the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Selecting CMIP6 GCMs for CORDEX Dynamical Downscaling over Southeast Asia Using a Standardised Benchmarking Framework
Abstract. Downscaling global climate models (GCMs) provides crucial, high-resolution data needed for informed decision-making at regional scales. However, there is no uniform approach to select the most suitable GCMs. Over Southeast Asia (SEA), observations are sparse and have large uncertainties, complicating GCM selection especially for rainfall. To guide this selection, we apply a standardised benchmarking framework to select CMIP6 GCMs for dynamical downscaling over SEA, addressing current observational limitations. This framework identifies fit-for-purpose models through a two-step process: (a) selecting models that meet minimum performance requirements in simulating the fundamental characteristics of rainfall (e.g., bias, spatial pattern, annual cycle, and trend) and (b) selecting models from (a) to further assess whether key precipitation drivers (monsoon) and teleconnections from modes of variability are captured [El Niño-Southern-Oscillation (ENSO) and Indian Ocean Dipole (IOD)]. GCMs generally exhibit wet biases, particularly over the complex terrain of the Maritime Continent. Evaluations from the first step identify 19 out of 32 GCMs that meet our minimum performance expectations in simulating rainfall. These models also consistently capture atmospheric circulations and teleconnections with modes of variability over the region but overestimate their strength. Ultimately, we identify eight GCMs meeting our performance expectations. There are obvious, high-performing GCMs from allied modelling groups, highlighting the dependency of the subset of models identified from the framework. Therefore, further tests on model independence, data availability, and future climate change spread are conducted, resulting in a final sub-set of two independent models that align with our a priori expectations for downscaling over CORDEX-SEA.
- Preprint
(11718 KB) - Metadata XML
-
Supplement
(5125 KB) - BibTeX
- EndNote
Status: open (until 29 Jul 2024)
-
RC1: 'Comment on gmd-2024-84', Anonymous Referee #1, 11 Jul 2024
reply
General comments
This paper describes a standardized benchmarking framework for selecting CMIP6 GCMs for CORDEX downscaling over Southeast Asia. The topic is important because Southeast Asia faces a high risk of flooding due to climate change, yet fewer models or frameworks are available for characterizing regional changes in precipitation compared to other regions such as Europe and the US. The authors did a great job highlighting the differences between their approach and those in the literature, which mainly rank GCMs according to specific evaluation matrices. The logic of this paper is very clear, and it is very well written. I only have a few minor points for the authors to consider.
Technical corrections
L44: GCMs’
L51, L55: should be ‘WCRP’?
Section 1: an overview of the paper structure should be added to the end of this section so the readers know what they expect in each section.
L120-121: do you mean ‘We do not consider models which have a horizontal grid spacing greater than…’. Or by ‘greater’ do you mean finer resolution than 2 degrees?
L123-124: incomplete sentence.
L174: you may want to remove theta from the first half of the sentence and explain it as wind direction(?)
L198: did you define DMI somewhere above?
L218: what do you mean by ‘significant sign’?
L320-322: not sure if I follow the definition or description of the benchmarking threshold. Do you find the six wettest and driest modelled months and require the four wettest and driest months from observations to be within those six modelled months? Then how is the threshold determined?
L340-342: did you show the observational trend somewhere or can you cite references for this claim?
Citation: https://doi.org/10.5194/gmd-2024-84-RC1 -
RC2: 'Comment on gmd-2024-84', Anonymous Referee #2, 22 Jul 2024
reply
Review on
Selecting CMIP6 GCMs for CORDEX dynamical downscaling over Southeast Asia using a standardised benchmarking framework
By Phuong Loan Nguyen, Lisa V. Alexander, Marcus J. Thatcher, Son C. H. Truong, Rachael N. Isphording, and John L. McGregor
General comments
In this study, the authors proposed an approach to select suitable GCMs for dynamical downscaling. This approach includes a standardized benchmarking framework that consists of two steps. One is based on minimum performance requirements in terms of the reproducibility of simulated precipitation. The other is associated with the representation of simulated key precipitation drivers and teleconnections. The second step seems to be unique and reasonable. However, there are some concerns as mentioned comments written below. The most important one may be the method for determining threshold values of metrics to judge whether a model well reproduces precipitation itself, key precipitation drivers, and teleconnections.
Major comments
- L147:
Perhaps the authors forgot to put section 2.2.1 just after this line. Putting here an explanation of fundamental metrics, such as MAPE and Scor, would be preferable.
- L158:
Maybe a good model performance based on key physical process in the historical climate does not always guarantee a good performance in terms of future climate. This is the same situation as the case of MSMs, as the authors mentioned.
- L244:
Maybe relative change would not always be a good indicator. Wouldn’t it be OK if the authors could also check the difference between the two (future minus historical), in particular, in a dry season?
- L251:
Using satellite data, such as TRMM and CMORPH, enables the authors to validate simulated precipitation over ocean as well.
- L289:
How about using RMSE as a metric to validate simulated precipitation. What do the authors think about it?
- L313:
Do the authors think that further validation is needed by using another observational product, such as CHIRPS?
- L315:
The method for determining threshold values seems to be important, as the authors mentioned here. Wouldn’t it better to determine the number of models that would be used for downscaling first, and then, to choose models in order of better performance? In this case, the authors do not need to determine threshold values.
Minor comments
- L131:
It would be better to write the resolution of ERA5 here, which would be helpful for readers.
- L170:
There seems to be no description of the abbreviation of MAPE.
- L173:
Could you explain the advantage of this metric? How about a metric as follows:
Sqrt((Ui-Ui,ref)**2+(Vi-Vi,ref)**2)
- L174:
Typo? Should we delete “theta i theta ref”?
- L274:
The threshold values seem to be somewhat subjective. What made the authors deduce these values.
- L307:
“Consequently” would not the right word here because the performance of biases does not always result in that of correlation.
- L340:
There seems to be a decreasing trend.
- L407:
Figures in bias seem to be preferrable for clear understanding of this discussion: overestimation of the wind intensity relative to ERA5.
- L509:
The linear relationship is not necessarily needed because it is between the changes of temperature and precipitation, not between temperature and precipitation themselves.
- L533, L543:
The number of clusters seem to be somewhat subjective. It would be preferrable to describe what is behind these specific numbers.
Citation: https://doi.org/10.5194/gmd-2024-84-RC2
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
234 | 66 | 17 | 317 | 31 | 9 | 11 |
- HTML: 234
- PDF: 66
- XML: 17
- Total: 317
- Supplement: 31
- BibTeX: 9
- EndNote: 11
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1