A high-resolution perspective of extreme rainfall and river  flow under extreme climate change in Southeast Asia

Hariadi, Mugni Hadi; van der Schrier, Gerard; Steeneveld, Gert-Jan; Sutanto, Samuel J.; Sutanudjaja, Edwin; Ratri, Dian Nur; Sopaheluwakan, Ardhasena; Klein Tank, Albert

doi:https://doi.org/10.5194/hess-28-1935-2024

Articles | Volume 28, issue 9

https://doi.org/10.5194/hess-28-1935-2024

© Author(s) 2024. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/hess-28-1935-2024

© Author(s) 2024. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 28, issue 9

Research article

|

06 May 2024

Research article |

| 06 May 2024

A high-resolution perspective of extreme rainfall and river flow under extreme climate change in Southeast Asia

Mugni Hadi Hariadi, Gerard van der Schrier, Gert-Jan Steeneveld, Samuel J. Sutanto, Edwin Sutanudjaja, Dian Nur Ratri, Ardhasena Sopaheluwakan, and Albert Klein Tank

Download

Final revised paper (published on 06 May 2024)
Supplement to the final revised paper
Preprint (discussion started on 16 Jan 2023)
Supplement to the preprint

Interactive discussion

Status: closed

RC1: 'Comment on hess-2023-14', Anonymous Referee #1, 07 Feb 2023

Review on “A high-resolution perspective of extreme rainfall and river flow

under extreme climate change in Southeast Asia” by Hariadi et al.

The manuscript provided a synthesis on the streamflow response under HigResMIP climate change scenario. Southeast Asia was chosen for testing the hydrological model called PCR-GLOBWB. The authors focused on how the extreme rainfall may influence the changes in low and high flow events. The manuscript may need a substantial revision before publication in HESS journal.

The message conveyed by the authors was not clear, does extreme rainfall influence changes in low and high flows, or increased low flow under climate change scenario? Or something else. I suggest the author to better rephrase it throughout.
The research gap being addressed in the current form seemed too weak and not clear. Para 2 and 5 in the introduction may summarized previous facts. But it still not clears why the research is important to do. Also, does it any benefit using CMIP for the analysis compared to other datasets.
It seemed that the authors would like to deliver the message on the importance of catchment properties to flows dynamics under climate change scenario (L15). But it may not be supported by strong methods and findings. Is there any calibration/validation on the hydrological model? If yes, which stations were used for this? Findings in Figs 3-6 may not show the changes in low/high flow, but it indicates the changes in water depth/storage in the river networks. River discharge is measured in fixed station not along the river network.
I appreciate that authors mentioned about uncertainty of the work. My question is how the uncertainty influences the conclusion/findings. Better elaborate on it will improve the readability of the manuscript.
Please elaborate what PCR-GLOBWB model can do and how does it work? Are there any assumptions for the model.
L131-137 not part of data, I suggest the authors to provide new sub-section about statistical analysis.
Hydrological drought was firstly mentioned in L55, which may not coherence with the previous paragraph. Please revise to better connect with the previous one. Does hydrological drought become a focus of study? I think is not as not many supports afterwards.
L386-395 shall be in the method section to represent the catchment properties influence. Also, how steep or shallow hydrograph was determined was not found anywhere in the method.

L2 does any climate change is not extreme?

L9 more prominent, how much?

L129 the authors used Hamon method for calculating potential ET may need elaboration (e.g. why not other methods?) the method may not commonly used n Southeast Asia.

L145 aggregated?

L158 put a comma after Indonesia

L534 PhD thesis?

Citation: https://doi.org/10.5194/hess-2023-14-RC1

AC1: 'Reply on RC1', Mugni Hariadi, 13 Mar 2023

We would like to thank the reviewer for valuable suggestions and comments. In this document, Prefers to the page number and L refers to the line number in the recent paper. For example, P3L65-70, refers to page 3, lines 65-70.

General comment: The manuscript provided a synthesis on the streamflow response under HigResMIP climate change scenario. Southeast Asia was chosen for testing the hydrological model called PCR-GLOBWB. The authors focused on how the extreme rainfall may influence the changes in low and high flow events. The manuscript may need a substantial revision before publication in HESS journal.

reply: We would like to point out that this manuscript is not intended to test the performance of the PCR-GLOBWB hydrological model but we used the model as a tool to simulate the streamflow. This manuscript focused on the impact of extreme climate change (RCP 8.5) on extreme rainfall and furthermore on the low and high streamflow. The model validations are available from the previous paper by Sutanudjaja et al. (2018). Previous studies by Trambauer et al., (2014) and Ward et al., (2013) show that PCR-GLOBWB is reliable for extreme studies.

Major comments:

No	Comment	reply
1	The message conveyed by the authors was not clear, does extreme rainfall influence changes in low and high flows, or increased low flow under climate change scenario? Or something else. I suggest the author to better rephrase it throughout.	We understand the reviewer's concern about the message of the paper. However, we stated our main finding in the abstract “the impact of climate change is more prominent in a low flow event than in a high flow” (P1L9-10). Moreover, we also discuss this in Section 4.2, e.g., ”the impact of changes in climate indices in SEA also affects the changes in hydrological extremes. The increase of CDD over Northern MEB results in declining LFD over the central part of MEB” (P12). We will change the title of the sub-chapter for clarification.
2	The research gap being addressed in the current form seemed too weak and not clear. Para 2 and 5 in the introduction may summarized previous facts. But it still not clears why the research is important to do. Also, does it any benefit using CMIP for the analysis compared to other datasets.	The reviewer is correct that paragraphs 2 and 5 summarized the previous studies and this study differs compared to those in the sense that we used the latest version of CMIP model results (HighResMIP) and the PCRGLOBWB model to simulate the hydrological extremes. The use of HighResMIP models generates more accurate monsoon characteristics and extreme rainfall results compared to the downscaling result of CMIP5 (CORDEX) (Hariadi et al. (2021,2022)). We will add this statement to address this feedback.
3	It seemed that the authors would like to deliver the message on the importance of catchment properties to flows dynamics under climate change scenario (L15). But it may not be supported by strong methods and findings. Is there any calibration/validation on the hydrological model? If yes, which stations were used for this? Findings in Figs 3-6 may not show the changes in low/high flow, but it indicates the changes in water depth/storage in the river networks. River discharge is measured in fixed station not along the river network.	The importance of catchment properties to streamflow changes due to climate change is not the main message of our paper. The objective of our paper is to evaluate the changes in future extreme rainfall and streamflow. However, we found that not all river basins in SEA follow the changes in extreme rainfall and our result suggests that the catchment properties/memory may play a significant role here. Therefore, we discuss the importance of catchment properties in the discussion. We will revise our manuscript accordingly thus the main message of our is clearly stated and the discussion on catchment properties will not overcome the main message. We used the river recession constant (C) to indicate the catchment properties. The method and the map of the recession constants are available in the supplement material (Supplementary 7). The validation of the PCR-GLOBWB hydrological model is available in the previous paper by Sutanudjaja et al. (2018) and in Section 2.2.2 (P4L131-137). The river discharge (streamflow) is calculated using the kinematic wave for the routing method (P4L130), which allows flow and area to vary both spatially and temporally within a conduit. Thus, the PCR-GLOBWB simulates the river discharge in m3/s for all river networks. We will add this statement in the revised version.
4	I appreciate that authors mentioned about uncertainty of the work. My question is how the uncertainty influences the conclusion/findings. Better elaborate on it will improve the readability of the manuscript	The analysis is based on one climate change scenario (RCP 8.5) which means that the result is more applicable to this scenario. Although, IPCC (2021) reported a relatively small difference between RCP scenarios in the near future (until 2050) rainfall projection. In addition, the hydrological simulation is excluded the water demand, which mean is not consider the change in water demand in the future. This is because we would like to focus on climate forcing. We will elaborate more on how the uncertainty influences the conclusion/findings in the manuscript.
5	Please elaborate what PCR-GLOBWB model can do and how does it work? Are there any assumptions for the model.	We will add more explanations of the PCR-GLOBWB model in chapter 2.2.2. The PCR-GLOBWB is essentially a leakybucket type of model [Bergström, 1976] applied on a cell-by-cell basis. Daily for each grid cell, PCR-GLOBWB calculates the water storage in two vertically stacked soil layers (max. depth 0.3 and 1.2 m) and an underlying groundwater layer, as well as the water exchange between the layers and between the top layer and the atmosphere (rainfall, evaporation and snow melt).
6	L131-137 not part of data, I suggest the authors to provide new sub-section about statistical analysis.	L131-137 is the validation of the PCR-GLOBWB hydrological model summarized in the previous paper (Sutanudjaja et al., 2018). We will make it clear in the manuscript that the detailed validation is available from the previous paper and it is publicly available.
7	Hydrological drought was firstly mentioned in L55, which may not coherence with the previous paragraph. Please revise to better connect with the previous one. Does hydrological drought become a focus of study? I think is not as not many supports afterwards.	We thank you for the useful feedback. Indeed, the reviewer is correct that this study analyzes the low flow and not necessarily drought. We will revise this part.
8	L386-395 shall be in the method section to represent the catchment properties influence. Also, how steep or shallow hydrograph was determined was not found anywhere in the method.	L386-395 is brief information for the river recession constant. The detailed method for how calculating the river recession constant is presented in the supplement material (Supplementary 8).

Small comments:

No	Small comment	Reply
1	L2 does any climate change is not extreme?	We refer the extreme climate change to RCP 8.5
2	L9 more prominent, how much?	We used prominent to highlight that the change of low flow is higher compared to high flow. The value is available in chapter 3.2 (P9L253-308)
3	L129 the authors used Hamon method for calculating potential ET may need elaboration (e.g. why not other methods?) the method may not commonly used n Southeast Asia.	We only used rainfall and temperature in the analysis, this limit us to using more advanced methods like the Penman-Monteith method.
4	L145 aggregated?	The number of events is cumulated per year.
5	L158 put a comma after Indonesia	We will add the comma
6	L534 PhD thesis?	It is a master's thesis. We will revise the manuscript

Citation: https://doi.org/10.5194/hess-2023-14-AC1

RC2: 'Comment on hess-2023-14', Anonymous Referee #2, 13 Feb 2023

Review on hess-2023-14 “A high-resolution perspective of extreme rainfall and river flow under extreme climate change in Southeast Asia” by Hariadi et al

General comments:

Based on HighResMIP CMIP6 models this study came up with the projection of extreme rainfall and river flow under extreme future climate change over Southeast Asia. While the study is useful, I have some concerns that the authors need to consider and address.

Specific comments

1. I think the literature review was not broad enough. As far as I know CORDEX – SEA has carried out multimodels downscaling projection over Southeast Asia. In fact, according to one of their papers (Tangang et al. 2020-- Climate Dynamics,55, 1247-1267), the projected future changes are consistent with what described in this study. However, I find there was lack of discussion on the comparison of the two simulations i.e. CORDEX-SEA vs HighRESMIP.

2. The mean rainfall and extremes are very seasonally dependent. Yet, this study only considers annual extremes. I propose the authors to consider the changes in different seasons in addition to annual timescale.

3. This study appears to have lack of model validations. I think it is useful to validate the HighResMIP models in terms of their ability in simulating the seasonal mean climate, including the monsoonal circulations.

4. The study used bias-corrected outputs to evaluate the projected changes. Yet, there was no analysis (other than Fig 1) on how bias-correction changes the future projected as compared to the raw outputs. It seems from a recently published paper ( Ngai et al. 2022 -- Weather and Climate Extremes, 37, 100484) inconsistency in the magnitude and direction of future change can occur between the bias-corrected and the raw outputs of CORDEX-SEA simulations. The authors need to provide information how bias-correction can affect the projected climate change signals in HighResMIP models.

5. I also find lack of validation in the river flow simulation during the historical period. I think this is needed before the models can be confidently used for projections.

Minor comment:

1. I suggest the authors to use “Malay Peninsula” instead of “Malaysian Peninsula”. Alternatively “Peninsular Malaysia” can be used if the authors was referring to the west Malaysia. However, if the landmasses they are referring to include the southern Thailand then “Malay Peninsula” would be most appropriate.

Citation: https://doi.org/10.5194/hess-2023-14-RC2

AC2: 'Reply on RC2', Mugni Hariadi, 13 Mar 2023

We would like to thank the reviewer for the comments, valuable suggestions, and acknowledgement of this study. In this document, P

refers to the page number and L refers to the line number in the recent paper. For example, P3L65-70 relates to page 3, lines 65-70.

No	Comments	reply
1	I think the literature review was not broad enough. As far as I know CORDEX – SEA has carried out multimodels downscaling projection over Southeast Asia. In fact, according to one of their papers (Tangang et al. 2020-- Climate Dynamics,55, 1247-1267), the projected future changes are consistent with what described in this study. However, I find there was lack of discussion on the comparison of the two simulations i.e. CORDEX-SEA vs HighRESMIP.	The comparisons of CORDEX-SEA vs HighRESMIP are conducted in our previous paper (Hariadi et. al., 2021 and 2022). We will elaborate more on the comparison between CORDEX-SEA vs HighResMIP in the revised version.
2	The mean rainfall and extremes are very seasonally dependent. Yet, this study only considers annual extremes. I propose the authors to consider the changes in different seasons in addition to annual timescale.	We do agree with the reviewer that means rainfall and extremes are very seasonally dependent. However, in this paper, we would like to focus on the general changes in extreme events for both rainfall and streamflow.
3	This study appears to have lack of model validations. I think it is useful to validate the HighResMIP models in terms of their ability in simulating the seasonal mean climate, including the monsoonal circulations.	The performance of the HighResMIP model and the comparison between monsoon characteristics and rainfall indices are available in the previous paper (Hariadi et. al., 2021 and 2022). We will summarize the findings in the introduction (P3L64-75).
4	The study used bias-corrected outputs to evaluate the projected changes. Yet, there was no analysis (other than Fig 1) on how bias-correction changes the future projected as compared to the raw outputs. It seems from a recently published paper (Ngai et al. 2022 -- Weather and Climate Extremes, 37, 100484) inconsistency in the magnitude and direction of future change can occur between the bias-corrected and the raw outputs of CORDEX-SEA simulations. The authors need to provide information how bias-correction can affect the projected climate change signals in HighResMIP models.	We thank the reviewer for the feedback. We used the same bias correction method as Ngai et al. (2022), so the impact of bias correction on the simulations can be expected to be significant, as Ngai et al. (2022) showed. In addition, the objective of the paper is to document the changes in hydrology. Quantifying the impact of bias correction is out of the scope of this study. Furthermore, we will refer to Ngai et al. (2022) to show the impact of bias correction on the simulations in the revised manuscript.
5	I also find lack of validation in the river flow simulation during the historical period. I think this is needed before the models can be confidently used for projections.	The detailed validation of the PCR-GLOBWB hydrological model is available in the previous paper by Sutanudjaja et al. (2018). We refer to this in P5L131-137. In addition, previous studies by Trambauer et al., (2014) and Ward et al., (2013) show that PCR-GLOBWB is reliable for extreme studies.

Small comment:

I suggest the authors to use “Malay Peninsula” instead of “Malaysian Peninsula”. Alternatively “Peninsular Malaysia” can be used if the authors was referring to the west Malaysia. However, if the landmasses they are referring to include the southern Thailand then “Malay Peninsula” would be most appropriate.

Reply:

We thank the reviewer for the feedback. We agree with the reviewer and thus we will use “Peninsular Malaysia” for referring west Malaysia in the revised manuscript

Citation: https://doi.org/10.5194/hess-2023-14-AC2

Peer review completion

AR: Author's response | RR: Referee report | ED: Editor decision | EF: Editorial file upload

ED: Reconsider after major revisions (further review by editor and referees) (21 Apr 2023) by Yi He

Dear authors,

Thank you for responding to the reviewers’ comments. Before I can proceed with a decision, I would like to suggest you revise your responses to the follows.

RC1 #4 regarding uncertainties.
There should be three main sources of uncertainties when quantifying climate change on streamflows. They are climate forcing inputs, hydrological model structure, hydrological model parameters. Other uncertainties can include water demand but that is not relevant when modelling natural streams or when you are assuming no anthropogenic interference. Your manuscript has discussed only uncertainties coming from the forcing inputs and hydrological model structure. The equifinality of hydrological model parameters is another important source of uncertainty.

RC2 #2 We do agree with the reviewer that means rainfall and extremes are very seasonally dependent. However, in this paper, we would like to focus on the general changes in extreme events for both rainfall and streamflow.
RC2 made a valid point here that seasonal changes are important to consider and would add value to your manuscript.

RC2 #5
You stated: In addition, previous studies by Trambauer et al., (2014) and Ward et al., (2013) show that PCR-GLOBWB is reliable for extreme studies.
Trambauer et al., (2014) Comparison of different evaporation estimates over the African continent does not seem to show PCR-GLOBWB is reliable for extreme studies.
Ward, P., Dettinger, M., Jongman, B., Kummu, M., Sperna Weiland, F., and Winsemius, H.: Flood risk assessment at the global scale-the role of climate variability, in: EGU General Assembly Conference Abstracts, pp. EGU2013–1390, 2013
This is an abstract and insufficient to support your argument here.
Please provide strong evidence to show that PCR-GLOBWB is reliable for extreme studies.

Sincerely,
Yi He, HESS Editor

Hide

AR by Mugni Hariadi on behalf of the Authors (29 May 2023) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (12 Jun 2023) by Yi He

RR by Anonymous Referee #3 (19 Nov 2023)

Suggestions for revision or reasons for rejection

This study provides a synthetic analysis on the extreme rainfall and river flow in response to climate change using HighResMIP and a high-resolution hydrological model. The numerical experiments are well designed. The authors have done a good job of synthesizing the available studies when interpreting their results. This study is potentially valuable, but I do have a few concerns.

The message conveyed in the introduction is not clear enough. In particular, the science question is well described, which however is not connected to the motivation very well. Based on my reading, it seems that the motivation is to use CMIP6, as a successor of CMIP5 and CORDEX-SEA, to investigate the impact of climate change, although this science question has been addressed previously and extensively. This gives me the impression that this study is a modeling study instead, in which I am expecting to see more analysis on the improved performance of CMIP6 over its predecessors. I suggest the authors either think about adding new experiments (maybe in specific regions) to compare the results of using different climate forcing, or elaborate the motivation in a more clarified and concise way.

There lacks a validation on the hydrological simulations. I understand that the model was previously validated in the same regions. Are those simulations using the same forcing as your study? It would be helpful to clarify this, as the hydrological simulation depends significantly on the atmospheric forcing being used.

In the discussion (especially section 4.1), I appreciate the authors’ thorough analysis against other available studies in the same region. When making comparison, it may be even better to explain what factors may cause the difference between your results and others’. What are the relevant uncertainties? There is a separate section on the sources of uncertainty. But that is more like a general discussion.

Figures need reorganization. Specially, in figure 1, are there statistical quantities (in addition to whisker plot) that you can use to better demonstrate the improvement from the bias correction? In the second column of figure 2, the range of colormaps need to be adjusted to improve readability. Some figures relevant to the discussion (such as those in section 4.3) need to be moved from the supplement to the main context.
Line 297, It is more appropriate to either move the figures relevant to the discussion here or move the discussion to supplement.
In section 2.1, a map with some texts showing geographic regions is recommended.

Minor comments:

Section 2.2.1, when describing the difference among climate data (e.g., CORDEX-SEA and HighResMIP), it may be good to mention their corresponding resolution.
“historically forced atmosphere run of HighResMIP (HighResSST), the Hist-1950” What is the difference? A bit more elaboration is helpful.
Line 125, "high skill" is weird wording and is not accurate.

Line 131, It is acutually possible for GCMs to simulate at high resolutions. So it may be good to delete "what is possible".

Line 148, “The model runs in 5 arcmins spatial resolution, which is about 10 km by 10 km at the equator.” This does not seem to be a high resolution for GHMs as claimed early in this paragraph.

Line 150, What remapping method do you use?

Line 168-173, This paragraph is not necessary. Or you may want to move it upper.

Line 200, “DJF”. This is defined later in page 10.

Line 222, “is” -> does

Line 223, delete “that”

Line 247, How is the probability calculated?

Line 256, Please rewrite this sentence to improve clarity.

Line 321, This explanation is unnecessary for the readers of HESS.

Line 323, The change of probablity can be up to 200%? Same question for figure 4-6. It is useful to define how the probability is calculated more clearly in the methodology section.

Hide

ED: Publish subject to minor revisions (review by editor) (28 Nov 2023) by Yi He

AR by Mugni Hariadi on behalf of the Authors (15 Jan 2024) Author's response Author's tracked changes Manuscript

ED: Publish subject to minor revisions (review by editor) (05 Feb 2024) by Yi He

AR by Mugni Hariadi on behalf of the Authors (14 Feb 2024) Author's response Author's tracked changes Manuscript

ED: Publish as is (26 Feb 2024) by Yi He

AR by Mugni Hariadi on behalf of the Authors (05 Mar 2024)

Short summary

We utilize the high-resolution CMIP6 for extreme rainfall and streamflow projection over Southeast Asia. This region will experience an increase in both dry and wet extremes in the near future. We found a more extreme low flow and high flow, along with an increasing probability of low-flow and high-flow events. We reveal that the changes in low-flow events and their probabilities are not only influenced by extremely dry climates but also by the catchment characteristics.