Precipitation Estimation from the NASA TROPICS Mission: Initial Retrievals and Validation

Round 1

Reviewer 1 Report

This is a straightforward manuscript, outlining the TROPICS mission, the methodology (i.e. the Precipitation Retrieval and Profiling Scheme PRPS) and a preliminary evaluation against both the GPM IMERG precipitaton product and instantaneous MRMS precipitation rates over the continental US.  

Overall, this mission is exciting, taking another step forward in developing remote sensing precipitation products.  As discussed, the TROPICS mission offers the potential of deploying a relatively cheap but robust constellation.  It's great to learn of these initial results. 

My criticisms are minor in nature.   Ultimately, I feel this manuscript was a bit rushed, lacking some finer detail.  I suspect that people with limited experience in satellite retrievals will be challenged.  Many Figure and Table captions need to be improved.  

Specific comments:  Ensure the initial definition of abbreviations/acronyms are included:  NE(DELTA)T, ATMS, Tb, MHS, MWHS-2 GMI, ME, SAPHIR, GEO-IR are all examples. (At least in my read through this.). 

It was a bit redundant, discussing the 12 channels on each cubesat. (Lines 69-72; 90-102; 139-143.) 

Figure 2 could be improved substantially with spatial scales.  Ideally it would be great to extend the image with a visible image and the IMERG precipitation.  

The text refers to the 183.31 GHz channel, but the table and the figure detail a 184.41 channel with the other systems operating at 183.31 GHz.  

Table 1: For the ATMS column, the 183.31 GHz is repeated 3 times with increasing range to meet the TMS frequency.  This should be better explained.  Also, is there an entry missing on the GMI column for channel 9? 

The summary of the PRPS retrieval scheme is, in my opinion, the opposite of the goldilocks principle: it's just wrong.  (Sorry.) It either needs to be more complete or more condensed, leaving the full details in the references provided.  

Figure 3.  How much data / observations comprise every point?  

Figure 5.  So much of the observations come from light precipitation. It's impossible to appreciate the saturation in this figure.  Is it possible to plot on using logarithmic scales to better appreciate the small values.  The lines will become curved, but that's easy to appreciate.  

Table 2:  The caption could be improved.  

Figure 6.  Here again, the caption needs to be improved.  You should explicitly state the four case studies here.  

I recommend that you take a moment to make this work more accessible to a wider audience with less expertise than yourselves.  

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 2 Report

This paper presents the preliminary results of TROPICS, a new mission for satellite estimates of precipitation. The authors describe the sensors utilized in this mission as well as the algorithm that will be used to estimate precipitation in a thorough and easy-to-understand manner. It is of great scientific interest and also quite tempting, since the observing system is intended to be low-cost. My overall recommendation is to be accepted for publication as is, after necessary checks for possible typos or mistakes.

For example, on page 4, line 114, “184.31GHz” should probably be “184.41GHz”.   

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 3 Report

General comments:

The research study titled “Precipitation estimation from the NASA TROPICS mission: initial retrievals and validation” by Kidd et al. encompasses (1) the description of the TROPICS mission and its millimeter-wave sounder instrument (TMS), (2) an overview of the algorithm and input data used to obtain precipitation rates, and (3) an initial validation of the retrievals obtained from the Pathfinder satellite compared against IMERG and MRMS. In summary, the TROPICS mission has the advantage of having retrievals at a high revisiting frequency (i.e. median of 50 minutes) due the use of a constellation of satellites at low-orbit, improving temporal resolution. The initial retrievals showed good agreement with IMERG rain intensities and with the spatial distribution of instantaneous surface radar retrievals from MRMS. The research study is of interest to the scientific community and provides insights about the upcoming satellite products for estimating precipitation. However, it is not clear how the retrievals are within 25% of the IMERG product. A robust explanation of how this 25% threshold is met and improvements to Figure 5 are needed. 

 

Major comments:

-    In Line 277 the authors specify that the retrievals should be within 25% of IMERG. It is not clear in the text how this +-25% limit is calculated. The description in Figure 5 and the plots provide some insights but an explicit description is needed.
-    For example, Figure 5 (line 305) shows the upper and lower limits. For an IMERG value of 2.0 mm/h the lower limit shows a value of 1.5 mm/h (i.e. a 0.5 mm/h or -25%) but the upper limit shows a value greater than 2.6 mm/h (a little more than +25%) which shows a skew of the limits to larger values. This also seems to suggest that the difference is larger than +25%. If the computation of the +-25% limit lines is done differently, it would be ideal to have it clearly stated to avoid ambiguity.
-    Figure 5 can include a metric of the percentage of points that fall within the +-25% limits. It seems a considerable number of points fall outside the limits, especially for precipitation rates smaller than 0.5 mm/h. If a large number of points are outside the limits, how the 25% condition is being met?
-    Why is the best-fit line forced through origin? The best-fit line (y = mx +b) might not go through the origin (0,0) and can be misleading. The best linear fit (not forced through origin) gives a better idea of the biases or differences between data sets. Additional statistical metrics can be also added such as NRMSE, mean bias, or R-Squared.

Minor comments:

-    Several acronyms and variables are not defined the first time they appear with the manuscript (e.g. ATMS, SPHIR, Tb, Tbs). The readers can figure out or find out most of them, but it would be ideal to have them properly defined in the manuscript.
-    The slope of the best-fit line (Figure 5) in both plots is less than one, meaning that PRPS-TROPICS was slightly underpredicting precipitation (in respect of IMERG). Figure 4 shows a few clusters of these values (red) in areas such as Northern Mexico, California, and Texas. Are there initial conjectures of in which areas/climates/env. conditions the retrievals are better or worst than in others?
-    Can a summary or short description of the output data product (level-1 or Level-2 if applicable) be included in the manuscript? It is not required but if the description is available, it can help future/potential users.

Author Response

Please see the attachment

Author Response File: Author Response.docx