 2017.

Wu, X.; Wen, J.; Xiao, Q.; You, D.; Dou, B.; Lin, X.; Hueni, A. Accuracy Assessment on MODIS (V006), GLASS and MuSyQ Land-Surface Albedo Products: A Case Study in the Heihe River Basin, China. Remote Sens. 2018, 10, 2045.

Reviewer 2 Report

This paper described validation of the applicability of albedo products of GLASS, MODIS, and GlobAlbedo at two observation stations in TP. It is important to examine how these albedo products provide accurate results through in-situ measurements for the purpose of the long-term monitoring of surface albedo in TP. Practically, I think all products provided should be validated by the product organization before it is released to the general public. In that sense, this paper has a merit of the publication in RS. However, I feel that some points should be clarified and discussed before publication.

1. It is natural that if there is a rare snow/rain event in this region where snow doesn’t usually fall, such effects will be included in 16-day composite data or monthly data at a certain rate. In fact, surface albedo widely changed especially in autumn and winter seasons (Figs. 8 and 9). In such cases, validation results in the fall and winter season make no sense. Instead, it should be validated using (many) scene data with different sites. In the first place, results in winter season couldn’t be reliable for the reasons: the issues on the current atmospheric correction algorithms (L240) and the high solar zenith angle in the MODIS product (L174). Thus, such data should be removed from the validation or carefully reviewed and used. To summarize, I will recommend that authors should validate carefully each albedo product using data selected (Fig. 5).
2. In Figs. 3, there is a systematic bias error of each albedo product at the Maqu station. (1) Are there any possibilities that the BRDF (kernel) model used in each product was not suitable for the grassland vegetation in the Maqu station? (2) This large bias can cause a large bias (error) in estimating energy and water cycle processes on the TP. The authors need to call attention to this issue in this paper. If the authors cannot use the suitable BRDF model in the albedo retrieval, the authors need to consider how to calibrate the albedo products which have a merit for the user.
3. There is no information about the instrument employed in the in-situ measurement. Please describe the instrument employed here for the validation, including its accuracy, calibration and so on in details.
4. Please add explanations for which months correspond to the season authors defined.