Description of the UCAR/CU Soil Moisture Product
Round 1
Reviewer 1 Report
Manuscript ID: remotesensing-788986
Title: Description of the UCAR/CU Soil Moisture Product
Clara Chew and Eric Small
Special Issue: Applications of GNSS Reflectometry for Earth Observation
The paper describes a dataset that provides soil moisture retrieval from the CYGNSS mission using GNSS-Reflectometry. The manuscript is very clear and well explained with a rigorous analysis. The paper is innovative since it provides a new large dataset the UCAR/CU CYGNSS Soil moisture product available since 2017.
I only have minor comments to finalize the manuscript before publication. In addition the manuscript is very appropriate for publication in the Remote sensing special issue entitled “Applications of GNSS Reflectometry for Earth Observation”.
Minor comments:
P2, line 43: in the references [4–7] you could refer as well to the GLORI airborne campaign https://www.mdpi.com/2072-4292/10/8/1245.
P12, line 346: Please detail quickly those different networks or add some references.
A mention of the PBO H2O GNSS soil moisture derived product could be useful as well for a reader not familiar with those products.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 2 Report
The paper describes an analysis of CyGNSS data for the retrieval of soil moisture. A rather long time series analysis has been proposed and calibrated using SMAP data. The paper is interesting and well written, it reflects the well-established experience of the authors on this specific topic.
Before considering the paper for publication in this journal, I would suggest performing some important revisions, mainly to increase and highlight the scientific content and novelty of the work.
The literature survey could be improved, some recent papers strictly connected to this analysis may have been missed.
Some inputs to improve the paper are given below:
To my knowledge, GPS satellites transmit Right polarized signal. The Left polarization is usually considered in reflection since it is stronger than the R one, at least up to 70 deg incidence. Please, revise the relevant sentence.
Please, make sure to clearly highlight the scientific novelty of the work with respect to refs. 10 and 19, as well as with respect to the following work
Clarizia, Maria Paola, et al. "Analysis of CyGNSS data for soil moisture retrieval." IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 12.7 (2019): 2227-2235.
As concerns the role of the topographic variations mentioned at pag. 3, please consider that this aspect has been recently discussed in the following papers, also considering the sensitivity to soil moisture. I would consider important a connection of your analysis with this works.
Dente, Laura, et al. "Space-borne GNSS-R signal over a complex topography: modelling and validation." IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing (2020).
Campbell, James D., et al. "Modeling the effects of topography on delay-Doppler maps." IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing (2020).
A connected study about the role of the topography, of the spatial resolution and of the inhomogeneities has been also described in the following works. Please, consider connecting them to your analysis.
Comite et al. "Bistatic Coherent Scattering From Rough Soils With Application to GNSS Reflectometry." IEEE Transactions on Geoscience and Remote Sensing 58.1 (2019): 612-625.
Gleason et al., “Geolocation, Calibration and Surface Resolution of CYGNSS GNSS-R Land Observations,” Remote Sensing, 2020
As concerns the following expression:
‘Using SMAP data for calibration of course comes with many drawbacks, the major one being that SMAP soil moisture retrievals are not actual soil moisture observations and have their own error and uncertainties.’
I would consider understood that SMAP values are only estimation and not ground-truth data, therefore I would not use the word ‘actual’. Just an opinion, anyway.
I agree that \Gamma retrieved from (1) is an ‘effective reflectivity’, and I would keep this definition throughout the paper. It is better to not only talk about reflectivity, which is not rigorously the case. This would also help to make smoother the difference with respect to the definition used by the authors in ref. 19, where they consider the scattering essentially incoherent.
I am just wondering why the authors have not consider the following map/reference for water bodies, which looks a bit more recent and perhaps more updated than ref. 22
Allen, George H., and Tamlin M. Pavelsky. "Global extent of rivers and streams." Science 361.6402 (2018): 585-588.
I would suggest using a dynamic range for figures 10 from 0 to 0.3, this would help better appreciating the agreement.
Have the authors compared their results with the time series proposed in ref. 19 and/or by Clarizia et al. 2019?
I really appreciated the discussion given in section 4. It helps the reader to get a big picture of the open issues. I am wondering why the authors, here, have not considered the technique used in ref. 19 to detect fully coherent reflections. Also, a possible way to improve the results could be to set a threshold to eliminate strong reflections (above a certain valued to be estimated from data). It is understood that water bodies provide the highest values of the effective (or equivalent) reflectivity for this technique.
Finally, it is not completely clear to me if the calibration of the L2 product must be regularly ‘refreshed’ or it can be assumed established by initially relying on certain set of training data. This is an important aspect to clarify and to understand: CyGNSS data will always need SMAP data to be calibrated? I understand anyway the temporal resolution is improved with this approach.
Personally, I think one the biggest source of mismatching is the signal fluctuation due to inhomogeneities and the loss of coherence (i.e., decorrelation) of the scattered signal.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 3 Report
Please see attachment.
Comments for author File: Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 2 Report
Thank you for considering my comments.
Author Response
Thank you for reviewing the paper!
Reviewer 3 Report
Thank you very much detailed response of my comments and revision. Most of comments have been clarified or addressed in the text. I have a few more (minor) suggestions before publication.
- Readers may misinterpret median ubMRSE of 0.049 cm3/cm3 in the abstract unless the standard deviation of ubMRSEs and median correlation coefficient (with its standard deviation) are provided even though the standard deviation of ubMRSEs is provided in Table 1 in the revised manuscript. My suggestion is (1) add standard deviation of ubMRSEs and median correlation coefficient (with its standard deviation) to the abstract and (2) add median correlation coefficient (with its standard deviation) to Table 1.
- It is kind of missed in my first review in relations to median ubMRSE calculations and my suggestion above. If those ubRMSE values with the sites (values <0.1 cm3/cm3 [as mentioned in quality flag 2]) that do not vary significantly are lot, your median ubMRSE will be naturally small since the dynamic range will be small. It is much important to show the performance against sites with higher soil moisture variability. I understand the rationale of keeping the long table, but I still suggest that you also provide histogram of ubMRSEs to make the interpretation of the results much more obvious.
- Page 1 Line 38. Please provide references for mentioned “several ongoing efforts”.
- Figure 11. Yes, I agree that CYGNSS retrievals will capture soil moisture variations around the time frame where it rains since it has higher temporal resolution. Figure 11 demonstrate this unique capability very well. However, this is the only contribution (unique aspect) of this product with respect to SMAP product. As a result, I would have expected to see more analysis on this aspect of the product (perhaps keep the time interval for entire 2018 (or any another year)). In any case, it would help reader better appreciate the value of this study if the contribution of this product is stated explicitly in the abstract.
-Figure 9. It seems like most of the ISMN sites are located in the CONUS. There are two in Australia, one in Africa, and two in South America as I see in this figure. Isn’t it fair to say that the validation is performed over CONUS for a global product? It is hard to answer, but the question will be how well the ISMN sites in CONUS represent the rest of the CYGNSS coverage. I think that it would better bring this to users’s attention upfront.
Author Response
Please see the attachment.
Author Response File: Author Response.docx