Determining Tropical Cyclone Center and Rainband Size in Geostationary Satellite Imagery
Round 1
Reviewer 1 Report
This study investigated the tropical cyclone (TC) center and rainband size in geostationary satellite imagery (Advanced Baseline Imager (ABI)). Based on the brightness temperature (TB) observations, TC center, rainband inner radius (RIR) and rainband outer radius (ROR) can be obtained. The RIR describes the size of TC inner-core region, and the ROR reflects the maximum radial extent of TC rainbands. The authors suggested that the root-mean-square differences of ABI-determined centers for tropical storms and hurricanes are 45.35 and 29.06 km, compared with the best track data. Additionally, they explained the implication of the TB-determined TC center, RIR and ROR, and related these parameters with Rossby wavenumbers. In general, the manuscript is well written and results of this study is reasonable. The manuscript can be accepted for publication after some minor revisions.
Specific comments:
Abstract: They authors should clarify the TC samples where all these general statements are concluded from. Is this a case study or statistical analysis based on regular observation?
How many TCs have been included in this study? Are they typical enough to make the conclusions convincible?
Some necessary TC case description is needed in Abstract, and the authors need to examine the whole manuscript to avoid overreaching conclusion.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This study adopted the Advanced Baseline Imager (ABI) onboard the U. S. Geostationary Operational Environmental Satellite to determine tropical cyclone (TC) center positions and rainband sizes. It is undeniable that the center positioning and size measurement are important in operational practice. The variety of center-positioning methods and size identifications have been proposed in previous literatures. In this study, an azimuthal spectral analysis method is employed to obtain an azimuthal symmetric center of a TC. However, it is unclear whether this center positioning has an advantage over other methods. Besides, the inner and outer rainband radii are estimated based on radial gradients of TB observations at eight azimuthal angles. In fact, these radii seem to be of less practical merits in forecasting operation. Overall, the discussions in this study aren’t of enough scientific significance to be worth acceptance for publication. In the current form, I don’t think the quality of this study meets the publication criteria in this journal. The specific comments are listed as follows:
1. A number of TC center-positioning methods were provided in previous studies, including local extreme method, weighted grid points method, and the minimum azimuthal variance method. How well the center-positioning method proposed in this study perform by comparison with other methods should be discussed.
2. Hu and Zou ever tried to locate TC center by the same method using ATMS and MHS onboard POEs. In this study, they adopted the alternative GOES-16 for the same purpose. It also requires an evaluation of these two methods of TC center positioning.
3. In L211, “the one which produces the maximum value of the mean amplitude of wavenumber-0 component”. Here, “maximum” or “minimum”? In general, the TB amplitude is small corresponding to rainband convection.
4. In L235, “the average TB of the surrounding environment of a TC”. This surrounding environment refers to the average TB between 800-1000 km away from the center. How sensitive are the results to the annulus selection?
5. In Fig. 3, it’s not surprising that the average wavenumber-0 amplitude corresponding to the ABI-determined center is the highest among those from three different centers, because this method does find the maximum average of TB. Therefore, I think it makes no sense to make this comparison. It is natural that the ABI-determined center is much closer to the best track than the first-guess one due to the physically-different definition. The first-guess position is the minimum value of TB observations, generally corresponding to the eyewall location. Therefore, it is not necessary to show Fig. 5. I suggest deleting it. The same case is Fig. 7.
6. Among all the tryout centers, the one which produces the maximum value of the mean amplitude of wavenumber-0 component (symmetric component) is taken as the TC center. Therefore, the relative importance of the azimuthal wavenumbers 0-3 amplitudes with the ABI-determined center and best track can be foreseen as shown in Figs. 12 and 14. Namely, the wavenumber-0 amplitude centered on the ABI-determined center is larger than that centered on the best track center, while it is opposite for other wavenumbers. Nonetheless, it doesn’t necessarily conclude that the method of ABI-determined center is superior to other center-positioning methods.
7. Physically, rainbands are characterized by spiral and narrow filamented convection. The criterion exerted in eight azimuthal angles fail to comprehensively reflect the overall rainband pattern. To my opinion, the identification of RIR and ROR based on eight azimuthal angles can’t reasonably represent rainband size.
8. I am not clear what is the point to identify RIR and ROR. In fact, the radii of inner- and outer-rainbands have no relationship with other commonly-used operational size parameters, such R17 or ROCI. These operational parameters are more practical for disaster prediction.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Review of remotesensing-1750842-v1
Manuscript Title: Determining Tropical Cyclone Center and Rainband Size in Geostationary Satellite Imagery
Authors: Yanyang Hu and Xiaolei Zou
The manuscript describes and apply a methodology developed to determine the tropical cyclone (TC) center from the IR channel (10.3 μm) of the ABI radiometer on board of GOES-16/17. The method was already developed for the ATMS data and now has been transferred to the ABI in order to exploit its higher spatial and temporal resolution. In addition to the TC center, the radius of the inner and outer TC bands are also provided. The method is based on the azimuthal spectral analysis and its output is compared with the TC best track method developed by the National Hurricane Center (NHC). The results, even if based on the analysis of only two hurricanes, show a good agreement in the TC center location, which evidences a slight dependence on the TC strength. Furthermore, the mean amplitude of wavenumber-0 is always higher than that provided by the NHC best track.
The manuscript is well organized and written. The quality of some figures has to be improved and the following specific comments have to be address to make the manuscript suitable of publication on the Remote Sensing journal.
Lines 91-92: the sentence is incorrect. The IR TB measurements have a good relationship with the top of the cloud, while more difficult is the estimation of the precipitation intensity from IR TB (e.g. multi-layers clouds are a limitation for IR detection as well as shallow precipitation, which are better detected by MW). Please correct the sentence.
Figure 1: please modify the position of the label Irma within the figure. From your description at lines 178-180, it is clear the Irma occurred southern than Jose.
Figure 2b and lines 223-254: basically, the two radii in the NW-SE direction determined by your method correspond to the red open circles (which are difficult to see on the plot, I suggest to make it bigger or the line width bold). If I am correct, I suggest to report the values in a table or somewhere within the text, this could significantly help the reader.
Lines 372-373: this seems to be the trend, but you need to apply your method to a larger dataset in order that this sentence could be confirmed or not.
Tables S1 and S2 are very useful. In my opinion, a plot showing the evolution of the best track center and the ABI-determined center should be added. You could replace Figure 6a and 8c, respectively which are quite complicated to read (i.e. it is basically impossible associate each blue and black circle to the corresponding TC center).
Lines 407-412: for Jose, the trend of RIR and ROCI is quite different, since, from September 12th, ROCI slowly but constantly increases, while RIR as a sharp increase between September 12th and 13th and then constantly decreases. Could you interpreter and explain this difference?
Lines 412-414: why do you say “more obviously”? it could be obvious for you but not for a reader.
Lines 453-455: you cannot state this for sure. What you see (more or less for all six time stamp even if you should specify to which one you are referring) is a TB depression, from which you cannot distinguish between convective core and anvil (for example).
Figure 12a: it has to be improved. It is quite impossible to distinguish between open magenta triangle and magenta marker.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
