the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Model calibration and streamflow simulations for the extreme drought event of 2018 on the Rhine River Basin using WRF-Hydro 5.2.0
Abstract. Recent drought events have significantly affected navigation through the Rhine River and the transportation of goods due to low water levels. It has become imperative to analyze the conditions in which these events occur to establish actions to develop adaptive measures and avoid monetary losses. The main focus of this study is to calibrate the hydrological model for extremely low water levels and to test its performance during the 2018 drought event in the Rhine River basin. WRF-Hydro was developed to complement the land-atmosphere interactions with the meteorological model Weather Research and Forecasting (WRF), and it has been mainly used to study flood events. In this study, we simulated the Rhine River basin’s streamflow using the meteorological ERA5 reanalysis dataset as input data. The calibration period is 2016–2017, during which the influence of several parameters on the streamflow was evaluated and contrasted with the daily observed discharge values at gauging stations along the river. Land use cover and terrain slope were used to create spatially distributed parameter maps, thus avoiding the calibration process of testing a range of values, substantially reducing computational demands. During calibration, the importance of modeling realistic outflow values of Lake Constance became apparent due to its significant contribution to the upper Rhine basin. However, WRF-Hydro’s lake module resulted in an overly strong dampening of streamflow. Lake Constance was, therefore, represented without the lake module, resulting in more realistic hydrographs and statistical scores. Overall, the calibration and validation process demonstrated that WRF-Hydro is capable of reproducing the variability of the discharge along the Rhine and capturing low water levels observed during the 2018 drought event. These results suggest that WRF-Hydro is a suitable model for analyzing recent and future drought events in the Rhine River basin under different climate conditions.
- Preprint
(1938 KB) - Metadata XML
-
Supplement
(280 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on gmd-2024-134', Anonymous Referee #1, 03 Sep 2024
The comment was uploaded in the form of a supplement: https://gmd.copernicus.org/preprints/gmd-2024-134/gmd-2024-134-RC1-supplement.pdf
-
AC1: 'Reply on RC1', Andrea Campoverde, 10 Nov 2024
The comment was uploaded in the form of a supplement: https://gmd.copernicus.org/preprints/gmd-2024-134/gmd-2024-134-AC1-supplement.pdf
-
AC1: 'Reply on RC1', Andrea Campoverde, 10 Nov 2024
-
RC2: 'Comment on gmd-2024-134', Anonymous Referee #2, 23 Sep 2024
This paper tries to setup a simulation system based on WRF-Hydro Hydrological model to simulate the extreme drought event along the Rhine in 2018.
the ERA5 reanalysis dataset was used as input data to simulate the runoff in the Rhine River Basin. During the calibration period from 2016 to 2017, the influence of several parameters on the runoff was evaluated, and spatially distributed parameter maps were created. During the calibration process, it was found that the outflow values of Lake Constance had a significant contribution to the upper Rhine Basin. However, the lake module of WRF-Hydro had an overly strong damping effect on the runoff. Therefore, the lake module was not used in the subsequent analysis.
This paper also introduces the scope and characteristics of the Rhine River Basin, as well as the configuration and calibration process of WRF-Hydro, including the calibration of hydrological parameters and the testing of the lake scheme. The results show that the model can capture the low water level in an acceptable manner, and the soil moisture analysis further proves the realism of the model. Although the lake scheme of WRF-Hydro has some shortcomings, the model is still suitable for analyzing recent and future drought events in the Rhine River Basin. In the future, the model will be applied to the climate-hydrological analysis of artificial but realistic extreme drought events.
This paper clearly and concisely introduced the application of WRF-Hydro in drought simulation in the Rhine River Basin, discussed the calibration of different parameters and their impact on the model results, and especially explored the influence of the activation of the lake module on the discharge process. While there still remains some major questions need to be discussed:
-
2018 is the target year for the study, but during the parameter calibration, 2018 was also used as the validation period. Usually, it might be better to set the validation period and the target year separately. This way, the validated model parameters can be better applied to the simulation of the target period; this is especially true if the model is used for forecasting.
-
The year 2018 is the target year of the study in the paper, and an extreme drought happened in this year. In the article, the years 2016 - 2017 were used as the time for parameter calibration, but there was no extreme drought during these two years, which will inevitably lead to the calibrated parameters may not be applicable. This may also be one of the reasons why the results are not ideal. Additionally, the periods of model calibration and validation may be longer for a better and stable result.
-
The final simulation results do not seem ideal. Usually, an NSE result above 0.7 would be better, but in the paper, many of the results are below 0.5, and even there are some negative values. Although the hydrological situation in a larger basin is indeed more complex and difficult to simulate perfectly, is it possible to make further attempts to adjust the results better?
-
The ERA5 data was used to drive the model operation, but it was also used as observational data to verify the model results, which is usually not recommended. It is advisable to see if there are other data, especially observational data, for verification, so that the resultswould be more reliable and convincing.
-
The title of this paper is "Model calibration and streamflow simulations for the extreme drought event of 2018 on the Rhine River Basin using WRF-Hydro 5.2.0", but there is not much analysis in the article on the simulation results of the extreme drought event in 2018. More space is devoted to the calibration of model parameters and the discussion of parameters related to the lake process. It is suggested to analyze in detail this extreme drought event and the model's simulation of this event.
Citation: https://doi.org/10.5194/gmd-2024-134-RC2 -
AC2: 'Reply on RC2', Andrea Campoverde, 10 Nov 2024
The comment was uploaded in the form of a supplement: https://gmd.copernicus.org/preprints/gmd-2024-134/gmd-2024-134-AC2-supplement.pdf
-
-
RC3: 'Comment on gmd-2024-134', Anonymous Referee #3, 25 Sep 2024
The manuscript presents an application of the WRF-Hydro model for the Rhine River basin focusing on the model’s ability to reproduce the river flow dynamics during the drought event of 2018. The model was calibrated for the period 2016-2017. Additionally, the upper soil moisture was compared to ERA5-Land. The manuscript concludes about decent result of model calibration and drought event simulation. It was found that the lake simulation scheme implemented in WRF-Hydro deteriorates the results. When disabled, higher statistical scores are achieved. The manuscript is well-structured and concisely written.
I have major concern related to the scientific novelty of this study. In L107, the authors motivate their study by the fact that the two-way coupled WRF/WRF-Hydro has not been tested for the River Rhine. I find this motivation too weak. Neither from methodological point of view, nor from the case study perspective I found significant advances in model development and application. The profound review presented by the authors clearly demonstrates that many other hydrological models have been previously setup and applied in the Rhine basin. Also, WRF-Hydro has been setup, calibrated and applied in many different catchments, though seemingly not in the Rhine basin. The calibration methodology seems to be based on a two-stage parameter adjustment across their plausible ranges. This sounds like a very standard procedure applied in many modelling study.
The fact that the lake scheme did not work properly and needs to be switched off is not deeply investigated and critically discussed. Why does it lead to poorer results? How can it be potentially improved? It is not sufficient to point out to previous studies that identified the same flaw.
Overall, I am not convinced that this study presents a significant model advancement. Neither, it is clear what we can learn from the presented application. I regret to suggest the rejection of the manuscript. Maybe the authors could develop a more appealing test case for the Rhine basin and consider a journal focusing on regional studies, e.g. Journal of Hydrology: Regional Studies. Alternatively, a methodological improvement focusing on developing a more profound lake and reservoir scheme seems promising.
Citation: https://doi.org/10.5194/gmd-2024-134-RC3 -
AC3: 'Reply on RC3', Andrea Campoverde, 10 Nov 2024
The comment was uploaded in the form of a supplement: https://gmd.copernicus.org/preprints/gmd-2024-134/gmd-2024-134-AC3-supplement.pdf
-
AC3: 'Reply on RC3', Andrea Campoverde, 10 Nov 2024
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
252 | 90 | 64 | 406 | 25 | 3 | 3 |
- HTML: 252
- PDF: 90
- XML: 64
- Total: 406
- Supplement: 25
- BibTeX: 3
- EndNote: 3
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1