Evaluation and Calibration of MODIS Near-Infrared Precipitable Water Vapor over China Using GNSS Observations and ERA-5 Reanalysis Dataset
Round 1
Reviewer 1 Report
The paper is a good contribution to the study of PWV, by means of the comparison of the accuracy of the MODIS near-infrared PWV products with some other well know techniques.
Author Response
Comment #1:
The paper is a good contribution to the study of PWV, by means of the comparison of the accuracy of the MODIS near-infrared PWV products with some other well know techniques.
Response: Thanks for your comment.
Reviewer 2 Report
The manuscript «Evaluation and calibration of MODIS near-infrared precipitable water vapor over China using GNSS observations and ERA-5 reanalysis dataset» is completed scientific work. Authors describe in detail the features of the spatial distribution of PWV and good comparison results are shown. We think the results may be used partially for choosing a site for the future mm/submm telescope.
-I think that correlation coefficient should be rounded to two decimal places (for example, line 346, GNSS-PWV over China. The correlation coefficients between RS-PWV and ERA-PWV are in a range from 0.9 to 0.99, the mean correlation is 0.97.). Also authors should use not term correlations but correlation coefficients (for example, line 346).
-Authors should add doi to the references. For example, https://doi.org/10.1029/2009RG000301 for 2 reference source, https://doi.org/10.1175/1520-0442(1997)010<2643:AWVOC>2.0.CO;2 for 4 reference source, https://doi.org/10.1029/2001JD001302 for 9 reference source, https://doi.org/10.1256/wea.126.02A for 11 reference source, https://doi.org/10.1002/joc.3412 for 13 reference source, 10.1109/RMC50626.2020.9312233 for 14 reference source and etc.
- Figures 8, 9,10,12, 13 correspond to the spatial distributions of quantities. Titles of figures should be in a consistent format (recomendation).
Author Response
-I think that correlation coefficient should be rounded to two decimal places (for example, line 346, GNSS-PWV over China. The correlation coefficients between RS-PWV and ERA-PWV are in a range from 0.9 to 0.99, the mean correlation is 0.97.). Also authors should use not term correlations but correlation coefficients (for example, line 346).
Response: Thanks for your comments.
(1) The three decimal places of the correlation coefficients in Line 346 is for consistency with that in other places (e.g. Lines 309, 392, 436).
(2) “correlations” was replaced by “correlation coefficients”, as advised.
-Authors should add doi to the references.
Response: Doi was added in the reference.
- Figures 8, 9,10,12, 13 correspond to the spatial distributions of quantities. Titles of figures should be in a consistent format (recomendation).
Response: Figures 8, 12 13 correspond to the spatial distributions of quality indices in the whole study period, the titles of which are in a consistent format. Figures 9 and 10 correspond to spatial distributions in different seasons, so their titles are different from Figures 8, 12, 13.
Reviewer 3 Report
Review On "
Evaluation and calibration of MODIS near-infrared precipita-2 ble water vapor over China using GNSS observations and ERA-3 5 reanalysis dataset "
In this work, the authors attempted to evaluate and improve the MODIS PWV products in the near-infrared band using reliable GNSS PWV estimates and ERA5 data over China. Results of such studies can be attractive to different users of the MODIS water vapor in the study area.
The paper shows a nice attempt in this field but however there are important points which must be revised or addressed.
Given that you have received the necessary reviews; I will present my main point of view about this research.
Recently, an article was published with the following title
Bai, J., Lou, Y., Zhang, W., Zhou, Y., Zhang, Z., & Shi, C. (2021). Assessment and calibration of MODIS precipitable water vapor products based on GPS network over China. Atmospheric Research, 254, 105504.
, which is very similar in terms of subject and study area. My suggestion is that the respected authors clearly state the difference between their work and their achievements with the Bai et al, 2021.
I was going to complete my reviewing report on this article in the next few days, but to speed up the publication process, I presented my general opinion.
Author Response
Response: Thanks and the article was added in the reference.
Bai evaluated the performance of MODIS PWV based on GNSS and ERA5 data over China, and presented a linear calibration model for MODIS PWV using the GNSS PWV as reference data. The main differences of this study from that of Bai are in the selection of sample data and the method used to construct the calibration model.
Bai adopted a linear regression model developed based on the sample data of GNSS PWV over the whole study area (i.e. China) for calibrating MODIS PWV at each MODIS swath epoch. The model contains fewer parameters than our model. However, the relationship between GNSS PWV and MODIS PWV in different geographical locations are different, thus the performance of Bai’s calibration model may perform differently in different regions, especially in the regions where a few GNSS stations are deployed (see the correlation coefficients in Figure 6 (c)). A test based on GNSS PWV at three epochs showed the RMS of the differences between MODIS PWV and GNSS PWV reduced by around 30%.
Our grid-based calibration model was constructed based on a harmonic model fitting the differences between MODIS PWV and ERA5 PWV at each grid point (rather than a region). This is the reason for the grid-based model to perform better in the calibration of MODIS PWV (53% reduction in RMS).
Round 2
Reviewer 3 Report
I suggest that the authors enter their response into the final manuscript
in Introduction and methodology sections
Author Response
I suggest that the authors enter their response into the final manuscript in Introduction and methodology sections.
Response: thanks for your suggestion. The description about the methodology was added in the methodology section (see Line 474-483). However, Bai's model used GNSS PWV over China as the sample data, so we don't think it's suitable to detailedly describe Bai's model in the Introduction section, but it was still added in the reference (see reference [54]).
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
Line 49 - It is no clear the reference [8] for precipitable water vapor (PWV) definition.
Line 86-89 - Not easy to understand. Is it a tittle?
Line 234 - Maybe the used NWP model could be name.
Line 279 - Consider if units in figure caption could be introduced.
Author Response
Line 49 – It is no clear the reference [8] for precipitable water vapor (PWV) definition.
Response: Two new references have been added.
- Barreto, A.; Cuevas, E.; Damiri, B.; Romero, P.M.; Almansa, F. Column water vapor determination in night period with a lunar photometer prototype. Atmos. Meas. Tech. 2013, 6, 2159–2167, doi:10.5194/amt-6-2159-2013.
- Bevis, M.; Businger, S.; Herring, T.A.; Rocken, C.; Anthes, R.A.; Ware, R.H. GPS meteorology: remote sensing of atmospheric water vapor using the global positioning system. J. Geophys. Res. 1992, 97, 15787, doi:10.1029/92jd01517.
Line 86-89 - Not easy to understand. Is it a tittle?
Response: The sentence was amended (Line 93–97). “Previous performance evaluation of the MODIS PWV products over several typical regions in China showed that MODIS PWV presented a typical error in the range from 5 mm to 12mm, referring to the GNSS PWV products from the International GNSS Service (IGS), radiosonde data from the Integrated Global Radiosonde Archive (IGRA), and sunphotometer data from the Aerosol Robotic Network (AERONET)”.
Line 234 - Maybe the used NWP model could be name.
Response: Amended as advised (Line 256–257). “the only way is to use assimilated values from the reanalysis dataset from NWP models (e.g. ERA-Interim and ERA5) for the conversion.”
Line 279 - Consider if units in figure caption could be introduced.
Response: Amended as advised.
Reviewer 2 Report
Interesting article on the evaluation of one of the MODIS Terra satellite products. The article is innovative and important for the study of the troposphere and climate change. The results of the assessment are shown in many figures (Fig. 8 is of poor quality). Good agreement of the MODIS product with reference sources proves its high quality (after calibration).
Comment for Authors: In equation 14, the argument of the cosine function for the annual and semi-annual components are the same - they should be different.
Author Response
Comment for Authors: In equation 14, the argument of the cosine function for the annual and semi-annual components are the same - they should be different
Response: Thanks for you suggestion. The equation was amended as suggested (Line 506).
Reviewer 3 Report
Comments:
Line 42: Is this compare to oceans volume??? The authors need a another simile, because water in the ocean is not relevant to PWV.
Line 45: Water vapor is not the most important greenhouse gas, how about CO2?
Line 51: Please add examples and citations
Line 71: In recent years? MODIS TERRA was launched in 1999.
Line 74: is the most widely used? I thought it was microwave radiometry (for example AMSR) because microwave measurements are not influenced by clouds.
Line 86: This sentence is missing something, Preliminary performance evaluation of MODIS products .... shows that .... (for example)
Line 125: The authors are essentially forcing MODIS to behave like ERA5, no surprise there is an improvement, this is the weakest point in the paper.
Also, are they no radiosondes over China since 1999 when MODIS TERRA was launched? If they are, why not use them as another dataset to compare, for example to evaluate ERA5 without using its meterological values to compute GNSS PWV …
Why are the authors only evaluating 2013 -2018? Why not 1999 to 2018?
Why are the authors only evaluating MODIS TERRA, why not use MODIS AQUA as well.
Line 144: the infrared is available during day and night.
Line 144: the NIR uses solar reflected NIR radiation from the ground clouds and glint, hence only available by day... It seems the authors are confusing the retrievals. Please check. Also the NIR is barely used, it is preferred to use the IR one.
Line 146: the Authors just said that the NIR has poorer accuracy than the IR retrievals. Please double check and add citations.
Line 172: Please show a figure showing the impact of this correction, i.e., the difference between the ERA 5 PWV uncorrected and the corrected one. Considering there is no much water above the tropopause (i.e. 100hPa) I do not expect this make a big difference.
Figure 1: What is the insert?
Figure 2: Please also show the climatological 2013- 2018 values for ERA-5 to get an idea of the actual PWV expected values. Also show the standard deviation to get an idea of the variability of PWV over China. Units of colorbars? is this mm?
Remove insert, It is not readeable.
Line 284: What do the authors mean by this? The ERA5 resolution and the MODIS resolution are not the same, the authors need to downsample (interpolate) the ERA5 to the MODIS center grid point to the comparison.
Figure 4: Remove insert
Figure 5: please unify the colorbars ranges in all figures so that the reader can compare directly among them. For example, the biases figures should have a colorbar encompassing all values show in all biases figures (for example, -4 to 14), etc.
Line 404-407: What is this?
Line 418: I thought they were 246 stations
Line 423: What do the authors mean by this? The ERA5 resolution and the MODIS resolution are not the same, the authors need to downsample (interpolate) the ERA5 to the MODIS center grid point to the comparison.
Figure 8 has really poor quality, please improve
Author Response
See attached pdf.
Author Response File: 
Author Response.pdf
Reviewer 4 Report
The manuscript is devoted to evaluation and calibration of MODIS (satellite) data using satellites as well as ERA-5 database which is result of assimilation and partially based also satellite. In other words, the data were calibrated using only atmospheric model data from ERA-5 and Global Navigation satellite systems. There are a lot of papers on comparing atmospheric models and satellite observations, for example [3,4,5,6] Still, it is necessary to use the data of the radiosoundes for the assessment and calibration of PWV. This is the first significant drawback of the manuscript.
Moreover, the study of the spatial distribution and dynamics of PWV is essential for operating and future ground-based telescopes working in mm / submm ranges. Today there are a number of projects of such telescopes for their placement in China and possibly in Russia and Asia [1]. The authors do not talk about this, i.e. the relevance of the data presented has not been sufficiently discussed. The introduction should be improved, the results of other scientists studying PWV over China should be considered, for example:
[1] Khaikin V. et al. On the Eurasian SumMillimeter Telescopes Project // 2020 7 th All-Russian Microwave Conference, 2020. DOI: 10.1109 / RMC50626.2020.9312233.
[2] Gong S. et al. Analysis on precipitable water vapor over the Tibetan Plateau Using Feng Yun-3A Medium Resolution Spectral Imager Products / Journal 'of Sensors DOI: 10.1155 / 2019/6078591.
[3] Wang S. et al. Evaluation of precipitable water vapor from five reanalysis products with ground-based GNSS Observations / Remote sensing. 2020.
[4] Wang Y. et al. Evaluation of precipitable water vapor from four satellite products and four reanalysis datasets against GPS Measurements on the Southern Tibetan Plateau. Journal of Climate. 30 (15). 2017.
[5] Evaluation of satellite and reanalysis precipitable water vapor data sets against radiosonde observations in central Asia. Earth and Space Science 2019.
[6] GNSS-derived PWV and comparison with radiosonde and ECMWF Era-Interium data over mainland China / JASTP. 182 (2019) 85 - 92.
These references should be taken into account.
The manuscript does not show the spatial distribution of PWV over the analyzed region. This essential aspect of PWV research has not been discussed. Of course, the authors set the goal of evaluation and calibration of data. But to evaluate and calibrate PWV authors use an accepted and well-known statistical apparatus. In our opinion, the academic value of the work is not great. Unfortunately, we must recommend the manuscript for major correction. A prerequisite is the use of calibration and PWV estimation by applying radiosonde data. It is desirable that the authors give the distributions of the PWV value over the region and answer the following question:
i) The ERA-5 database was updated recently (spring 2021). Did you use the data after correction?
ii) Due to the nature of the measurements, the ERA-5 can significantly overestimate the data in rough terrain conditions. In statistical analysis, did you use data without preliminary processing (without taking into account the relief, the absolute height of nodes)?
Thus, the authors should more clearly describe the relevance of the work in comparison with the available research and respond in detail to significant comments.
Author Response
See attached pdf.
Author Response File: Author Response.pdf
