the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 06 Jun 2024
Regional modelling of extreme sea levels induced by hurricanes
Abstract. Coastal zones are increasingly threatened by extreme sea level events. Storm surges are one of the most hazardous components of these extremes, especially in regions prone to tropical cyclones. This study aims to explore factors affecting the performance of numerical modelling in simulating storm surges in the tropical Atlantic region. The maxima, duration and time evolution of the extreme storm surge events are evaluated for four historical hurricanes by comparison against tide gauge records. The ADCIRC and NEMO ocean models are intercompared using a similar configuration in terms of domain, bathymetry and spatial resolution. These models are then used to perform sensitivity experiments on oceanic and atmospheric forcings, physical parameterizations for wind stress and baroclinic/barotropic modes. NEMO and ADCIRC show a similar skill to simulate storm surges induced by hurricanes. Storm surges simulated with ERA5 atmospheric reanalysis forcing are generally more accurate than those using parametric wind models for simulated hurricanes. The inclusion of the baroclinic processes improves storm surge amplitudes in some coastal locations such as along the southeastern Florida peninsula (USA). Experiments exploring different wind stress implementations and the interactions between storm surges, tides and mean sea level however have shown a minimal impact on storm surges induced by hurricanes.
- Preprint
(18595 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on nhess-2024-100', Coleman Blakely, 15 Jul 2024
General Comments
This manuscript details an examination of two different hydrodynamic models
(ADCIRC and NEMO) and their applications in modeling of storm surge due to
tropical cyclones. Tropical cyclones were modelled both using ERA5 wind
reanalyses and a selection of parametric wind models. Sensitivies between ADCIRC
and NEMO, as well as sensitivities to wind model and drag law, are evaluated.
The results of the various experiments indicate that storm surge, defined as the
non-tidal residual in this study, are relatively insensitive to the use of
ADCIRC vs NEMO. However, there are some differences when comparing wind product
used. This study shows that the reanalysis winds generally outperform the
parametric wind models.Overall, I believe that this manuscript presents some useful information with
regards to modeling storm surge caused by tropical cyclones. The experiments
were clearly carefully designed to examine the sensitivities of interest. The
combination of an inter-model as well as an intra-model comparison was
particularly interesting. I do have some comments regarding scientific questions
and experiments as detailed in the "Specific Comments" section; however, I think
the scientific merit of this manuscript is generally sound.While the manuscript as a whole is generally interesting and explores some
useful/impactful questions, there are numerous grammatical/typographical issues
that make the story told hard to follow. I have detailed some specific examples
of readability issues in the "Technical Corrections" section; however, these
examples should not be taken to be a comprehensive list of typographical
problems. I strongly suggest a thorough editing of the manuscript to improve
readability and accessibility.I recommend major revisions, primarily related to presentation quality, and
look forward to reading a revised version of this manuscript.Specific Comments
Section 2
When describing the selection of tide gauges used in the evaluation of storms
you note that you select gauges that are within a 300 km radius of the hurricane
in lines 91-92. In lines 95-96 you note that gauges registering storm surges of
less than 15 cm are also excluded. Why not simply use the surge-based criteria
alone rather than also having the distance criteria? I just find the distance
criteria to be rather arbitrary whereas the surge criteria (assuming you base
the selection off of observations instead of model results) is more rational.When describing the parametric wind models in lines 108-126, you do not touch on
the various parameters within each model and what you set them to. For example,
in the Chavas et. al. (2015) model there are several parameters that need to be
prescribed by the user. Among these are the radiative-subsidence rate (W_cool),
surface drag coefficient (C_d), and the ratio of enthalpy and momentum exchange
coefficients (C_k/C_d). The other parametric wind models also have
user-prescribed coefficients. You do not touch on what values you selected for
these parameters. I think it would be useful to add a subsection to Section 2
where you give a short description of each parametric wind model and the values
you selected in order to make the experiments herein more reproducible by
others.Section 3
When describing the numerical grids of ADCIRC vs NEMO in Section 3.1, you do not
note the total number of degrees of freedom of the two grids. While it is not
front and center in the manuscript, you do provide wall-clock times in the
appendix. Without knowing the number of mesh vertices it is hard to get a sense
of the true computational cost/how it will scale with more computational power.Section 4
In lines 276-277 you note that, "ADCIRC simulations tends to underestimate the
maximum [storm surge] compared to tide gauge data." Could you compute the mean
error in addition to the mean absolute error to give a sense of the differences
between NEMO and ADCIRC in this regard? It is hard to quantify based off of
Figures 4 and 5 alone without hard numbers. Using the mean error would allow for
the performance to be quantifiably evaluated instead of qualitatively. In the
discussion at line 437, you also note that, "[i]n general, both models
underestimate the storm surge amplitudes."You note in lines 351-355 that different parametric wind models perform are best
for different storms. I feel that this might relate to the selection of
parametric wind model parameters. While I do not think you need to do a
sensitivity analysis of the wind model parameters (although you could), I think
it is worth noting that this may be part of the issue.In lines 357-359 you say that the parametric wind models underestimate more than
ERA5, could this be due to the far-afield winds not being captured in the
parametric models? Does this underestimation go away if you only look at
stations, for example, within the 30 knot wind radius of the center of the
storm?Section 6
I think that adding a description of future work would be quite valuable. For
example, NEMO lacking wetting/drying capability will drastically limit its
usefulness as a storm surge prediction model due to lack of ability to capture
inundation of floodplains. I assume that adding a wetting/drying algorithm is
part of future work but it might be useful to state here. Additionally, I think
it might be useful to compare the computational performance of NEMO vs ADCIRC,
especially since the manuscript highlights the usefulness of NEMO in simulating
storm surges.Technical Corrections
Lines 22-34: This paragraph is hard to follow. It seems to try and
highlight two different things: 1) how tropical cyclones cause storm surge, and
2) the four storms evaluated in this manuscript. While I was able to understand
the message of the paragraph, I had to read the whole thing two or three times
to truly parse what was being communicated. As an example, rather than saying,
"Wilma is the most intense Atlantic hurricane by lowest pressure on record,
formed on October 15, 2005, reaching sustained winds of 295 km/h before making
landfall in southwestern Florida on October 24, 2005.", it would be more clear
to write something along the lines of, "Hurricane Wilma, which formed on Ocober
15, 2005 and made landfall in southwestern florida on October 24, is the most
intense Atlantic hurricane on record as measured by atmospheric pressure." I
think if might be useful to separate the description of each storm at the very
least into multiple sentences. Another approach would be to remove the explicit
storm descriptions and instead add more of the detail to Table 1.Lines 54-56: In the sentence starting with, "Parametric wind models," the clause,
"enable to compute a large number of simulations," is grammatically incorrect.Line 64: Another study that looks at the inclusion of baroclinicity in
evaluating storm surge is Pringle et. al., 2019
(doi:https://doi.org/10.1029/2018JC014682).Lines 100-102: You describe ERA5 as having a resolution of 0.25 degrees and
ERA-Interim as having a resolution of 79 km. While it is relatively simple to do
the conversion I would suggest including a ballpark resolution in kilometers for
ERA5, e.g., "The atmospheric variables have a horizontal resolution of 0.25
degrees (~25 km)..."Lines 108-126: A good chunk of this paragraph could be moved to the
introduction, particularly the description of the history of parametric wind
models. The description of IBTrACS should remain in this section. Alternatively,
as mentioned above, having a separate subsection with each parametric model
described in more detail would be useful.Line 180: "Tidal amplitudes are relatively moderated in the region..." This is a
little awkward. Perhaps clarify what you mean by "moderated".Line 213: "This experience is conducted with the ADCIRC..." Should be
"experiment".Line 235: You say that variables are described, "in the NEMO description part."
Give a section number instead to make it easier to locate where this information
can be found.These are just some examples of places where I found the readability of this
manuscript to be lacking. I think that some thorough editing will really bring
out the message and findings of the article and make it much more accessible to
the reader.Citation: https://doi.org/10.5194/nhess-2024-100-RC1 - AC1: 'Reply on RC1', Alisée Chaigneau, 05 Sep 2024
-
RC2: 'Comment on nhess-2024-100', Anonymous Referee #2, 17 Jul 2024
This manuscript focuses on the performance of NEMO and ADCIRC ocean models in simulating storm surges in tropical east Atlantic.
It concludes that both ocean models can simulate storm surge in a similar way.
It also concludes that ERA5 atmospheric reanalysis forcing gives better results than parametric wind models and also that inclusion of baroclinic processes in the simulation improves the result significantly. Different wind stress parametrizations, and the surge interaction with tide and mean sea level have shown little to minimal impact.
Globally I would say that this study is well conducted, and gives some interesting results.
However, I would suggest a general review to improve the flow and readability.
Sometimes it gets hard to follow.
I have one specific comment:
In lines 219-230 you describe how you investigated the impact of wind stress parameterizations.
I find this section very hard to follow. I also would say that is poor in terms of content.
I would re-write this part describing both equations in more detail, especially Eq.1, that has no description of the variables whatsoever.
I appreciate the fact that you performed a different simulation with a variable Charnock parameter depending on wave parameters. You should explain why you performed this additional simulation, mentioning why sea roughness is dependent on wave parameters, explaining what variables have an impact, and cite the authors that investigated this process.Also, I could not find (Smith and Banke, 1975) on your reference list. There could be others missing.
I recommend to check the list very carefully.Citation: https://doi.org/10.5194/nhess-2024-100-RC2 - AC2: 'Reply on RC2', Alisée Chaigneau, 05 Sep 2024
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 316 | 111 | 81 | 508 | 15 | 18 |
- HTML: 316
- PDF: 111
- XML: 81
- Total: 508
- BibTeX: 15
- EndNote: 18
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1