the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 31 Mar 2025
Escalating typhoon risks in Shanghai amid shifting tracks driven by urbanization and sea surface temperature warming
Abstract. Tropical cyclones (TC) in the western North Pacific, known as typhoons, cause significant socioeconomic damage in East and Southeast Asian countries. The city of Shanghai in China is highly vulnerable to TC damages. For example, Typhoons Bebinca and Pulasan recently (September 2024) struck the city, resulting in widespread flooding, power outages, and the evacuation of more than half a million residents, while also breaking local rainfall records. Despite these threats, there is limited knowledge about the variability and mechanism of TC activities in this region under climate change and urbanization. Here, we use the Weather Research and Forecasting (WRF) convection-permitting model to simulate five TC events that made landfall along the southeastern coast of China and severely impacted Shanghai between 2018 and 2022. Different scenarios are conducted, including considering a future expansion of Shanghai's urban area and increases in sea surface temperature (SST) by 1, 2, and 3 °C. The results indicate that while SST warming significantly shifts TC tracks away from the city, the local risk continues to increase due to substantial enhancement of rainfall intensity and wind velocity. Warmer SST increases air temperature and decreases sea level pressure, thereby facilitating the formation and development of TC sizes and their dynamic systems. Furthermore, we find a consistent southward shift of the TC tracks that can be linked to the Fujiwhara effect, a phenomenon that occurs when two typhoons interact, causing a mutual counterclockwise rotation. Compared to SST changes, urbanization has limited influence on TC tracks and structures. The increase in surface roughness due to urban expansion reduces wind velocity but enhances the rainfall intensity within Shanghai, further exacerbating local risk. These findings could improve our understanding of typhoon variability under the combined effects of urbanization and climate change, as well as the risks they pose to Shanghai and other megacities in TC-prone regions.
- Preprint
(19418 KB) - Metadata XML
-
Supplement
(6353 KB) - BibTeX
- EndNote
Status: open (extended)
-
RC1: 'Comment on egusphere-2025-1002', Anonymous Referee #1, 21 Jun 2025
reply
General Comments
This study investigates the potential impacts of urbanization and rising SSTs on tropical cyclones impacting Shanghai based on convection-permitting WRF simulations of five tropical cyclones. The authors find a significant impact from increasing SSTs, including consistent increases to cyclone radius, maximum rainfall rate, and 10-m wind speed over the cyclone area that are more variable over Shanghai itself, as well as a southward shift in the cyclone track. They also find that urbanization has only a small impact on tropical cyclones, with almost no effect on large-scale cyclone characteristics but a slight increase in rainfall and decrease in wind speed over Shanghai itself. The latter half of the paper also includes an analysis of potential mechanisms, concluding that increased SSTs enhance lower tropospheric temperatures, wind speeds, and pressure anomalies, which increase cyclone size and intensity, and produce a southward shift of the cyclone tracks through enhancement of the Fujiwhara effect.
The results are clear, well-presented, and compelling. The paper is also beautifully written and was a genuine pleasure to read. The authors did a good job of evaluating potential cyclone impacts across multiple metrics, distinguishing between large-scale changes and those affecting Shanghai in particular, and proposing and demonstrating plausible mechanisms for the effects they observed. My biggest substantive comments have to do with the figures, some of which I find to be insufficiently explained and which use color schemes that are confusing and even, in some cases, deceptive. Most of my remaining comments are looking for additional clarification on certain choices that were made in the methods or analysis rather than objections to those choices. Overall, I am recommending minor revisions.
Specific Comments
Event selection – time period. On line 86, why focus on the period from 2018-2022 for selecting TCs? Why not, for example, the five most destructive TCs impacting Shanghai in the last twenty or even thirty years instead?
Event selection – trajectory. You specified that the TCs under consideration had to directly impact Shanghai (line 87), but given the southward track shift you observed under higher SSTs isn’t it possible that cyclones that would otherwise have passed north of the city could shift to hit it instead? This also comes up on line 262, as it seems to me that the southward shift may spare Shanghai from some landfalls but could equally easily expose it to cyclones that would otherwise pass harmlessly north.
WRF resolution. On line 104, you describe the vertical resolution as 45 layers going up to 50 hPa. Are those levels terrain-following, hybrid, or pressure-following? And is the layer spacing constant or does it change with height? At least a few studies (Wu et al 2019 in Acta Oceanologica Sinica; Ma et al 2012 in Asia-Pacific Journal of Amostpheric Sciences) indicate that the vertical resolution in specific parts of the atmosphere can have a noticeable effect on TC simulation with WRF.
Urbanization case definition. Why choose the lowest urban development scenario, SSP1, for your increased urbanization case (line 112-113)? This seems designed to under-estimate the potential impact of urbanization.
SST case definition. Near the end of the paper, you acknowledge a potential pattern effect of SST change which is in contrast to the uniform warming you imposed (line 280-281), so why did you go with a uniform warming case (line 118)? Particularly when you already have the pattern effect, at least in a climatological sense, in Figure S1.
Significance or Confidence Intervals on TC metrics. Figures 5 and 6 look a lot at changes in key TC metrics from the CTR case, and line 151-152 discusses deviations from the control track in km. Is there any way to establish statistical significance of these changes, using either the bias from observations or some spread in these metrics from an ensemble of TC simulations? For example, some of the magnitudes discussed are quite small (line 186-187 discusses a “consistent” change in maximum rainfall and maximum wind speed in Figure 6, but those changes are on the order of 1-4 mm/h and 1-2 m/s, which seems very small to attribute to a significant impact of urbanization). I realize this could be challenging to quantify, but if there is any way to do so I think it would make it much easier to interpret Figures 5 and 6.
Domain for calculating I_max and W_max. In Figure 6, the maximum rainfall and wind speed are calculated just over the Shanghai domain, but in Figures 5 and 7 it is not clear to me the domain that was used to find these maxima. Is it within the radius R during the timesteps that Shanghai was also in that radius, similar to Figure 2c?
Color scale for Figures 5 and 6. I don’t understand the color mapping used in these figures, and on fairly close inspection actually find them quite deceptive. As far as I can tell, each panel (single heat map) uses a shared colorbar, which has a diverging colormap where the lowest values are red and the highest are blue. But the intention of the figure seems to be to compare changes from the control simulation, so in my view there should really be a separate color scale for each row (otherwise the highest-intensity values will always be on the most and least intense TCs, so you don’t really see changes even when they do exist across the middle of the pack), such that the CTR simulation is grey for each TC, any increase from that is red, and any decrease is blue, with the two sides symmetric so that the intensity of the color indicates either the absolute or relative (which may be better, since the text mostly discusses changes as a percent of the control case) change from the control case. As is, there are some very confusing cases where the text describes, for example, a three-fold increase in the radius of Jongdari in the SST3 scenario (line 160), but the corresponding box in Figure 5a is the palest shade in the Jongdari row, which would imply the opposite. Either way, the color scaling should be explained in the caption, particularly since it is not provided as a colorbar on the figure.
Diverging colormap for uniformly increasing values. A minor point, but S1 uses a diverging red-blue colormap even though the underlying values are uniformly positive and represent a change in temperature over time. I found this unnecessarily confusing, and it took me a while to realize that the blue areas are still an increase in SST, just a smaller one.
Definition of Delta in Figure 7. It was not clear to me what the Delta in rainfall and wind represents. In “after landfall”, for example, is the change calculated as a change over the Shanghai domain from the moment of landfall to one time step (1 hour) after landfall? 6 hours after landfall? In that case, line 196 is confusing to me, as it states that an enhancement of rainfall both before after landfall indicates that Jongdari decays slowly after reaching land; if rainfall increases after landfall, it’s not decaying at all yet, right? And if the change is actually between the CTR case and a SST/U3km case, I’m still not sure what time(s) “before” and “after” refers to. This is another case where I think a more detailed explanation in the figure caption of what the Delta I_max and Delta W_max^10m actually represent is needed.
Fujiwhara effect seen in all cases. The finding that the Fujiwhara effect played a role in the response of all 5 TCs to increased SSTs was particularly interesting to me (line 221-222). Was there no secondary pressure low in the CTR case, and one developed in all 5 cases under increased SSTs? Or did the secondary low already exist but just got stronger/closer to the primary TC? Is there any explanation for why this change in the effect strength happens under increasing SST? Explaining why the Fujiwhara effect gets stronger would help to support the claim made in line 296-297, particularly since about 61% of current TCs occur as doubles (line 228), which is a large number but wouldn’t necessarily produce any change in the other 39% of TCs unless the increase in SSTs actually makes double-TC events more frequent rather than just changing the range/strength of interaction.
Influence of El Nino. In line 266, you state “anomalously warmer SST, so-called El Nino, significantly influences typhoons on a large scale.” However, the support for this point that follows seems to be based on a single El Nino year (2023) and subsequent TC season. I think you either need more references to support the claim that El Ninos in general can influence typhoons on a large scale or to soften that claim to a possibility rather than a surety.
Technical Corrections
Line 44: “there are not many evidences” -> “there is not much evidence”
Line 92: “followed the TC central location” -> “following the TC central location”? I’m not sure I understand this sentence structure, though
Line 105 and in Table 2: “Yonsei University scheme (YSU) scheme” -> “Yonsei University (YSU) scheme”
Figure 7 caption: “landfall from Shanghai” -> “landfall in Shanghai”
Line 203: “T” -> “T^850”
Line 225: “The typhoon” -> “Typhoons” and “has been found normally moving” -> “have been found to move” or similar
Line 238: “amount of water vapor context” get rid of “context”
Line 242: “enhances the upward” -> “enhances upward”
Line 243: the citation should be in-text, no parentheses
Line 249-250: “marking a record-breaking four-time landfall” not sure what this refers to; did the TC make landfall four times? Or was the record broken by a factor of four?
Line 273: “tropical regions” New York and Tokyo (2 of the 3 listed cities) are mid-latitude, not tropical
Citation: https://doi.org/10.5194/egusphere-2025-1002-RC1
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 225 | 213 | 14 | 452 | 18 | 12 | 23 |
- HTML: 225
- PDF: 213
- XML: 14
- Total: 452
- Supplement: 18
- BibTeX: 12
- EndNote: 23
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1