the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Downscaling the probability of heavy rainfall over the Nordic countries
Abstract. We used empirical-statistical downscaling to derive local statistics for 24-hr and sub-daily precipitation over the Nordic countries, based on large-scale information provided by global climate models. The local statistics included probabilities for heavy precipitation and intensity-duration-frequency curves for sub-daily rainfall. The downscaling was based on estimating key parameters defining the shape of mathematical curves describing probabilities and return-values, namely the annual wet-day frequency fw and the wet-day mean precipitation μ. Both parameters were used as predictands representing local precipitation statistics as well as predictors representing large-scale conditions. We used multi-model ensembles of global climate model (CMIP6) simulations, calibrated on the ERA5 reanalysis, to derive local projections for future outlooks. Our analysis included an evaluation of how well the global climate models reproduced the predictors, in addition to assessing the quality of downscaled precipitation statistics. The evaluation suggested that present global climate models capture essential covariance, and there was a good match between annual wet-day frequency and wet-day mean precipitation derived from ERA5 and local rain gauges in the Nordic region. Furthermore, the ensemble downscaled results for fw and μ were approximately normally distributed which may justify using the ensemble mean and standard deviation to describe the ensemble spread. Hence, our efforts provide a demonstration for how empirical-statistical downscaling can be used to provide practical information on heavy rainfall which subsequently may be used for impact studies. Future projections for the Nordic region indicated little increase in precipitation due to more wet days, but most of the contribution comes from increased mean intensity. The west coast of Norway had the highest probabilities of receiving more than 30 mm/day precipitation, but the strongest relative trend in this probability was projected over northern Finland. Furthermore, the highest estimates for trends in 10-year and 25-year return-values were projected over western Norway where they were high from the outset. Our results also suggested that future precipitation intensity is sensitive to future emissions whereas the wet-day frequency is less sensitive.
- Preprint
(13236 KB) - Metadata XML
-
Supplement
(49524 KB) - BibTeX
- EndNote
Status: open (until 30 Jul 2024)
-
RC1: 'Comment on egusphere-2024-1463', Anonymous Referee #1, 20 Jun 2024
reply
Review of “Downscaling the probability of heavy rainfall over the Nordic Countries”, by Benestad et al.
This manuscript is addressing an important topic and is well-written. However, I would recommend major revisions to give more details on the main body of the manuscript rather than in the SI. I think that the manuscript should be rewritten for the broad readership of HESS, or alternatively be submitted to an more specialized journal.
The essence of the approach presented is to build a statistical model that relates large-scale information (EA5 or CMIP6) to station-level data. Then, this model is used to estimate the future climate at stations locations, based on CMIP6. I am not sure this approach can be termed downscaling, as in my mind downscaling is the passage of data from a coarse grid to a finer grid. Here the correct term, in my understanding, would be something like change of support. In particular, the data are only produced at the stations locations, meaning that it is less effective in areas with a lower stations density (such as northern Norway/Finland).
Moreover, the details of the methodology are very condensed and not comprehensible for a reader that is not already familiar with it. The method is summarized very shortly in section 2.2, and mostly referred to through citations to other papers by the authors, with the reader is not necessarily familiar. Therefore, it is difficult to grasp it by only reading the manuscript. A schematic of the methodology would be helpful.
Very few details are given on the kriging procedure used. Which variogram parameters were used, and what does this tell us about the spatial dependance - which is more discussed from the angle of the variance carried by the principal components, a very indirect way of looking at spatial correlation. Furthermore, I expect the spatial dependance to be non-stationary, e.g. very different in western Norway than in other parts of the domain. How is this addressed?
For instance, section 2.2 mentions the use of EOFs for the large-scale data and PCA for the station data. It is not clear why a different analysis was done for each data type.
The validation of the results is mostly done visually, with a few assessment metrics given along the text in section 2.3. I would expect results of the cross-validation to be given as detailed tables.
I guess that many of the comments above can be addressed by referring to the extensive supplementary material, but I believe that an article should be self-contained and should not require readers to go through some 160 pages of annexes to understand the main points.
Figure 5: are these statistics calibrated? In other terms, for a Normal distribution I would expect about 64% of the data to fall with the 1-standard deviation confidence interval. This does not seem to be the case here. Can this be commented?
The discussion section is long and windy. Some of it could be moved to the introduction (e.g. ll.290-300), or to the conclusion, or removed.
Typos:
l.83: Russian -> Russia
l.91.92: Specify the cause of that: some runs are not available at a daily resolution
l.257: …the presence *of* various…
l.304 (and other places): references to unpublished work should be removed,
Figures 3 and 4: the projection seems incorrect, resulting in deformed maps
Figure 6: The legend foes not match the figure
Citation: https://doi.org/10.5194/egusphere-2024-1463-RC1 -
RC2: 'Comment on egusphere-2024-1463', Anonymous Referee #2, 04 Jul 2024
reply
The subject is valuable, the data used is pertinent, and the paper is interesting and well-written, but major revisions are necessary. The document is not self-sufficient, and relies too much on the supplementary material.
An example is the difficulty to assess the method's quality due to insufficient description of the methodology and evaluation parts.
These sections should be enhanced by adding more details, including figures.
Key details to add in the method part should include the amount of data used to calibrate the regression model (i.e., train the model), the amount of data used to evaluate the model's performance, the size of the data used as predictors (inputs), and the size of the data used as predictands (outputs).
The evaluation on ERA5 data is crucial and should be comprehensively included in the main document. References to supporting material should be limited to more detailed evaluations. For example, clarifications are needed for the statement about the “close match between the rain gauge data and ERA5” (Line 134). It should specify whether this evaluation was done on data used to calibrate the regression model.
Other comments
L71-77 : It seems the authors are describing a bias/variance tradeoff. What I understand is that it is desired to avoid fitting the predictands too closely with statistical techniques to avoid overfitting (high variance, where models become overly sensitive to data used for training/calibrating). It would be helpful to clarify if the exactitude (precision and accuracy in capturing detailed patterns - in calibration data, akin to high variance ?) versus robustness (the ability to generalise to new data, capturing essential patterns) tradeoff is similar to a bias/variance tradeoff. Additionally, the authors describe their strategy as “obtaining a model with high robustness at the cost of precision” but do not detail what happens after this choice. I would expect some specifics on the chosen model in this introduction.L144: “The evaluation also involved testing the ability of the GCMs in reproducing the predictors in a skillful way” needs clarification. I assumed the GCMs were the predictors used for projections. From the context, it seems the GCMs are compared against ERA5 to verify if the model trained with ERA5 data could be applied using GCM predictors.
Typos
L83 : “the highest point being 2062 above” missing unit
L257: missing “of” between various and meteorological
Citation: https://doi.org/10.5194/egusphere-2024-1463-RC2 -
RC3: 'Comment on egusphere-2024-1463', Anonymous Referee #3, 08 Jul 2024
reply
The study presents a statistical methodology on parameterizing the average annual probability of rainy days (fw), and the annual average daily precipitation (μ) at a rain gauge scale based on co-variates extracted by larger scale models (a procedure called downscaling in this paper). Subsequently fw and μ were used to parameterize daily rainfall extremes, assuming an exponential distribution of daily depths and a parametric form of IDF curves based on fw and μ. Under present climate the statistical methodology was trained using rain gauge observation and ERA5 reanalysis. The methodology was then used as an extrapolation of the future, where climate covariates were obtained from CMIP6 simulations.
The statistical methodology presented here is definitely of interest to the hydrological community, particularly for applications involving design against extremes. However I do find that the paper needs substantial improvements to be suggested for publication.
Major comments:
- The first major concern about the paper is the presentation of the methodology. While the methodology itself is straightforward, its presentation is not clear. Large parts of the methodological description point to previous publications from the lead author, instead of actually describing the method, making it very hard to follow. The use of R-code references and jargon in the methodology itself is also not particularly helpful. Very important information that needs to be much more clearly presented is: (a) what is the spatial resolution of each of the data sets? (b) which covariates were used from the climate models and at which spatial/temporal scales? (c) Why were those covariates chosen? This information needs to be part of the main manuscript and not the supplementary information.
- GCMs and ERA5 do not have the same spatial/temporal scales. Most important ERA5 is a data assimilation scheme that integrates observations, in contrast GCMs. The study needs to address how information transfer from ERA5 to the rain gauge scale is similar to that from the GCM to rain gauge scale. Particularly as ERA5 carries a lot of information, not only from modelled atmospheric/land-surface/ocean dynamics, but also in-situ and remote sensing observations.
- I agree with the first reviewer that the term downscaling might be confusing in the context of this study. Downscaling would in most cases refer to methodologies refining/disaggregating rainfall from large to smaller scales. What is presented here is the estimation of fine spatial scale statistics based on climate models. The authors need to clarify that early in the manuscript.
- Extreme rainfall statistics were derived under the assumption that rainfall is exponentially distributed. It is known that an exponential distribution with light tails can significantly underestimate extremes. The authors need to better support this decision (beyond mathematical tractability).
- An assessment of the performance of the exponential distribution, as well as the skill of the parametric form of the IDF curve needs to be presented in the main manuscript for the entire study domain.
- I am not particularly confident that the structure of the supporting information is helpful for understanding the study. I would strongly encourage the authors to separate the code development/application part in a technical reference from the remaining of the supporting material.
Citation: https://doi.org/10.5194/egusphere-2024-1463-RC3
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
153 | 54 | 19 | 226 | 33 | 8 | 8 |
- HTML: 153
- PDF: 54
- XML: 19
- Total: 226
- Supplement: 33
- BibTeX: 8
- EndNote: 8
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1