the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 05 Apr 2024
A storyline-based approach towards changing typhoon intensities over the Pearl River Delta under future conditions using Pseudo-Global Warming
Abstract. During the Pacific typhoon season of 2023, the Pearl River Delta (PRD) was hit by a series of typhoons, causing unprecedented precipitation as well as fatalities and significant damages. There are indications that these events may intensify under climate change. However, the unfolding of similar events in the future is yet to be fully understood. We therefore conducted an investigation of historical typhoon events affecting the PRD using Pseudo-Global Warming. Within this framework, we perturbed the historical initial and boundary conditions obtained from ERA5 reanalysis and handed to the regional model WRF according to the climate change signal projected by the CMIP6 ensemble under SSP5-8.5. We pursued a storyline-based approach, in which each of the 16 selected CMIP6 models acted as the basis for an individual storyline of future typhoon intensity. Additionally, we created two storylines based on thermodynamic drivers to create scenarios with excessively favorable and unfavorable conditions for typhoon intensification, resulting in 18 storylines that were assessed for seven representative typhoons. WRF was set up in a two-way nesting framework using domains of 25 km and 5 km resolution, of which the latter was used for further assessment. For each typhoon, the simulations were initiated at the start of the first intensification phase and statistically assessed until landfall. Minimum sea level pressure, maximum wind speed, mean and maximum 1-hourly precipitation rates as well as the integrated kinetic energy (IKE) as an advanced measure for typhoon damage potential were used to determine typhoon intensity. Results indicate a general increase in typhoon intensity across all metrics for six of the seven inspected typhoons. This increase is notably higher for specific storylines, and the projected increase in the extreme values of the inspected metrics significantly exceed the median change of all storylines. This indicates that the true potential range may lie above what would be expected under a median approach. The results suggest a maximum decrease of up to 15 hPa for minimum central pressure, 11 m s-1 increase for maximum wind speed, 2.5 mm h-1 for mean and 50 mm h-1 for maximum precipitation rates. The two additional storylines revealed an even higher intensity increase in the form of central minimum pressure decreases and wind maxima increases for two typhoons, but mostly resembled the span provided by the 16 GCM-based storylines. These results can support the optimization of the development of protective measures considering the improved range of intensity potential.
- Preprint
(4144 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on hess-2024-95', Anonymous Referee #1, 02 May 2024
This paper presents numerical simulations for a series of typhoon events using the WRF model and PGW methods to extract the SSP585 far future scenario global warming signal from 16 CMIP6 models, which is added to the ERA5-based historical climate as the IC BC of the model. The authors have invested significant effort in model simulation and demonstrated methodological rigor; the article is well-written and easy to read. It highlights changes in intensity and precipitation in seven typhoon cases under future scenarios, compared to historical control runs, as well as the impact of varying surface and low-level atmospheric temperatures on typhoon intensity. The topic is not related to hydrology or urban hydrology, I'm not sure if it coincides with the scope of HESS journal. There are some merits in the study, but there are substantial flaws that prevent the manuscript from publication. The discussion lacks depth concerning the physical mechanisms of climate change impacts on typhoons, especially for the typhoon dynamic and thermodynamic changes. The paper primarily presents changes in typhoon intensity and precipitation as projected by the WRF model and PGW method under a warmer climate, but it fails to systematically explore the underlying dynamic and thermodynamic reasons for these changes. This omission limits the depth of understanding of the complex physical processes involved. Moreover, the conclusions presented are not particularly novel and innovative, focusing mainly on the increase in typhoon intensity and precipitation due to global warming. Addressing these issues could significantly improve the paper’s quality.
- The introductory section is cluttered and lacks a logical structure, and it is more of a thesis introduction than that of a research article. Furthermore, while the authors spend considerable time discussing existing research gaps in current PGW studies, they barely touch upon the specific effects of climate change on typhoon intensity and the physical mechanisms behind it. The authors mention previous studies but do not systematically explain the results and progress made in this field. For instance, whether climate change will lead to stronger typhoons in the western Pacific, how much will the typhoon intensify with one degree SST warming, and how climate change can affect typhoons' intensity and precipitation. There should already be plenty of research covering these topics. Strengthening the introduction would make it easier for readers to grasp what has been accomplished and the scientific basis of it.
- On page 5, line 155, the manuscript states that 28 typhoons were initially selected for performance assessment, but only seven were ultimately chosen for detailed study. This selection process raises questions about the model’s ability to accurately simulate typhoon tracks and intensities. As the author mentioned in the introduction: “How well are historical typhoons affecting the PRD represented by our selected model set up regarding track accuracy and intensity?”. If only seven out of 28 typhoons demonstrate reasonable model performance, this suggests potential limitations in the model’s reliability for simulating such events.
- On page 8, line 250, the use of spectral nudging is mentioned, but it's unclear whether it was applied across all domains. Typically, spectral nudging is used at the outermost model domain to minimize its impact on simulation results. Clarification of how spectral nudging was implemented in this study is needed for a better understanding of its influence on the results.
- The analysis of Figure 4 focuses only on the mean bias of maximum wind speed and minimum sea level pressure, which may not fully capture the model’s reliability and accuracy (results are also very sensitive to the time period chosen for time averaging). A more robust approach would be to compare the time series of the model’s maximum wind speeds and minimum sea level pressures against best track data, to assess both intensity and temporal accuracy more comprehensively.
- Section 3.2.2 discusses experiments comparing loS-hiA (low surface - high atmospheric temperatures) and hiS-loA (high surface - low atmospheric temperatures) scenarios. While the comparison is interesting, the deviation from realistic conditions and the lack of in-depth analysis on the physical mechanisms influencing typhoon dynamics is a major drawback. The authors only describe time series plots of typhoon intensity and precipitation, which is seriously lacking in the analysis and explanation of the mechanism from the point of view of typhoon dynamics, thermodynamics, and physics. A reasonable research article should not only pose scientific questions to identify problems but also provide well-founded explanations and reasons, rather than merely describing the observed model outputs.
- Following up on the previous comment, after reviewing sections 3 and 4, I am concerned that the results presented in this article do not sufficiently support the research paper. While it is evident that the authors conducted a large number of experiments using the WRF model, the results section primarily showcases time series plots of typhoon intensities and precipitation (including the Integrated Kinetic Energy) from these experiments. These findings are not particularly novel and innovative, as several studies have already employed the WRF and PGW methods to investigate the impact of climate change on typhoon intensity and precipitation, often arriving at similar conclusions. My suggestion is for the authors to conduct a comprehensive analysis of the model results, comparing the simulation outcomes of PGW signals derived from individual CMIP6 models. It is helpful to determine whether the variations between different experiments stem from discrepancies in CMIP6 model global warming signals and to assess how these differences specifically affect typhoon intensity. The authors should also delve into the reasons behind changes in typhoon intensity and precipitation, detailing the specific dynamics and thermodynamics of typhoons and how these alterations influence typhoon behavior, which is a crucial aspect of this research. For example, in comparisons such as loS-hiA and hiS-loA, the study should explore how varying temperatures introduced through PGW modify the typhoon structure and dynamics, rather than merely describing the time series of typhoon intensity.
- As noted in Section 4.3, line 636, the author acknowledges that “The selected sample size of 7 typhoons is too small to draw a generalized conclusion on typhoon intensity over the PRD under climate change.” If the model accuracy can’t be improved, I will recommend that the authors extend their analysis to include all 28 typhoon cases, regardless of whether the WRF accurately reproduces the typhoon’s path and intensity very accurately. This broader analysis could enable the authors to draw more general and comprehensive conclusions. For instance, they could determine how much warming per degree of SST might lead to an enhancement of typhoon intensity; and whether the increase in typhoon rainfall due to SST, atmospheric temperature, and typhoon intensity, following the Clausius-Clapeyron (CC) relationship. The authors should also compare these findings with those from existing references to provide a more robust evaluation of the impacts of climate change on typhoons.
Citation: https://doi.org/10.5194/hess-2024-95-RC1 - AC1: 'Reply on RC1', Jianhui Wei, 04 Oct 2024
-
RC2: 'Comment on hess-2024-95', Anonymous Referee #2, 06 May 2024
Olschewski et al., investigate tropical cyclones (TCs) over the Pearl River Delta and their physical changes under climate change using the storylines approach. They select historical TCs and change their initial and boundary conditions using the Pseudo-Global Warming method.
The method used in this study is interesting, as it takes advantage of the flexibility that storylines offer to carry out a thorough analysis, going beyond mean/median changes in projections. I particularly liked the extra 2 storylines exploring favorable and unfavorable thermodynamic conditions, as this type of thinking helps overcome potential limitations on climate models. The different results obtained for the thermodynamic storylines support the added value of exploring them. I also like the inclusion of integrated kinetic energy (IKE) as an additional metric aimed at offering some proxy for impact. I believe the manuscript contributes to the field and is compatible with this journal. Having said that, I have some comments and questions that could provide more context to the work.
Spectral nudging:
- In the methods section (line 252), spectral nudging is used to minimize spatial variability. I found the idea of combining both approaches interesting. Did you run/check simulations without applying spectral nudging? I wonder how influential this step is in your setup. I think this merits more discussion on the implications of using and not using the spectral nudging.
- While it makes sense to remove spatial variability to make comparisons more straightforward, a spatial change in TC track as consequence of climate change would be an interesting finding and relevant for impacts. If there are results available for the runs without the spectral nudging, you could consider analysing them as well.
- Since you included spectral nudging in addition to PGW, you could discuss the role (e.g., advantages, drawbacks, and when to use each one) of each method - PGW, spectral nudging, and their combined use - in simulating TCs and their use with storylines.
Temperature levels:
In figure 2 you demonstrate the resulting delta change in temperature for each model from April to October. I am intrigued to understand the direct relation between the temperature change per model for each storm and the change in relevant metrics, such as precipitation. For instance, in Figure 2b CanESM5 (yellow) shows the largest delta change for 3 months, which made me expect to see the same model causing the largest changes for precipitation across all storms. Instead, only for one case, Hagupit, we see that. Could it be that the other storms occur in months where CANESM5 is not as intense as other models? I believe adding an extra (SI) Figure comparing some results (maybe mean precipitation) to the increase in temperature per model could offer extra insights into the mechanisms behind the increase in impact and the spread across models (do warmer models lead to more changes?).
Contextualization:
I see that the focus of the discussion section was in comparing the results for TC estimations in the DPR with other studies. However, I miss some discussion on your approach and results from a storylines perspective. There are other storyline works that explored TCs in DPR (Qiu et al., 2022), TCs using spectral nudging (Goulart et al., 2024) or even using a similar Pseudo-Global Warming method (Dullaart et al., 2024). I believe also framing the study within the storylines field of work can provide a better contextualization and visibility to your work.
References:
Qiu, J., Liu, B., Yang, F., Wang, X., and He, X.: Quantitative Stress Test of Compound Coastal-Fluvial Floods in China's Pearl River Delta, Earth's Future, 10, e2021EF002638, https://doi.org/10.1029/2021EF002638, 2022.
Goulart, H. M. D., Benito Lazaro, I., van Garderen, L., van der Wiel, K., Le Bars, D., Koks, E., and van den Hurk, B.: Compound flood impacts from Hurricane Sandy on New York City in climate-driven storylines, Nat. Hazards Earth Syst. Sci., 24, 29–45, https://doi.org/10.5194/nhess-24-29-2024, 2024.
Dullaart, Job CM, et al. "Improving our understanding of future tropical cyclone intensities in the Caribbean using a high-resolution regional climate model." Scientific Reports 14.1 (2024): 6108.
Minor comments:
In general the writing is a bit long and complex, some more direct text could make the flow of the manuscript better.
Line 41: biggest natural risk factors -> I think biggest is not the best adjective here.
Line 120: “This is based on the findings of Shen et al. (2000), Hill and Lackmann (2011), and Tuleya et al. (2016) who found that an increase in thermodynamic atmospheric stability and increased sea surface temperatures counteract with regards to typhoon intensity” – The text is a bit confusing, and the sentence ends with a comma.
Line 164: “initial and boundary conditions from ERA5 are applied on a grid-cell basis” – does it mean for each grid cell you apply a specific delta factor? So it means that the delta factor is calculated for every grid cell and every timestep of the simulation?
Line 366: “simulations the agreement of the historical and simulated tracks is high” -> Is this already including the spectral nudging? If so, it could be a bit more clear.
Line 425: I think it could be rewritten to make it more clear and direct.
Line 624: Missing punctuation in “; Kantha, 2006) Based on”
Limitations (line 628): Since you already aimed for some impact proxy using the IKE, you could mention that other impact approaches (metrics, models etc.) can also enhance the study from an impact perspective, such as wind and flood modelling.
Citation: https://doi.org/10.5194/hess-2024-95-RC2 - AC2: 'Reply on RC2', Jianhui Wei, 04 Oct 2024
-
CC1: 'Comment on hess-2024-95', Jimmy Fung, 26 May 2024
The comment was uploaded in the form of a supplement: https://hess.copernicus.org/preprints/hess-2024-95/hess-2024-95-CC1-supplement.pdf
- AC4: 'Reply on CC1', Jianhui Wei, 04 Oct 2024
-
RC3: 'Comment on hess-2024-95', Anonymous Referee #3, 27 May 2024
In this manuscript, the authors run a set of pseudo-global warming (PGW) simulations to investigate how 7 tropical cyclones (TCs) in the Pearl River Delta (PRD) region might change under warming perturbations from 16 different CMIP6 models (plus two additional "extreme" perturbations). They report results regarding storm pressure, wind, precipitation, and integrated kinetic energy. They find that storms generally increase in intensity across all metrics for 6 of the 7 storms.
The application of climate storylines has become popular in recent years. The PGW framework is one way such storylines can be performed that targets regional scales. I agree that there is utility in such simulations, particularly around understanding how climate may change in these regions and communicating these to downstream individuals interested in climate data (e.g., planners, emergency managers, etc.).
TCs and climate are important topics, but there are opportunities to deepen the analysis, better interpret the science, and make the findings more impactful. As is, I find the paper somewhat uninspiring. The actual analysis of the storms themselves is fairly shallow, with only basic qualitative comparative evaluation and not providing a deeper interpretation of the dynamical structure (e.g., how the changes in stability impact the storm's axisymmetric and asymmetric circulations). Even the evaluation of aspects such as the radius of maximum wind and precipitation distributions could be interesting. Conversely, the sample of 7 TCs doesn't feel large enough to draw any meaningful conclusions and the authors do not provide any real concrete reasons for their decision to eschew a larger sample size (28 storms). Of note, there is a lack of statistical tests to evaluate if any of the changes are robust. The general findings mainly serve as further confirmation of the numerous PGW/TC simulations published over the past 10 years or so.
The paper reads a bit like using a tool (PGW simulations) to find a nail (TCs in the PRD). It currently is written in a way that reads very linearly and more like a technical document than a scientific paper. As is, I think the paper needs a fairly large overhaul before it can be considered for formal publication in a journal. The good news is I think this can be achieved with a better interpretation and deeper understanding of the data as opposed to additional model simulations. My major comments are described below and I think need to be critically addressed before any publication.
Of note, the data availability requirements of HESS (https://www.hydrology-and-earth-system-sciences.net/policies/data_policy.html) do not appear to be satisfied by the statement "All data used to conduct this study will be made available upon request." The simulation data (or at least a subset of it able to recreate the results) should be uploaded to Zenodo or some other repository. Given the relatively small nature of the simulations (i.e., regional domain and short time windows), this shouldn't be an issue with some careful consideration of exactly what data should be archived.
Major comments:
There isn't an objective metric for choosing 7 of the 28 storms. The authors describe evaluations of intensity and track, but only superficially and rapidly jump from Fig. 4 (all TCs) to Fig. 5 (the 7 TC subset). The obvious concern here is -- if the other 21 TCs were poorly enough simulated to not be included in this analysis, is the version of WRF applied here really "fit for purpose"? That is, if it struggles to simulate the TCs, why do we believe the PGW signal is credible? Simply put, I think this decision needs a clearer rationale. If the other storms were excluded due to poor simulation quality, it would be helpful to provide a detailed analysis of these issues and discuss how they might affect the credibility of the remaining PGW simulations.
There is no statistical significance testing. This appears to be a fairly big oversight in my mind -- even basic tests like KS or a t-test could shed some light on how robust these changes are. The authors speculate about this occasionally in the manuscript (talking about the "majority" of members that increase in intensity, for example) but more quantitative analysis is needed.
The IKE section is very underdeveloped (only 15 lines in the text). The evaluation is very superficial and, in my opinion, could be improved with fairly little effort. For example, Neoguri has a larger signal in the maximum wind versus the IKE. This would imply the structure of the TC is compensating (smaller?) for the IKE to change little even in the face of a large change in wind speed. Conversely, the IKE signal seems larger for Usagi, implying that the storm wind field is expanding (either in the inner core or the outer reaches). This comes back to my overarching concern that this paper reads very much as a shallow-ish description of model simulations without some deeper probing.
In general, the paper could be better served by providing at least some spatial evaluation (e.g., 10.1038/s41586-018-0673-2). For example, is the precipitation maximum just increasing because all precipitation rates are increasing, or is the fundamental structure of the TC changing? In the storms that weaken in the PGW runs, does this appear to just be internal variability in storm intensity (see literature surrounding rapid intensification and weakening in TCs due to inner core processes) or is this fundamentally a response to the large-scale environment (perhaps an evaluation of metrics such as maximum potential intensity would be worthwhile)?
Minor comments:
Lines 106-107. Are we sure CMIP6 versus CMIP5 provides value-added for PGW runs?
Fig. 2. While there is a sample of 16 GCMs. there really should be a discussion of model independence -- for example, there are multiple versions of CMCC, MPI, and NorESM2, with the only apparent difference being resolution. I would expect these sets of models to have very similar changes when compared to models with vastly different structural characteristics.
Line 229: 5km grid spacing with cumulus convection remaining on? At these grid spacings, do researchers typically apply cumulus parameterizations or turn them off?
Fig. 4. The authors should include a discussion of how the pressure-wind relationship simulated in models impacts these results (e.g., doi:10.6057/2018TCRR04.01)
Fig. 6. I assume for the precipitation rates (c,d), the authors first find a single value at each model timestep (e.g., an array that is 1 x ntimes for each member) and then include those in the statistics. That is, there is no inclusion of spatial components in these boxplots?
Figs. 7-8. It is very difficult to interpret these results outside of the thicker black, red, and blue lines. If the authors would like to provide more model-specific context, I would suggest either converting these lines to some form of shading or larger figures that could provide the ability to see some of the model lines (which currently are stacked very much on top of one another).
Lines 422. I am not sure what is particularly unique about Hato and Hagupi. Is it that the median value of the PGW runs falls in the 25-75% range of the ERA5?
Citation: https://doi.org/10.5194/hess-2024-95-RC3 - AC3: 'Reply on RC3', Jianhui Wei, 04 Oct 2024
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 672 | 124 | 255 | 1,051 | 21 | 17 |
- HTML: 672
- PDF: 124
- XML: 255
- Total: 1,051
- BibTeX: 21
- EndNote: 17
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1