the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Enhanced hydrological modelling with the WRF-Hydro lake/reservoir module at Convection-Permitting scale: a case study of the Tana River basin in East Africa
Abstract. East Africa frequently faces extreme weather events like droughts and floods, underscoring the need for improved hydrological simulations to enhance prediction and mitigate losses. One of the main challenges in achieving this is low-quality of precipitation data and limitations in modelling skills. Due to drought sensitivity, flood proneness and data availability, the upper and middle stream of the Tana River basin was used as a case to address some of the challenge. We performed convection-permitting (CP) simulations using the Weather Research and Forecasting (WRF) model, and utilizing the CPWRF output as a driver we conducted WRF Hydrological modelling (WRF-Hydro) integrated with the lake/reservoir module. The CPWRF precipitation outperforms the ERA5 using IMERG as the benchmark, particularly for the precipitation amount over mountainous regions and light precipitation events (1–15 mm day-1) in the dry seasons. The improved precipitation especially alleviates the peak false, when comparing the well-calibrated lake-integrated model driven by CRWRF output (LakeCal) to that by ERA5, with an NSE increase of 0.53. Additionally, the lake/reservoir module effectively mitigates the model-data bias, especially for dry-season flow and peak flow, when comparing the lake-integrated model (LakeCal) to the model without the lake (LakeNan), with an NSE increase of 1.67. The lake module makes river discharge more sensitive to spin-up time and affects discharge through lake-related parameters. Adjustments to the lake-integrated model’s runoff infiltration rate, Manning’s roughness coefficient, and the groundwater component have minimal impact on the dry-season flows. Dividing by the total NSE increase, hydrological modelling improvement is 24 % and 76 % from CPWRF simulation and lake module, respectively. Our findings highlight the enhanced hydrological modelling capability with the lake/reservoir module and CPWRF simulations, offering a valuable tool for flood and drought predictability in data-scarce regions such as East Africa.
- Preprint
(1990 KB) - Metadata XML
-
Supplement
(2327 KB) - BibTeX
- EndNote
Status: open (until 09 Dec 2024)
-
RC1: 'Comment on hess-2024-278', Anonymous Referee #1, 20 Oct 2024
reply
This manuscript presents an enhanced hydrological simulation using a resolution-refined climate model coupled with a more comprehensive hydrological framework. The improvements achieved through convection-permitting WRF simulations and the integration of lake and reservoir modeling are evident, particularly in addressing biases in peak flow and dry-season flow during discharge simulations. These findings emphasize the enhanced capability of hydrological simulations when using refined climate models and lake-integrated approaches.
The study focuses on two key topics: the added value of convection-permitting models and the integration of hydrological models with lake systems. This work holds great value in addressing challenges related to unreliable hydrological simulations, especially over Eastern Africa. It provides critical benchmarks for optimizing hydrological modeling, contributing to improved flood and drought forecasting and loss reduction in water management applications.
In my opinion, this paper gives an incremental advancement in the hydrological modelling field more than novelty, considering the convection-permitting modelling and lake-integrated hydrological simulation. The paper is well written, and quite comprehensive and well-structured. The topic is of interest and fits the journal scope, I suggest a few major revisions and some minor before publication in HESS. I believe a minor revision is needed and the revised manuscript will be much better. My comments are listed below.
Major comments
- The data for precipitation evaluation is from 2010 to 2014, while the data for discharge is from 2011 to 2014. I suggested using the same time series (i.e. 2010-2014) for precipitation and discharge evaluation.
- If it is possible, longer time period of data for precipitation evaluation is necessary. Using data over 2010-2014 to calculate the monthly precipitation is not enough, which is usually more than 10 years.
- It is not easy to understand how to calculate the attribution of discharge changes from WRF-refined precipitation and lake-integrated WRF-Hydro model in Sect. 4.3. Please add it to the method section.
- The Sect. 4.3 should be put at the front of the Discussion, since it explains why the hydrological simulation improves.
Minor comments
Line17: “limitations in modelling skills” in the abstract might be the defects or imperfections in the model algorithm, such as not involving lake processing. But the term “limitations in modelling skills” also covers the content of the forcing. So I suggest change ”limitations in modelling skill” to “limitations in model capacity”
Line34: Change “limitations of hydrological modelling” to “limited capacity of hydrological model”.
Line57: Change “realistic regional detail” to “the realistic regional detail”
Line58: Change “realistic regional details” to “refined-scale features”
Line67: Change “coarse resolution” to “the coarse resolution”
Line86: Change “suggesting” to “which suggests”
Line107-108: Please add references to the sentence “However, the region faces increasing risks of drought and flood, which are likely exacerbated by climate change. ”
Line118-119: Change “The research is to improve hydrological models for better water resource management and risk mitigation, supporting sustainable practices in regions vulnerable to water-related damages.” to “The research aims to improve hydrological models, which helps to better water resource management and risk mitigation, and supports sustainable practices in regions vulnerable to water-related damages.”
Line129: Please keep the hyphen consistent in “S 1.25°~N 0.50°, E 36.50°-E 39.75°”.
Line131-132: Before “We classified the terrain into mountainous regions above 1,600 meters and plains below 1,600 meters.”, please add “To analyze and evaluate the spatial distribution of precipitation related to the topography,”
Line133-135: Change “There are five reservoirs in the basin and along the Tana River (Table 1, Fig. 1 c). It is worth noting that the Garissa station is downstream Rukanga and the lakes between them are Masinga, Kamburu, Gitaru, Kindaruma, and Kiambere from the upstream to downstream. While the lakes don’t affect the streamflow at Rukanga, they do impact the discharge at Garissa.” to “There are five reservoirs in the basin and along the Tana River (Table 1, Fig. 1 c), including Masinga, Kamburu, Gitaru, Kindaruma, and Kiambere from the upstream to downstream. The five lakes are between Garissa station upstream and Rukanga downstream. It is worth noting that the lakes don’t affect the streamflow at Rukanga while they do impact the discharge at Garissa.”
Line143: Change “boundaries” to “boundary”
Line154: If one year is used as the spin-up year, the subsequent analysis in the results should be based on the simulated precipitation from 2011 to 2014 (which is 2010-2014 in the paper). However, I think the analysis in the results covering 2010 to 2014 is acceptable due to three reasons: 1) one month of spin-up is typically sufficient for WRF downscaling, 2) the time span of the precipitation data used is rather limited, 3) the precipitation analysis or evaluation focus on the MAM and OND, or the extremes. If you can extend the WRF downscaling (for example, from 2001 to 2014), it would be better.
Line367: Change “GIS pre-41.processing” to “GIS pre-processing”
Line376: Change “ in the evolution of discharge ” to “ based on the discharge ”
Line378: There are two things related to the term “peak-flow”. One is the largest simulated daily flow on the date when the largest daily flow occurs based on the observations. Another is the peaks shown in the discharge-date curve. Please distinguish between the two.
Line408: Change “demonstrate” to “demonstrates”
Line411: Please add references (i.e. figures) for “The water levels of the five lakes show the same spin-up time.”
Line411-412: Change “However, a larger lake seems to require more time to reach equilibrium. The lakes are interconnected, so the initialization time is determined by the longest spin-up. Therefore, despite the disparate sizes, the initialization times of the five lakes are the same.” to “Usually, a larger lake seems to require more time to reach equilibrium. Since the lakes are interconnected, the initialization time is determined by the longest spin-up. This results in the same initialization times of the five lakes despite of the dramatically disparate sizes”.
Line413-415:Please delete “ The bias from the LakeRaw simulation is considerable (> 80 %). This is due to that the parameters used in the LakeRaw are from the primarily calibrated LakeNan Model or GIS pre-processing (Methodology), which needs further calibration for the WRF-Hydro system.” Because there is no relationship between this part and the preceding text.
Line492: Please delete “still”.
Line499: Please double-check the figure ”-53”.
Line527: Change “under estimation” to “underestimation”
Line532: The label in Fig. 10 is a bit confusing. Please label them more clearly.
Line551: Change “Hydrological modelling improvement from convection-permitting WRF-simulated precipitation” to “Hydrological modelling improvement from lake/reservoir module”
Line554: Delete “. Factors”.
Citation: https://doi.org/10.5194/hess-2024-278-RC1 -
CC1: 'Reply on RC1', Ling Zhang, 12 Nov 2024
reply
Publisher’s note: this comment is a copy of AC1 and its content was therefore removed on 14 November 2024.
Citation: https://doi.org/10.5194/hess-2024-278-CC1 -
AC1: 'Reply on RC1', Zhang Ling, 14 Nov 2024
reply
Thanks you for your feeback. Please find our response attached.
-
RC2: 'Comment on hess-2024-278', Anonymous Referee #2, 26 Oct 2024
reply
The paper presents a hydrological simulation in a data-scarce region, utilizing a convection-permitting climate model and a lake-integrated hydrological model. It highlights and quantifies the improvements brought by convection-permitting WRF simulations and the inclusion of lake and reservoir modeling. A key contribution of the paper is its identification of specific areas of improvement, particularly at peak and low-flow points, and a detailed explanation of the underlying causes and attributions of these improvements.
The study addresses the added value of convection-permitting modeling in hydrological simulations and the integration of the lake/reservoir module within WRF-Hydro, both of which are topics that have been rarely explored. The findings provide fresh insights into the benefits of using a convection-permitting climate model and the lake/reservoir module, offering valuable benchmarks for optimizing hydrological modeling, especially in regions with limited data availability. As noted by the authors, this enhanced hydrological simulation has potential future applications in forecasting extreme hydrological events, such as floods and droughts.
The paper aligns well with the scope of HESS, as it focuses on hydrological simulation, especially in a data-scarce region. The use of the convection-permitting climate model and lake-integrated hydrological model addresses critical concerns in the climate-hydrology field. Furthermore, the study area, East Africa, is particularly relevant due to the scarcity of data, the complexity of hydrological simulations, and the frequent occurrence of extreme flood and drought events.
In conclusion, this paper offers significant value and is suitable for publication in HESS. However, I have several suggestions and comments for consideration, as outlined below.
Major comments
Although this manuscript is well-structured and comprehensive, with some well-written sections, it requires careful editing by professional English editors. Special attention should be given to sentence structure, as well as minor spelling and grammatical errors, to ensure that the study's goals and results are clear to the reader.
The conclusions should be presented with caution. The sensitivity of the simulated peak flows to the spin-up time, based on a single event in 2011. The conclusion may vary if different regions or other peak flow events are considered. I recommend adding further discussion on this point.
The authors should verify the model configuration. Figure 1 indicates that WRF is directly driven by ERA5, but lines 156-157 suggest a nested simulation domain. Please double-check this for accuracy.
When evaluating the WRF simulation, the authors focus primarily on bias. However, a smaller bias does not necessarily indicate a better simulation. I suggest using additional indices, such as a Taylor diagram, to support the conclusions.
The manuscript emphasizes the importance of CPM in East Africa. If the authors add a discussion of the added values of CPM with respect to its driving forces, this manuscript would be more informative.
Minor comments
Line17-18: Replace “the upper and middle stream of the Tana River basin was” with “the upper and middle streams of the Tana River basin were ”.
Line19-20: This sentence is ambiguous. Please revise it.
Line21: Replace “ using IMERG as the benchmark” with “ when benchmarked against IMERG
Line22-23 Change “alleviates the peak false” to “alleviates the false peak simulation”.
Line24: Change “NSE” with “NSE (Nash-Sutcliffe Efficiency)”.
Lin29-30: There are two terms “lake” and “lake/reservoir” which seem to represent the same thing. It would be better to unify them for consistency throughout the document.
Line29-30: Replace “highlight the enhanced hydrological modelling capability with” with “highlight the enhanced capability of hydrological modelling using”.
Line123: Replace “resulting” with “which results”.
Line129: Please replace “S 1.25°~N 0.50°” with “S 1.25°-N 0.50°”.
Line139: Replace “ the upper and middle stream of the Tana River Basi” with “ the upper and middle streams of the Tana River basin”.
Line:154: Usually, one month spin-up is sufficient for WRF downscaling.
Line182-183: Please add “(with the lake/reservoir module inactive)” behind “without the lake/reservoir module”.
Line185: Replace “of the Garissa discharge.” with “of simulated discharge against the observation at Garissa”.
Line208-210: Replace “which may affect the subsequent sensitivity analysis and hydrological modelling assessments.” with “which may potentially affect the result of subsequent sensitivity analyses and the performance of the hydrological simulation.”
Line260: Replace “For each lake test” with “For each test”.
Line274: Replace “Each lake” with “Each”.
Line306-307: Please change the unit “mm a-1” to “mm”. The unit should be corrected in the whole text.
Line368: Replace “GIS pre-41.processing” with “GIS pre-processing”.
Line408: Replace “demonstrate” with “demonstrates”.
Line411: The view “a larger lake seems to require more time to reach equilibrium.” depends. You should add some references.
Line499: Replace “-53” with “-53%”.
Line527: Replace “under estimation” with “underestimation”.
Line529: “WRF-refined precipitation” is not idiomatic. Please revise it.
Line544: Replace “Woodhams et al.'s research (2018) demonstrates” with “Woodhams et al. (2018) demonstrates”.
Line554: Delete “. Factors”.
Line565: Replace “Arnault et al.,” with “Arnault et al.”
Line574-575: Replace “ lake water levels may be not well presented” with “it shows a limited skill for simulating water level”.
Line577: Please replace “ with r2 ranging from near zero (0.005) to 0.25 for the five lakes.” with “ with r2 of the simulated discharge against the observation at Garissa less than 0.25 for all the five lakes.”
Line599: Replace “a seamless, consistent meteorological-hydrological modelling system” with “a seamless and consistent meteorological-hydrological modelling system”.
Line601-602: The sentence “which makes an NSE increase of 0.53 when comparing LakeCal to LakeCal-ERA5” is not clear. Please correct it.
Citation: https://doi.org/10.5194/hess-2024-278-RC2 -
CC2: 'Reply on RC2', Ling Zhang, 12 Nov 2024
reply
Publisher’s note: this comment is a copy of AC2 and its content was therefore removed on 14 November 2024.
Citation: https://doi.org/10.5194/hess-2024-278-CC2 -
AC2: 'Reply on RC2', Zhang Ling, 14 Nov 2024
reply
Thanks you for your feebacks. Please find our response attached.
-
CC2: 'Reply on RC2', Ling Zhang, 12 Nov 2024
reply
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
207 | 45 | 13 | 265 | 25 | 6 | 4 |
- HTML: 207
- PDF: 45
- XML: 13
- Total: 265
- Supplement: 25
- BibTeX: 6
- EndNote: 4
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1