Implementation of a Rainfall Normalization Module for GSMaP Microwave Imagers and Sounders
Round 1
Reviewer 1 Report
In this paper the authors developed the Method of Microwave rainfall Normalization (MMN) for mitigating the discrepancy in GSMaP rainfall estimates among passive microwave imagers and sounders. Such discrepancy is mainly due to differences in sensor specifications and retrieval algorithms. The MMN can be inserted inside the GSMaP algorithm between the gridding and PMW merging processes. By testing its performance on seven months of matchup data, the authors obtained good results in both rainfall amount and bias for all the PMW sensors and for both ocean e land regions. The main strengths of the MMN are: the correction of not only the GMI-observation region but also polar regions, and the correction of the CDF to utilize rainfall amount. On the contrary, this method has not been validated yet with ground dataset and cannot correct for no-rain/rain errors.
I think this paper meets the objectives of Remote Sensing Journal. It is clear from the authors results, that this methodology can improve the reliability and the applicability of the GSMaP.
In the following I provide some minor suggestions:
1. In section 3.1 I believe that some Equations could help the reader in understanding the methodology (especially at point 3).
2. In section 3.3 the authors introduce a “Normalization table”. However, this table is a figure (Figure 3). Furthermore, the mentioned Figure 3 is difficult to read, in terms of the chosen colours and style. As an example, related on Figure3(b), the authors at line 143 reported a distortion of 8mm/h for several MWSs.
I believe that the differences between the different PMW sensors are not appreciable.
3. At line 179 the authors introduce a second comparative analysis using all the observation samples. At line 181 they speak about “horizontal distributions”, and it is also repeated in the Figure 6 caption, but it is not clear what they refer with the term horizontal.
Author Response
Response to Reviewer 1 Comments
In this paper the authors developed the Method of Microwave rainfall Normalization (MMN) for mitigating the discrepancy in GSMaP rainfall estimates among passive microwave imagers and sounders. Such discrepancy is mainly due to differences in sensor specifications and retrieval algorithms. The MMN can be inserted inside the GSMaP algorithm between the gridding and PMW merging processes. By testing its performance on seven months of matchup data, the authors obtained good results in both rainfall amount and bias for all the PMW sensors and for both ocean e land regions. The main strengths of the MMN are: the correction of not only the GMI-observation region but also polar regions, and the correction of the CDF to utilize rainfall amount. On the contrary, this method has not been validated yet with ground dataset and cannot correct for no-rain/rain errors.
I think this paper meets the objectives of Remote Sensing Journal. It is clear from the authors results, that this methodology can improve the reliability and the applicability of the GSMaP.
>> We kindly thank you for your helpful comments and suggestions for our manuscript. This article is intended as a preliminary report as the analysis is based on the final test version from a short period of time. We plan to validate and to improve this method for the next step. We add the sentence in Conclusion.
In the following I provide some minor suggestions:
- In section 3.1 I believe that some Equations could help the reader in understanding the methodology (especially at point 3)
>> Thank you for your suggestion. Indeed, it may have been difficult for readers to understand only the description in text. We added Equation (1) to introduce corrected rainfall intensity from original one through the correction table.
- In section 3.3 the authors introduce a “Normalization table”. However, this table is a figure (Figure 3). Furthermore, the mentioned Figure 3 is difficult to read, in terms of the chosen colours and style. As an example, related on Figure3(b), the authors at line 143 reported a distortion of 8mm/h for several MWSs.
I believe that the differences between the different PMW sensors are not appreciable.
>> Thank you for your comment. We replaced Figure 3 to relationship of rain rate before the MMN correction (X-axis) and rain rate difference from corrected to uncorrected (Y-axis) the MMN correction. We believe the differences between the different PMW sensors become clearer.
- At line 179 the authors introduce a second comparative analysis using all the observation samples. At line 181 they speak about “horizontal distributions”, and it is also repeated in the Figure 6 caption, but it is not clear what they refer with the term horizontal
>> We just showing you the distributions of the bias on the map, and the term “horizontal” is unclear as you pointed out. Therefore, the term “horizontal” is deleted in text and figure.
Reviewer 2 Report
In this manuscript, the authors discuss the new method, method of microwave rainfall normalization (MMN) that has been used in the recent version of GSMaP. The manuscript is well written and is critical to understanding GSMaP. GSMaP is one of the widely used Level-3 precipitation products, and this manuscript will definitely interest the users of GSMaP and readers of Remote Sensing.
I have only minor suggestions:
- Lines:33-37 lack supporting reference.
- Lines 87-91: Is there a reason for using linear interpolation above 99 percentiles? If so, it would be nice to include that here.
- Line 111 and Figure 2: Why are mid-latitudes (55-50) and not (50-55)? This is vice-versa for the tropics. If there is no particular reason, please be consistent
- Figure 4 - MTA_MHS: At -70 degrees latitude, there is some spurious rain.
Author Response
Response to Reviewer 2 Comments
In this manuscript, the authors discuss the new method, method of microwave rainfall normalization (MMN) that has been used in the recent version of GSMaP. The manuscript is well written and is critical to understanding GSMaP. GSMaP is one of the widely used Level-3 precipitation products, and this manuscript will definitely interest the users of GSMaP and readers of Remote Sensing.
I have only minor suggestions:
>> We kindly thank you for your helpful comments and suggestions for our manuscript.
Lines:33-37 lack supporting reference.
>> We added the reference [7].
Lines 87-91: Is there a reason for using linear interpolation above 99 percentiles? If so, it would be nice to include that here.
>> Heavy rain may cause the correction to be unstable due to the small sample size. To reduce this unstable condition, linear interpolation was applied. We included this reason.
Line 111 and Figure 2: Why are mid-latitudes (55-50) and not (50-55)? This is vice-versa for the tropics. If there is no particular reason, please be consistent
>> We corrected from (55-50) to (50-55).
Figure 4 - MTA_MHS: At -70 degrees latitude, there is some spurious rain.
>> This spurious rain is also detected in COR_GMI (black line in Fig. 4) and appeared at 50W (Fig. 6b). This is misdetection of sea ice as a heavy rainfall. The difference appears to have increased in this latitudinal band because of the upward correction.
Reviewer 3 Report
The paper is certainly innovative and interesting because it deals with an issue that is of great relevance and usefulness to anyone dealing with climate and related topics. In the introduction, the issue needs to be better expressed by citing literature that highlights errors in precipitation estimation by the GPM constellation. for example:
Tan, J., Petersen, W. A., & Tokay, A. (2016). A novel approach to identify sources of errors in IMERG for GPM ground validation. Journal of Hydrometeorology, 17(9), 2477-2491.
Gentilucci, M., Barbieri, M., & Pambianchi, G. (2022). Reliability of the IMERG product through reference rain gauges in Central Italy. Atmospheric Research, 278, 106340.
Ning, S., Song, F., Udmale, P., Jin, J., Thapa, B. R., & Ishidaira, H. (2017). Error analysis and evaluation of the latest GSMap and IMERG precipitation products over Eastern China. Advances in Meteorology, 2017.
In addition, you should expand the literature
I suggest that you change the structure of the paper by dividing it into Introduction, methods, results, discussion, conclusion, the sub-sections you freely choose.
Why did you not use standardized model performance evaluation indexes? This could help to compare the performance of the model with other different models tested in the literature.
Place emphasis on the results obtained and the reason why your method turns out to be better.
Author Response
Response to Reviewer 3 Comments
The paper is certainly innovative and interesting because it deals with an issue that is of great relevance and usefulness to anyone dealing with climate and related topics.
>> We kindly thank you for your helpful comments and suggestions for our manuscript.
In the introduction, the issue needs to be better expressed by citing literature that highlights errors in precipitation estimation by the GPM constellation. for example: …
>> Thank you for your suggestion. We add some sentences about the highlights errors in precipitation estimation by the GPM constellation in the Introduction.
In addition, you should expand the literature
I suggest that you change the structure of the paper by dividing it into Introduction, methods, results, discussion, conclusion, the sub-sections you freely choose.
>> Thank you for your suggestion. We reconsider the structure of section and subsections as you suggested.
|
Before |
After |
|
1. Introduction
2. Data 3. MMN algorithm 3.1 Overview of the algorithm
3.2 Differences in CDF among PMW sensors 3.3 Normalization table 4. Evaluation of MMN method 5. Summary and discussions |
1. Introduction 2. Methods 2.1 Data
2.2 MMN algorithm 3. Results 3.1 Differences in CDF among PMW sensors 3.2 Normalization table 3.3 Evaluation of MMN method 4. Discussion 5. Conclusion |
Why did you not use standardized model performance evaluation indexes? This could help to compare the performance of the model with other different models tested in the literature.
>> This article is intended as a preliminary report as the analysis is based on the final test version from a short period of time. Some evaluation indexes are variated due to an insufficient number of samples, even a seven-month accumulation. Further evaluation, including standardized model performance evaluation indexes, will be a topic for future work, as long-term processing is still underway. We added this sentence in Conclusion.
Place emphasis on the results obtained and the reason why your method turns out to be better.
>> Thank you for your suggestion. As you commented, we reconsider the chapter layout, we divided the advantages and challenges of our method into the Discussion, and the summary of results into Conclusion for easier understanding.
