Chinese Nationwide Earthquake Early Warning System and Its Performance in the 2022 Lushan M6.1 Earthquake
Round 1
Reviewer 1 Report
The paper of Peng et al., deal with a very important and delicate topic: the building and evaluation of an Earthquake Early Warning System for China. Such a system, as properly mentioned by the authors, can be very useful in mitigating the effects of such devasting natural hazards in terms of human life and economic damage. Furthermore, this is very important for countries like China that have been affected by devasting seismic events mentioned by authors like Tangshan and Wenchuan in the last decades. So I found the topic of very high interest not only limited to the scientific community but also to decision-makers and even to the population. Despite this, high interest, the paper requires several improvements before it can be eventually accepted for publication. A general flavour of this paper is that the authors need to have in mind that they are writing a scientific paper and not a “flyier” to present or even sell to other country the developed system. Often the paper contains “glorious” affirmations (like we installed 15000 stations...) that are out of the scientific style required by the Remote Sensing journal. I fully understand the “propositive spirit” of the authors to develop an useful system to save population and I admire and congratulate for this spirit, but in the paper, you need to be more “objective” and overall add several scientific bases that are completely or partially lack in the present version. One of the most crucial is “how the earthquake parameters are estimated”? Please, see my further comments below for other points that I warmly suggest improving. Overall, you don’t need to convince me that you have developed a good Early Warning System, even if we desire so. In fact in a scientific paper you objectively present what you have developed, you evaluate the performance on a case of Lushan 2022 earthquake that is good, and finally, you discuss the performance proposing a path of future development of the present system. Partially you have done so, but partially “blinding” some aspects that need to be discussed more. One is that you have a very good estimation of earthquake parameters at the 5th warning issued after 16.5 seconds but at this time the seismic wave reached about a distance of 60 km from the hypocenter, so the warning is not so much useless as at this distance the effect of a M6.1 earthquake are anyway low. For this purpose, I suggest discussing more maybe with an additional dedicated graph (or table), the estimated seismic Intensity at several distance compared with the effective one, within the information of leading time. This information is partially contained in your 4 graphs of Figure 5. For example at station SC.AL001 you estimated a seismic intensity of about 4.7 5.7 s after the EQ and about 2 seconds after the warning the real SI reached this value, but unfortunately, in the following seconds the SI overpassed 6 and the updated estimations of SI occurs in the same time as the shake hit the place, so they are ineffective (you state this in the text, so this is not a critic but I would like you spend more time/deep discussion to analyse this aspect). In particular, you need to divide the regions for SI observed and compared with predicted and when this have been predicted. At this purpose a map with the position of seismic wave at the five warning could be useful, or even better you can produce 5 maps with the estimated SI compared with the position of seismic wave, as what is inside is not-effective and what is outside can be still usefull for a warning.
General and important points:
l In the first part of the introduction, you rightly present the urbanization of China and statistic about populations in big towns (lines 41-43). I would like the authors to add some considerations related to the scientific topic. I mean, among the 17 cities with more than 10 million populations, how many are in high seismic hazard risk zones? The same is for the 90 cities with more than one million people. You can extract such information from the Seismic Hazard Map of China.
l A general question is about the infrastructure. I am not sure to have understood, but the “Shanghai Tower Co., Ltd” is located in Shanghai or is it a distributed server network (cloud servers)? That’s a crucial point because if you transmit all the data, for example, from Sichuan to Shanghai, you need to cover a distance of about 2000 or 3000 km and send it back to the provincial centre I suppose in Chengdu, you need to transmit data for 4000 / 6000 km, and you lost some precious time. Can you clarify this aspect better to me and include a more detailed description in the paper, please?.
l In Figure 2 and lines 178-188 you describe the first decision-making module as Level-2 and final decision-makin module Level 1. Even if your description is coherent among the paper, normally the Level 1 is the first stage and Level 2 is the second processing, why did you invert? Please check, in case correct, but keep in agreegment with already published material on this network to avoid future “dangerous” confusion
l Line 190. Did you send an alert to the population if an earthqauke of M2+ is detected? This would create an “over” alert, as earthquakes smaller than 3 or 4 don’t produce any demage and if you send too much alert even if they are correct could create an opposite effect in the population. I mean if a person is alerted 10 times in a year for M=2.0 and then the persone receives an alert for M=5.5 maybe this person don’t take precaution because the previous earthquakes didn’t create any trouble, but an M5.5 could create severius damage, and not all the population can maybe distinguish the danger basing on the magnitude of the allert... So in my opinion, is strongly discouraged to alert for such low magnitude.
l In lines 211-213 you say that you deployed about 2300 systems to notify the early warning to critical buildings like schools, hospital, etc... did you provide a free app that anyone can download and receive alert, or even better than app (that requires internet connection) would be to send an SMS to all the phone interested by a sort of alarm. Such service of SMS alarm by phone is active in California for example.
l Often in this paper, the algorithm description is demandended to other papers. Please add at least a sketch of any of them. I refer, for example, to line 164 (the protocol for EEWS), lines 172-174 (how they detect such information just on p-waves?) to line 217 (FinDer) and so on...
l Magnitude type: In the paper, I don’t find any consideration about the type of magnitude you are considering/estimating. For example, the Chinese earthquake catalogue uses the MS magnitude that is typically higher than moment magnitude Mw. For example, the earthquake that you propose as an example has an Mw estimated by USGS of 5.8 which is closer to the one the Early Warning system finally estimated (Fig. 4 a and c). Is it due to the fact that you are estimating Mw and comparing with Ms? Please revise and improve the paper considering the magnitude type.
l At line 381 you properly mention the “SI prediction equation”. I would like you spend more sentences explaining which is this equation. Did you use attenuation relationships like Ambraseys, 1995?
o Ambraseys, N.N. (1995), The prediction of earthquake peak ground acceleration in Europe. Earthquake Engng. Struct. Dyn., 24: 467-490. https://doi.org/10.1002/eqe.4290240402
l Lines 394-396. I agree with this sentence and it’s in accordance with Figure 7. Anyway the goal of the Early warning is to reduce as much as possible the “blind zone” and produce alerts for sites with moderate-high intensity to have any usefulness. I fully understand that is impossible to make a null blind zone with an early warning system, because if you are at the epicenter the same time the sensor receives the seismic wave also the people/buildings will receive and computational and transmission time would even produce the alert with a delay. My request is just to discuss this point more.
l Figure 5. I think you don’t explain what is the pink/light orange band. I think it’s the leading time. Anyway if so, the end of the band is often wrong, in fact at SC.AL001 red line (oserved seismic intensity) overpassed the threshold (blu dashed horizontal line) 1 or 2 second before. The same is in the case of SC.A8306, maybe 3 seconds before. For SC.A2902 the band needs to be a little longer as the red line reached the threshold at around 20 s after origin time. SC.TY002 looks okay for me.
l Are you sure that you need depth estimation for an Early warning system? I mean that this parameter is not necessary to take action to mitigate/save population or critical structure, we just need location, time, and magnitude (or seismic intensity).
Specific points:
l Lines 19-21. “Firstly, ... information” This sentence is not proper for the abstract. You can explain the structure of your pare in the last part of the introduction, but not in the abstract. If you want you can replace the sentence with an overview on how the EEWS works or simply delete it.
l Line 24. I think it’s better in English “it was created 5.7 s after its occurrence” without “at”. The same at line 25.
l Line 27. I don’t understand what do you mean by “blind” in this sentence. Probably I can understand reading the rest of the paper, but the abstract must be clear without read all the paper.
l Line 136. Sorry can you explain this sentence better? What do you mean that you increased the number of cellular towers, so you not only installed several seismic sensors but also improved the coverage of 3G/4G/5G? Or do you mean that by installing the sensor at the cellular tower, you don’t need additional data connections that would reduce the bandwidth of the present cellular network? Furthermore, are you sure 3G in China is still working?
l Title of section 2.2. Please correct the typo in “Structure”.
l Line 145. What is it exactly FSU link? Is it the one used by cellular networks to route the normal call/data of mobile devices between the towers?
l Line 157. I think it’s better “due to” instead of “owing”
l Lines 156-158. As this is a scientific paper you need to explain how do these centers protect the network? Solar panels? Battery backup? How about for a possible collapse of the hosting building?
l Line 162. I suggest substituting “are set” with “acquire”.
l Line 164. “reduced” is not the right word if you do not introduce it before the data packet length. As I suppose to understand what you mean, I suggest replacing “can be reduced” with “can be as small as“
l Line 174. If you write “more than three stations” it means that 3 stations not trigger an alarm but 4 or more, yes. If three are sufficient (I think so) I suggest you to replace “at least three stations”
l Line 219. I think the authors mean “results at the same time, because”
l Lines 223-232. Sorry I don’t understand if any strategy contatin the previous one or they are independent. So when in line 234 you state that use strategy 3, does it measn you are also using strategies 1 and 2. If so , I would suggest to call criterium and not strategy, as the strategy is then the combination of several criteria, in your case from 1 to 3.
l Lines 244-245. Sorry I don’t understand this sentence. How can be a mean of maximum and minimum? I mean any station provide one estimation of magnitude and than you used maximum if their range is between 1.0, for example 5.1, 5.0, 5.4, 5.3, 5.4 and you select the maximum in this case 5.4. If the range exeede 1.0 you get the mean I guss. I have another question, please, why not the median that could be less affected by eventual outlier if one or more stations could be influenced by external factors?
l Line 247. Please add the origin time of the earthquake that is fundamental for the following evaluations of the performance of the system. It would be better all the time in the same format (Universal Time or Beijing time = UT+8).
l Line 251. I suggested writing “was estimated” because the determination of Seismic Intensity be seismic station is an indirect way. The right one is from the damages caused to the structures.
l Line 264-265. I don’t understand this sentence, because I didn’t understand the method description for this part. Please rewrite better. Do you estimate as magnitude 3.6 – 4.7 when it actually was 6.1?
l Line 276, In the caption please correct in “estimated magnitude” as correctly typed in axis label of Fig3b.
l Line 400. I think it is better to say “entered in the trial operation”
l Line 403 I suggest to add the year for the earthquake overall to distinguish from the one of 2013, So “recent Lushan 2022 M6.1 earthquake.”
l Line 405. I think you mean “ensuring”
l Line 418. What is the length of time window? The one inside you measrured the shake from the sensor? Please specify better.
l Line 435. If you consider a speed of 3.5km/s and 5.7 seconds of delay the seismic wave would have travelled for s = v t = 3.5 km/s 5.7 s = 19.95 km, so about 20 km, not 10 km. Please check.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
Earthquake monitoring and early warning is an effective method of earthquake disaster prevention and mitigation, and it is also a technology that many countries pay attention to. This manuscript provides the latest information on the monitoring capabilities of the China Seismic Network and introduces the China National Earthquake Early Warning System. The performance capability of the system in the 2022 Lushan 6.1 earthquake is introduced. The specific suggestions are as follows: (1) It is suggested to revise Figure 1. The three types of stations in Figure 1 cannot be well distinguished on the figure; (2) Can more cases be provided to analyze the performance of the system; (3) The discussion part should be separated from the conclusion. Also, these two parts need to be reorganized and rewritten. It is recommended that the author analyze the performance of this study and previous work, or the performance of other countries' networks from all aspects of the work of this paper. In addition, future work needs to be looked forward to. Therefore, my suggestion is a moderate revision.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Dear Authors,
I consider the manuscript addresses a cutting-edge topic. The paper is very consistent and well structured. However, before its acceptance, some issues should be undertaken:
-I think a native-English speaker must review the whole document since I have found some mistakes related to the preposition and article usage. Besides, the text "we presents" (a significant error) appears twice in the manuscript as well as some typos such as "Strcuture" (line 138). In style writing, spell out the number: "five" better than "5" when the number is [0-9] (l 239), and on similar occasions. In line 299, in "low magnitude estimates" a hyphen is missing ("low-magnitude estimates"), and so forth.
-Regarding the references, in general, major works are cited, but still, I have some concerns:
* In lines 38-40, the reference (from 2013) should be replaced with a newer one, or removed.
* When you write on seismic networks, you should cite some very recent papers related to other countries (such as Spain or Iran) strong motion networks.
*In lines 421-4222, "...the method proposed by Collombelli et al. [35] or Peng et al. [36]..." rather than "...the method proposed by [35, 36]..."
- In line 271, "no depth error", sounds unreal (maybe the numbers are the same, but the uncertainty cannot be null).
Kind regards and congrats on the work.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Dear authors,
Thank you so much for your detailed reply. I’m satisfied with your reply and the revised version of the manuscript that I think now is suitable to be published in Remote Sensing. Thank you also to have discussed and take into account my questions and suggestions. Reading the new version, I just found some last points that I would like you could revise, please.
Thank you!
Specific points:
l Lines 47. I think it’s better to say “intensity of degree 7 or above.”
l Line 172. I think it’s better to say “fails for example for a possible collapse”
l Line 203. I would suggest adding the first family name: “proposed by Horiuchi et al. [34]”
l Line 264. I tried to search the application on a Chinese app store but searching with English words, I didn’t’ find this app. Nevertheless, I put on automatic translation the 3 words (Emergency earthquake system), and it gives me “紧急地震信息” and with this I found, installed the app and see that it works, showing the last earthquake of today or two days ago and even more, distance to me and magnitude and some other information, but it’s all in Chinese, and I don’t understand sorry. Anyway, as I think your paper could also be useful in some way to present the system, I think you can add the Chinese name of the app in brackets so the reader of the paper can search and find the app if they want (as I did).
l Line 276. I think you mean “peak” not “pack”
l Line 474. I think it is better “as” instead of “because”
l Line 484. I think the word “deviation” is not appropriate. It’s okay if you mean the “standard deviation” but in this case, please write “standard deviation”, otherwise you can say “uncertainness”
l Line 490. If it is often large then how can be smaller than 0.5? Do you mean larger than 0.5?
l Line 507. I think it’s better say “select the depth as a fixed value” and not “select the optimal depth as the fixed value” this is because it’s not optimal if it’s fixed...
l Line 513. Please remove “widely”. Considering that earthquake depth generally is from few kilometers to about 700 km from 5km to 20 km is a small range not wide.
l Lines 543-545. This sentence is not well constructed in English. I try to suggest you a way: “Not considering the fault finiteness, would cause serious ground-motion underprediction for destructive earthquakes with magnitude greater than 7.0.”
l Lines 623-630. Please as this section is “Conclusion” transform all the verb tenses to past and not present. (e.g., Firstly, we introduced the seismic network...)
l Line 632 I think is right “estimation” instead of “estimate”.
l Line 634. This is not the “earthquake” but the “earthquake early warning system”
Author Response
Please see the attachment.
Author Response File: Author Response.pdf