the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Sensitivity analysis of mesoscale simulations to physics parameterizations: a case study of storm Ciara over the Belgian North Sea using WRF-ARW
- 1Department of Environmental and Applied Fluid Dynamics, von Karman Institute for Fluid Dynamics, Waterloosesteenweg 72, 1640 Sint-Genesius-Rode, Belgium
- 2Department of Mechanical Engineering, Vrije Universiteit Brussel, Boulevard de la Plaine 2, 1050 Ixelles, Belgium
- 1Department of Environmental and Applied Fluid Dynamics, von Karman Institute for Fluid Dynamics, Waterloosesteenweg 72, 1640 Sint-Genesius-Rode, Belgium
- 2Department of Mechanical Engineering, Vrije Universiteit Brussel, Boulevard de la Plaine 2, 1050 Ixelles, Belgium
Abstract. The Weather, Research and Forecasting (WRF) model includes a multitude of physics parameterizations to account for atmospheric dynamics and interactions such as turbulent fluxes within the planetary boundary layer (PBL), long and short wave radiation, hydrometeor representation in microphysics, cloud ensemble representation in cumulus, amongst others. A sensitivity analysis is conducted in order to identify the optimal WRF-physics set-up and impact of temporal resolution of re-analysis dataset for the event of sudden changes in wind direction that can become challenging for reliable wind energy operations. In this context, Storm Ciara has been selected as a case study to investigate the influence of a broad combination of different interacting physics-schemes on quantities of interest that are relevant for energy yield assessment. Of particular relevance to fast transient weather events, two different temporal resolutions (1-hourly and 3-hourly) of the lateral boundary condition's re-analysis dataset, ERA5, are considered. Physics parameterizations considered in this study include: two PBL schemes (MYNN2.5 and scale-aware Shin Hong PBL), four cumulus schemes (Kain-Fritsch, Grell-Devenyi, and scale-aware Grell-Freitas and multi-scale Kain-Fritsch,) and three microphysics schemes (WSM5, Thompson and Morrison) coupled with two geospatial configurations for WRF simulation domains. The resulting WRF predictions are assessed by comparison to observational RADAR reflectivity data on precipitation. In addition, SCADA data on wind direction and wind speed from an offshore wind farm located in the Belgian North Sea is considered to assess modeling capabilities for local wind behavior at farm level. For precipitation, results are shown to be very sensitive to model setup, but no clear trends can be observed. For wind-related variables on the other hand, results show a definite improvement in accuracy when both scale-aware cumulus and PBL parameterizations are used in combination with 1-hourly temporal resolution reanalysis data and extended domain sizes.
Adithya Vemuri et al.
Status: final response (author comments only)
-
RC1: 'Comment on wes-2021-137', Anonymous Referee #1, 27 Jan 2022
Summary:
This study explores a series of possible WRF physics configurations for a Storm Ciara in the North Sea on 10 February 2020. The authors vary temporal resolution of the initial and lateral boundary condition data, the size of the WRF domains themselves, the boundary layer scheme, the cumulus scheme, and the microphysics scheme. The WRF model output from each simulation is validated against observations of wind speed and wind direction at an offshore wind farm. Model precipitation data is also evaluated against Doppler radar observations at the offshore wind farm. The authors found that scale-aware cumulus and boundary layer schemes, along with 1-hourly input data and a larger WRF domain generally performed better for this case.
Recommendation:
Reject.
Major Comments:
- A thorough technical English edit is required. There are numerous issues throughout the manuscript, most of which are specifically mentioned in the Minor Comments from the Abstract through Section 2 as examples of the issues that need fixing. I only mentioned a couple items from Sections 3–5.
- Lines 95–98 provide the key to defining the niche that this article aims to fill. I think you need to do a better job honing in and repeatedly showing how your work fills that niche. I also think that validating against data at only a single point for a single storm is inadequate to fill that niche. That is really the biggest fundamental issue I have with this manuscript, and why I gave it a Reject instead of a Major Revision. A single case study can have value if you do more than validate against observations only at a single point, while validating at a single point can have some value if you evaluate multiple cases. Are there any other buoys or towers that are available for offshore wind validation? I am aware of FINO1 in the North Sea region, but I believe it is outside your d04, unfortunately.
- Validating precipitation at only a single point is of limited use, even over many cases, when your goal is to determine which model configuration gave the most realistic simulation of precipitation. If you want to validate model precipitation or reflectivity, then you should leverage the land-based radar that you do have data from, and do object-based validation with MODE (https://dtcenter.org/met-online-tutorial-metv8-0/mode) for a more comprehensive validation.
- Many of the figures need substantial revision. (I originally had these subpoints labeled a–d, but this text editor in the WES form renumbered them to 40–43 for some reason.)
- Fig. 1 (several of these comments also apply to Figs. 6, 8, 10, and 11): It is customary to plot coastlines or national borders in black or gray on most maps. Using a color from your colorbar (red) is simply confusing. Also, your filled contour colors do not match the colorbar. The thin ribbons of darker colors around fields of pastel colors also make this figure difficult to interpret with any confidence. Additionally, in this figure the colorbar label says “Precipitation” but has units of mm/h. Precipitation would have units of mm, but mm/h are units for precipitation rate. The caption also states that the figure is depicting radar reflectivity, which again is not quite the same thing as precipitation rate, though they of course are related to one another. The caption also states that the observed radar reflectivity (or really, radar-derived precipitation rate) is valid at 04:00, but line 137 says it is valid at 04:40. Which is it?
- Figs. 3 and 4: You should have a thin gray line in your legend if it is in your plot. In the x-axis label (also in Figs. 5, 7, and 9), also use a date format like 10 Feb 2020. 10/2/2020 will easily confuse American readers into thinking the date is 2 Oct 2020.
- Fig. 5: The orange dashed line is quite faint and difficult to see.
- Fig. 6 (most of these comments also apply to Figs. 8, 10, and 11): First, calling these Domain 1 and Domain 2 is misleading. These are really Domain Configurations 1 and 2; both these domain configurations have domains 1–4, so when you say domain 1 or domain 2, my mind automatically thinks of the outermost two WRF domains. Second, the radar contour lines all look the same color, which seems like a mistake. Third, restrict your WRF filled contour range to the equivalent of the radar-derived precipitation rate, or at least restrict the lower bound cutoff to something like 0.01 mm/h. Is it really raining at lower values than that, anyway? Values below your lower bound should be transparent/not plotted. That will also solve the undesirable issue of the entire domain being filled with a dark blue that makes other features difficult to discern while also being meaningless. Fourth, state in the caption what the star is.
- Line 135: To be completely honest, changes in wind direction of 40° do not seem like a huge shift—it is not even half of a quadrant. If 40° is a hugely consequential shift that wind farm operators need to be quite concerned about, then it would be helpful to provide some justification.
- Table 2: I suggest either reordering pairs A–K based on the order they are discussed in Sections 4.1–4.4, reordering Sections 4.1–4.4, or both (my preference). It makes more sense to me to look first at the size of the domain and the lateral boundary condition temporal frequency, before then comparing different physics schemes. Also, in Table 2 ensure that cell borders are turned on to separate the different experiment pairs in the cumulus pair column. Additionally, in the section titles for Sections 4.1–4.4, it would be helpful to include the experiment pair letters.
- Table 7: In the Average NED column for rows 10 and 11, you have 1.10 in green, 1.111 in yellow, and 1.32 in red. The values 1.10 and 1.111 are so close that it is misleading to make them such different colors. Is the difference between 1.10 and 1.111 in this metric even meaningful? What would be a meaningful difference in NED or Kantorovich distance? In any case, in Tables 4–7, I really think you would be better off keeping the color scale and ranges from Table 3.
- Lines 325–327: First, change “ensemble” to “ensemble mean” or “ensemble average”. Second (and more importantly), there are many papers and books that explain why the ensemble mean usually outperforms individual ensemble members (e.g., Wilks 2019, https://www.elsevier.com/books/statistical-methods-in-the-atmospheric-sciences/wilks/978-0-12-815823-4). It would be worthwhile to engage with some of that literature here, especially since your findings of the ensemble mean not being the best are contrary to what was expected. Do you have any insights as to why the ensemble mean performs comparatively poorly in the Kantorovich distance for wind speed and wind direction? This appears to be why the NED is not the best for the ensemble mean for the wind variables. Perhaps this is a side effect of the randomness introduced by having a sample size of only one event validated at only one point?
Minor Comments/Typos:
- Title: Storm Ciara is a proper noun, so “storm” should be capitalized here and throughout the manuscript.
- Throughout: Change “RADAR” to “radar”. It has been decades since radar was written in all caps in formal writing.
- Throughout: Whether you use Oxford commas or not, the journal guidelines state that you need to be consistent, but there is not consistency of usage in your article. There are many places where you do use Oxford commas (e.g., line 55), and many more places where you do not (e.g., line 12). I strongly prefer and encourage them because I think it enhances readability, but it is a personal style choice. Just be consistent.
- Throughout: Follow the journal’s guidelines on defining nearly all acronyms and abbreviations at first usage.
- Throughout: Change “[horizontal] resolution” to “[horizontal] grid spacing”. It is quite common for people to use resolution and grid spacing as synonyms, but they do not mean quite the same thing. A numerical model’s effective resolution is usually about 6–8 times the horizontal grid spacing.
- Throughout: Provide the time zone for all times in this article. I presume your times are in UTC, but it is never stated.
- Lines 2, 30: Change “Weather, Research and Forecasting” to “Weather Research and Forecasting”.
- Line 3: Change “long and short wave radiation” to “longwave and shortwave radiation”. See also line 152 (remove the hyphens).
- Line 4: Change “amongst others” to “and others”.
- Line 5: Change “WRF-physics set-up and impact of temporal resolution of re-analysis dataset” to “WRF physics setup and the impact of temporal resolution of a reanalysis dataset”.
- Line 6: Change “the event” to “occurrences”.
- Line 7: It would be nice to mention in the abstract when Storm Ciara occurred.
- Line 11: Change “Kain-Fritsch, Grell-Devenyi, and scale-aware Grell-Freitas and multi-scale Kain-Fritsch” to “Kain-Fritsch, Grell-Devenyi, scale-aware Grell-Freitas, and multi-scale Kain-Fristch”.
- Line 14: Change “data on” to “data of”.
- Line 23: Change “associated to” to “associated with”.
- Line 30: Change “open-source” to “public domain”. Also, since you use WRF v4.2, you should instead cite the WRF v4 technical note (Skamarock et al. 2019, https://doi.org/10.5065/1dfh-6p97). Consider also citing Powers et al. (2017, https://doi.org/10.1175/BAMS-D-15-00308.1).
- Line 48 and elsewhere: Add a comma after “i.e.” (Also add commas after “e.g.”)
- Line 49: Change “explicitly” to “explicit”.
- Line 52: Change “grid-column” to “grid column”.
- Line 62: Change “explores; redistribution of” to “redistributes”.
- Line 109: Change “Island of Mauritius” to “island of Mauritius”. Island is not part of the proper name for Mauritius, and thus is a common noun that should not be capitalized.
- Line 111: “concluded in no particular combination of WRF physics” — This is awkwardly worded. Please revise.
- Line 113: Change “sensitivity analyzes” to “sensitivity analyses”.
- Line 114: “time-lapse considered within the diurnal cycle” — This is awkwardly worded. Please revise.
- Line 124: “exposed” is an odd word choice here.
- Line 127: Change “on” to “over”. Change “transpiring” to “transiting” (or to “moving”, or something similar).
- Line 131: “in Ostend, located at the Belgian offshore coast” — Perhaps you mean “at the Belgian coast”?
- Line 132: Please define what you mean by “over the local region”. How local? D04 is not very big to begin with.
- Line 137: Change “RMI-B, for brevity” to “RMI-B; for brevity”.
- Line 140: Change the semicolon after Damiani et al. (2018) to “and”.
- Line 141: Change “addition to” to “addition of”.
- Line 143: Change “context of wind energy applications” to “context of offshore wind energy applications in this region”.
- Line 154: It should just be Tewari et al., 2004, not Mukul Tewari et al., 2004. Ensure that is corrected in your reference manager.
- Line 202: Change “encompasses” to “has” or “uses” (or similar). Also use the definite article “the” before each of these physics scheme names and datasets.
- Line 218: Change “an equivocal literature” to “equivocal findings”.
- Table 3 caption: Capitalize Euclidean.
- All throughout Section 4: I think you need to consistently refer to domain configuration 1 and 2, as each domain configuration has domains 1, 2, 3, and 4. If you say domain 1, the reader will think of your outermost WRF domain, which is domain 1 (d01).
- Tables 4–7 captions: You forgot to include NED in the captions. Also, change “subject to” to “with” or “comparing”.
- Line 322: This sentence is awkwardly worded.
- AC1: 'Reply on RC1', Adithya Vemuri, 24 Apr 2022
-
RC2: 'Comment on wes-2021-137', Anonymous Referee #2, 04 Feb 2022
General comments:
Overall, the manuscript is written in an understandable way. However, I see the need for additional revision for both language and parts of the content (more details in the specific comments section). The introduction is very detailed and provides sufficient references to connect the article to ongoing research and the state-of-the-art. While the methodology (evaluation metrics, model setups) has been described nicely and adequately, some details, especially about availability of measurements are missing. Furthermore, I think subsections for the different elements (model setup, evaluation metrics and description of validation data incl. pre- & post-processing and data availability) would be useful to keep the different components more separated.
Specific comments:
general remarks:
There are major inconsistencies in the way radar reflectivity and precipitation (rate) are handled in the manuscript. This is especially visible in some of the figures (captions mentions reflectivity while figure legends talk about precipitation with units of precipitation rates, e.g. Figure 1, 4), but also in the text (e.g. line 188-191). A major revision of all elements of the manuscript is needed by the authors with special attention to a consistent usage of atmospheric properties and units.
Throughout the paper, the term “horizontal resolution” has been used to describe the grid spacing of the WRF domains (among others in Table 1, line 145/146 and elsewhere). This is misleading since horizontal (effective) resolution and grid spacing are not equivalent for numerical models like WRF (see e.g. Skamarock 2004, https://doi.org/10.1175/MWR2830.1 for details). Please replace the term “horizontal resolution” with “grid spacing” where needed.
As mentioned earlier, the descriptions in the methodology section are quite convoluted with model setup, evaluation metrics, introduction and processing of measurements and WRF post-processing all described in the same section. I would suggest to introduce sub-sections for the model setup (incl. WRF post-processing of radar reflectivity), evaluation metrics and the measurements to enhance readability and to make space for more details, especially with respect to data availability for the radar and SCADA data (publicly available data, protected data or similar). If publicly available, please also state the access point of the data and provide more details about the wind farm.
remarks addressing specific lines:
Line 30: The reference (Skamarock et al. 2008) points towards Version 3 of the WRF model, but in your methodology section, you mention that you are using Version 4.2. Is there a particular reason why the Version 3 reference is used here and not Version 4?
Line 130: Reference(s) for the statements/quoted numbers of wind gust, travel path, effects etc. are missing. Please consider adding.
Line 151: “Subsequently, the model is run …”. Please consider reformulation since the current formulation could be misinterpreted as two separate independent simulations (one 24h long simulation and one 21h long simulation). I assume the WRF simulation has been run in one continuous block?
Line 153/154: “The land surface interactions are kept constant”. Please reformulate since it is the parameterization scheme that is kept constant, not the interactions themselves.
Line 213: Make sure that you are talking about the correct boundary conditions (temporal resolution of the LATERAL boundary conditions not the INITIAL boundary conditions). Please correct.
Technical corrections:
Language corrections:
I am not a native English speaker, so I restrict myself here to mostly obvious formulations and spelling errors that caught my eyes.
Line 2: remove “,” after “Weather”
Line 4: replace “,” after “microphysics” with “and” Line 5: while the resolution of the re-analysis products is important, it is the temporal resolution of the lateral boundary condition updates that is investigated here (which is not necessarily identical), so I would suggest a formulation like “update interval of lateral boundary conditions” instead of “temporal resolution of re-analysis data”.
Line 10: remove “:”
Line 11: replace "Grell-Devenyi" to "Grell-Dévényi" (as written in Line 59)
Line 12: remove “,” after “Kain-Fritsch”
Line 23: replace “associated to” with “associated with”
Line 30: remove “,” after “Weather”
Line 34: I am not sure what you mean by “expanse of physics parameterizations”. Do you mean variety? I would suggest reformulation.
Line 61: replace “Frietas” with “Freitas”
Line 62: “[…] explores; redistribution […]”. Sentence structure is unclear, please consider reformulation.
Line 87: replace “resolving” with “resolve”
Line 96: “XPIA” has not been introduced before. Please introduce the abbreviation.
Line 111: add “the” in front of “west Pacific Ocean”
Line 124: add “s” to “perspective”
Line 139: “potentially influential of the state of power”. Unclear, do you mean “influenced by”?
Line 147: remove “largest” since there is only one parent domain.
Line 168/169: replace “PBL” with “PBL scheme”
Line 195/196: Insert “temporal” before resolution
Line 233: remove “to”
Line 237/38: capitalize “D” in “domain” for consistent naming
Line 248: This sentence is difficult to understand, please reformulate. Do you mean the Shin-Hong PBL scheme simulation is in better agreement with the SCADA data?
Line 260: replace “on” with “with” after “dependency”
Line 273: remove “through”
Line 291/292: add “scheme” or “parameterization” after “cumuls”
Line 333: Unclear what “wind-farm power excursions” means. Please reformulate.
Tables:
Table 1: Number of nested domain is misleading. Following WRF naming conventions (see e.g. http://dx.doi.org/10.5065/1dfh-6p97), it would be one parent domain (d01) and 3 nested domains (d02-d04). Furthermore, it would be good to mention if 1-way or 2-way nesting has been used.
Table 1: Time-step information incomplete. It is not clear which domain uses the 20s time step. I would suggest to either mention the time-steps for the four domains explicitly or state the domain which uses the 20s time step and provide the time step ratio.
Table 1: Consider replacement of “update frequency” with “update interval” to be consistent with the values given in the second column (1h and 3h are intervals, not frequencies)
Table 3-8: Units for MAEs, NEDs and Kantorovich distances are missing, please add. I would also suggest to change the E notation of the Kantorovich distance (rather uncommon in scientific literature) to decimal notation. The differences in scales of magnitude are not large enough to justify such notation to the expense of readability.
Figures:
In general, the font sizes in some of the figures are too small (especially Figure 1, 2 and the time series plots). Please adjust to increase readability.
Figure 1: Mismatch in time statement in caption (4:00) and reference in main text (4:40, l. 137). Please double-check. A description of the meaning of the star-symbol and the red lines is missing in the figure caption (and other figures captions as well). Please also consider changing the coastline color, which is too similar to colors used in the color map of the variable you are plotting.
Figure 1,6,8,10,11: Since the focus region is rather small, please add latitude and longitude information to make it easier to locate features.
Figure 3: insert “are” before “shown”
- AC2: 'Reply on RC2', Adithya Vemuri, 24 Apr 2022
Adithya Vemuri et al.
Adithya Vemuri et al.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 332 | 117 | 10 | 459 | 3 | 4 |
- HTML: 332
- PDF: 117
- XML: 10
- Total: 459
- BibTeX: 3
- EndNote: 4
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1