An Improved Method for Rainfall Forecast Based on GNSS-PWV

Round 1

Reviewer 1 Report

GENERAL COMMENTS:

The article is written with a good English, but it would be useful to improve further through a complete rereading.

 

The description of the methods already available in the scientific community should be more detailed and discussed with reference to the adopted method.

 

In reference to the algorithms and methods widespread in the scientific community a general introduction and discussion should be given by highlighting the novelty aspects of this work. It seems that the proposed work is an application on a larger scale of already proposed methods and therefore, obviously, in this case the novelty character of the work is poor.

 

The paragraph: 2.2.4 Criteria for evaluation results should contain the description of the processing. For example an important parameter is the time window used for the prediction of rainfall, i.e the maximum temporal distance to which the observation of a change in PWV is referred for forecasting rainfall. The chosen window (12 hours) is not motivated. It should be an investigation  process to find the best window size. 

The procedure described in section 3.1 should be further investigated. As the Figure 6 suggest a strong correlation (mostly linear) between the PWV threshold and the number of detected rainfall event, a probability based approach should be investigated to understand if a probability of rainfall occurrence could be associated to PWV quantity, increment or derivative.

 

The work is only centered on indices such as POD, FAR etc… An analysis should be shown to find a quantitative relationship between the PWV related values and the amount of precipitation. 

 

The analysis of results (Section 5) suggest a further separation between different climatic areas with different  precipitation regimes. I would suggest to start from the analysis of single stations and aggregating those characterized by similar behaviour. How would the mean POD and FAR values change if introducing the climatic factor in the analysis? 

 

 

SPECIFIC COMMENTS

 

The introduction, up to the line 111, contain a list of work without the description of the main method that in bibliography used GNSS-PWV related data to predict rainfall in the next 6 hours.

 

It should be provided an introduction the methodologies adopted from the various introduced works (it is only mentioned that they are threshold based methods).

 

From line 117 starts the introduction of the method of the present work and it should be described the difference.

 

In one 131 the IGS is mentioned but, in paragraph 2.1.2 other dataset are mentioned. The authors should clarify what data have been used in this work.

 

A better description of ISD station data used in this work should be given, including the type of measurements used in this work.

 

Line 195 the total number of selected GNSS stations should be provided.

 

Some grammatical corrections should be included through a compete re-reading of the manuscript. Some examples:

Line 153: “the data is” should be corrected with “the data are”.

 

I am not completely convinced of formula  in Equation 1. This is referred to 

Li, H.; Wang, X.; Wu, S.; Zhang, K.; Chen, X.; Qiu, C.; Zhang, S.; Zhang, J.; Xie, M.; Li, L. Development of an improved 634 model for prediction of short-term heavy precipitation based on gnss-derived pwv. Remote Sensing. 2020, 12, 4101.

Where the PWV is expressed as a summation of terms referred to specific humidity, as confirmed also by:

Qin Zhang, Junhua Ye, Shuangcheng Zhang, Fei Han, "Precipitable Water Vapor Retrieval and Analysis by Multiple Data Sources: Ground-Based GNSS, Radio Occultation, Radiosonde, Microwave Satellite, and NWP Reanalysis Data", Journal of Sensors, vol. 2018, Article ID 3428303, 13 pages, 2018. https://doi.org/10.1155/2018/3428303

Alternatively the PWV could be expressed in function of mixing ratio observations as suggested by:

Tang X, Hancock CM, Xiang Z, Kong Y, Ligt H, Shi H, Quaye-Ballard JA. Precipitable Water Vapour Retrieval from GPS Precise Point Positioning and NCEP CFSv2 Dataset during Typhoon Events. Sensors (Basel). 2018 Nov 8;18(11):3831. doi: 10.3390/s18113831. PMID: 30413096; PMCID: PMC6263919.

 

However also the formulas referring to specific humidity or mixing ratio should be adopted and converted to process radiosounding observations should be directly expressed in terms of quantities measured by radiosounds (pressure, temperature, relative humidity, etc…) 

The author should express the PWV in function of quantities obtained from radio sounds.

 

The formula used in Equation 2 is wrong. Several expressions exist for the computation of ZHD. A correct formula could be found by consulting:

Samuel Osah, Akwasi A. Acheampong, Collins Fosu, Isaac Dadzie, "Evaluation of Zenith Tropospheric Delay Derived from Ray-Traced VMF3 Product over the West African Region Using GNSS Observations", Advances in Meteorology, vol. 2021, Article ID 8836806, 14 pages, 2021. https://doi.org/10.1155/2021/8836806

Please provide a reference for the adopted Formula.

 

Line 227 it should be specified how the height is computed from geopotetial height.

Line 228 It should be specified that Pd is computed as difference between P and Pw

 

Lines 247-252: The entire sentence should be rearranged. Moreover I have some doubts about the points described: 1) I think that a rainfall event should be discarded if there is not the possibility to identify correctly the start of the rainfall and if the measured rainfall quantity could be wrong. 2) The number of event may not be a good discriminator of poor precipitation. It is not clear to me why you need to have a large number of events on all stations, nor why a number of events less than 100 could lead to an analysis with worse results. The author should justify this and clarify the results of this work if all the possible stations would be included in the analysis.

 

Lines 254-261: The description of GNSS-PWV should be more detailed. It should be specified if the authors directly processed rind data, provide some information about the implementation of the software, listing the adopted choices, etc…

 

Lines 264-267: As the computation expressed in Equation 2 is wrong, all the work should be recomputed.

 

Line 269-270: Please provide a reference for the Equation 4.

 

Lines 304-305: These are very general consideration that should be accompanied by bibliographic references at least. However, it would be better to insert more sentences with more details, 

 

Lines 306-316: A reference to the most adopted thresholds, or some of them should be introduced, with some discussions about the novelty of the proposed methods, with motivation of the choices.

 

Lines 324-340: The threshold used for the identification of heavy rainfall should be provided and discussed/justified.

 

From Figure 3 it seems not possible to obtain thresholds really able to discriminate POD with respect to FAR, as their vales are very near and FAR is almost always greater than POD. This is a very critical aspect of adopting this method even though is not discussed in this paper. I would expect some consideration to  justify the acceptance of such high FAR or so low POD.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Please see the attachment.

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf