the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 05 Aug 2024
Inclusion of the ECMWF ecRad radiation scheme (v1.5.0) in the MAR model (v3.14), regional evaluation for Belgium and assessment of surface shortwave spectral fluxes at Uccle observatory
Abstract. The MAR model (Modèle Atmosphérique Régional) is a regional climate model used for weather forecasting and climate studies over several continents, including polar regions. To simulate how solar and Earth’s infrared radiation propagate through the atmosphere and drive climate, MAR uses the Morcrette radiation scheme. Last updated in the 2000’s, this scheme is no longer maintained and lacks of flexibility to add new capabilities, such as computing high resolution spectral fluxes.
This paper presents version 3.14 of MAR, an update that allows MAR to run with ecRad, the latest radiation scheme provided by the European Centre for Medium-range Weather Forecast (ECMWF). Operational in the ECMWF’s Integrated Forecasting System (IFS) since 2017, ecRad was designed with modularity in mind and is still in active development.
We evaluate the updated MAR by comparing its outputs over 2011–2020 for Belgium to gridded data provided by the Royal Meteorological Institute of Belgium (RMIB) and by the EUMETSAT Satellite Application Facility on Land Surface Analysis. Several sensitivity experiments have been carried out to find the configuration achieving the most balanced radiative budget as well as to demonstrate the updated MAR outperforms its former configuration. Moreover, a MAR simulation running ecRad with high resolution ecCKD gas-optics models has been conducted to produce spectral shortwave fluxes, which are compared to ground-based spectral measurements captured by the Royal Belgian Institute for Space Aeronomy (BISA) at Uccle (Belgium; 50.797° N, 4.357° E) in the 280–500 nm range from 2017 to 2020. Finally, as a first application of spectral shortwave fluxes computed by MAR running with ecRad, a method for predicting UV indices is described and evaluated.
- Preprint
(2002 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-1858', Anonymous Referee #1, 05 Oct 2024
A nice paper overall but think some improvements are needed. I like the inclusion of the UV index work. Perhaps more work is needed to show that ecRad is as good if not better than the old scheme. Also some of the results and calculations need further explanation. Some suggestions and questions are included below:
L29 Don't think this is needed: with the latter resulting from the Earth's surface being heated by the former.
Table 1: typo developed --- drop one d
Figure 1: Long dashed versus short dashed - I'd suggest you make this clearer. May be better to apply colour instead.
L127: or as a mean of coupling ---- means
L133-134: that manages spectral bands extending over a few dozens of nanometres, e.g. 25 nm -- slightly unclear
L139: By swapping RRTM-G with high resolution ecCKD gas-optics models, --- what's the computational cost?
L158: The number of aerosol species --- any limitations on this??
L161: Therefore, to make the most of ecRad radiation scheme --- "the" missing
L161: the greenhouse gases and aerosols forcing -- think aerosol should be singular
L164: over the 2000?s -- don't think apostrophe should be there
L171: This scaling factor is then applied on the ---- to the
L172: the scaled forcings are interpolated in the MAR grid ---- to the MAR grid
L173: averaging with respects --- respect
L160-175: Not sure if all of these details are needed here - perhaps could go in an Appendix either? It somewhat disrupts the flow of the paper.
L176: CAMS aerosol specification, consisting of 11 hydrophilic or hydrophobic aerosol species --- are there not 14 species?
L178: being that CAMS data provided by the --- the CAMS data
L180: between poles --- the poles
L181: be covering the high resolution MAR grid --- will cover
L181: or simplicity's sake --- for the sake of simplicity
L186: the pressure value of 90 hPa --- change to a pressure of 90 hPa
L194: near-)surface --- perhaps put in a space
L192-196: Para should include how you got around the Ozone issue. Maybe merge paras.
L200 - that took account --- that take account of the properties
Para around L200 - are you saying the the vertical grid is different only for the rad calculations?
L206: and specific humidity --- and the specific
L208: and stay consistent --- to stay
L209: Finally, it is assumed the temperature --- that the temperature
Perhaps stick to Celsius rather than mixing Kelvin and deg C?
L210-212: The temperature OF the stratospheric layers can be inferred by finding the plausible
tropopause in the column (which is not necessarily AT the top of THE said column), adding 60 K to obtain the temperature at the
stratopause, then linearly interpolating TO the intermediate values. (see capitalised words)Figure 3: in the MAR grid --- to the MAR grid
L215: with the same method as in Sect. 3.1 --- using the same
L219: a total of ozone --- total ozone (remove of). Also in L220
L226 -- remove also at end of sentence
L228: an underestimated cloudiness, the radiation scheme also requiring cloud fraction values for each grid cell among its inputs. --- does not read well.
L229: computer model is whole research -- is a whole
In general I'd suggest getting a native English speaker to read through the paper as there are words missing here and there etc.
Figure 4 caption -- too long a sentence - confusing to read.
L232: with an -- using an
L235: sent to ecRad -- used by
L236: The parameterizations from -- replace from with "by"
L237 -- replace them with these --- them refers to people
ut are good at predicting the total cloud cover, which is currently the most important
requirement for MAR regarding cloudiness, given that research involving the MAR model focuses mostly on (near-)surface
processes, --- I don't agree - cloud condensate has a huge impact on the surface variable, more so than cloud cover - compare a high thin cloud to a low thick cloud, both having 8 octa.L251: (i.e., droplets, ice crystals and snowflakes), --- is graupel and hail accounted for in these?
L256: the parameterizations from --- from to of
State disadvantages of ------- such as more advanced diagnostic parameterizations (Weverberg et al., 2021b, a) or prognostic solutions
(Tompkins, 2002), both requiring more implementation work. The inclusion of these more complex solutions in the MAR
model is left for future work.L257 - parametrization is the correct spelling I believe
L261-262: Change to "for the sake of simplicity"
L264 - remove in ecRad at end of sentence. Assume you mean the extra 4% overall in the computation time.
L265 - sentence a bit long and clunky. Are you saying your can't use the LW scattering with Tegen - unclear from the sentence. I think default
settings should be included in the paper rather than referring to a wiki, which can be updated at any stage.The phrase simplicity's sake appears many times - not good use of English.
L269: at once --- change to "at a time" And sentence is too long, split into two.
L271: Again a reason to have a table of the default settings - unclear from reading whether TripleClouds is the default or you have chosen a
a non-default option. Change McICA to McICA scheme.L278: Change to past tense - have been
L280: Need to be tested.....
L281: comma missing after fw.
L285: to lower values
L288: since version .... remove "the"
L291: change the others to just "others".
L292: is as follows.
L293: and 24 pressure ....
L296: tuned not the correct word. Perhaps say configured.
L299: we completed two tasks. Take care of the tense used throughout - should be past as you refer to experiments you have done.
L300: Second, we evaluated..... also the sentence is too long.
L304/305: respectively without and with a heat fluxes tuning mechanism inherited ---- with and without a heat flux tuning mechanism
L307: known radiative fluxes biases ---- known radiative flux biases ..... w.r.t. - perhaps don't abbreviate
Table 2 caption: for the sake of readability. Heat flux rather than the plural
Re ecCKD - which version did you use? The new sped-up one by Peter Ukkonen (refactored) or the old one?
L310: The very first ecRad --- remove the word very.
Make more use of commas throughout the paper to enhance readability.
L311: adjustments --- typo
L312: Starting with ...., were added ---- explain this better - have you changed the number of model levels in the runs?
from E3 to E6) test the --- rm from ... tested the.L314: parametrization .... for the fw parameter of ecRad ---- of the fw parameter.
L315: tense. ecCKD gas-optics models ---- model not plural
L329: daily mean surface shortwave downward fluxes --- daily mean surface downward shortwave fluxes and daily mean surface downward LW fluxes.
L331: with the daily mean temperature being based on observations at 2 m above grass --- this is the normal so you prob don't need to state it.
L338: and Middle-East ---- and the Middle-East.
L336 - sentence way too long and clunky.
L339: due to a lack of a gridded --- due to the lack
L341: relative error below 10% compared --- where does the 10% come from, should it be 20?
L345: only with a different projection system -- only on a different projection. MSG satellite -- singular
L346: degrees
L347: for each couple of time series --- for each pair of time series. Written a bit misleadingly.
L348: which can be visualized on the grids of the data products, have been both saved and averaged, ---- don't think you need to mention they
have been saved etc.L350: with respects to ---- with respect to.
From here on I won't correct the language as I think the full paper can be checked by a native English speaking author.
L352-353: sentence needs rephrasing.
L355: Paragraph not well written. Language vague - perhaps include more of the values. Not conclusive about what's happening in each experiment.
More detail needed in order to distinguish the options.Some of the biases quoted are surely less than the bias on the satellite product in the first place.
Mention of the extra levels here - should that be described much earlier in the paper?
Looking at Table 3 - diffs are not major and within range of uncertainty in the obs.
L365 - the diffs are so small, I would say you can't really draw such conclusions from them.
L370 + paragraph: Can you justify more why you can use fw=0.5? At higher res the inhomogeneity should be closer to 1?
L365-380: This discussion is a bit weak I think - I don't think too much weight can be put on the small differences.
L385-395: explain these results a bit better and why you are confident in them? They are not all stat significant though so difficult to draw such
conclusions. Are you have if ecRad is comparable to Morcrette, so as long as it's comparable it's OK, and you switch to the more advanced scheme.Fig 5/6: Might be good to show areal plots for all experiments - nicer than the summary stats and can also see any geographical variation.
L400 etc: I don't think you should argue that you're happy to have both + and - biases compared to all in one direction.
Table 5: Might be good to include all ecRad expts here and not just 3. Perhaps include expt number in the table to avoid having to decribe again in
the textL415 - have you (or can you) test the refactored ecCKD?
L424: Can you explain this - longer time elapsed but yet ecRad more efficient
L423-430 - written confusingly. Please elaborate more on the numbers and your conclusions re speed of ecRad.
I don't follow how ecRad is much slower but overall runtime more or less similar. What causes that?L433: You mention the 26% but have not put it in context compared to the other expts.
L458: Define BISA
L485 Paragraph a bit confusing - write more clearly
L480: diurnal measurements --- do you mean daylight measurements?
Fig 7: Can you comment on the overestimation in MAR compared to OBS across all bands? I know it's small but seems consistent.
Fig 8: Would it be better to filter by clear days to try to figure out what's causing the difference? The cloud days are more complex as you
will always have cloud mismatches. Have you looked at any clear sky indices for MAR vs Obs over the time period? Clear sky index would clearly
show whether you are lacking cloud in certain situations.L560 and before/after: Perhaps all the calculations should go in an appendix so as not to distract from the results and discussion.
L575 - can you comment on why MAR can't capture maxima? Is it the resolution? How often is the radiation scheme called - has this been experimented with?
OK I see you answered later but perhaps add comment to the section.Citation: https://doi.org/10.5194/egusphere-2024-1858-RC1 -
RC2: 'Comment on egusphere-2024-1858', Anonymous Referee #2, 17 Oct 2024
The authors have done a nice job with implementing the newest version of ecRad in the MAR regional atmospheric model. The paper describes the testing of this concisely and gives an nice use case with computing the UV index.
Minor comments:
- Line 4: "... and lacks of flexibility to add new capabilities, ..." --> "... and lacks the flexibility to add new capabilities, ..."
- Figure 1: The available cloud optics schemes for liquid and ice clouds should be listed.
- Line 141: I don't understand the text in the parentheses. Can you reformulate this?
- Section 3.1: "Updated greenhouse gas and aerosol forcings". The SSP scenarios include aerosol scenarios. Did you include these? If not, then explain why not! The direct, and in particularly the indirect, effects of aerosols are of similar magnitude as the greenhouse gases regarding contemporary climate change (Forster et al. 2024).
- Table 3: Are the RMIB gridded precipitation data based on radar and rain gauge data, on interpolated rain gauge data, or on modelled precipitation data? If it is based on one of the latter two, I would suggest to refer to the resulting statistics as "Difference" and "RMSD".
- Table 3: For the LW data "Bias" has been exchanged with "Difference" - presumably due to the more uncertain nature of the MSG LW radiation data. To be consistent, "RMSE" should here be replaced with "RMSD".
Citation: https://doi.org/10.5194/egusphere-2024-1858-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 285 | 74 | 113 | 472 | 6 | 6 |
- HTML: 285
- PDF: 74
- XML: 113
- Total: 472
- BibTeX: 6
- EndNote: 6
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1