the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Evaluating future hydrological changes in China under climate change
Abstract. Projecting and understanding future hydrological changes in China are critical for effective water resource management and adaptation planning in response to climate variability. However, few studies have investigated runoff variability and flood and drought risks under climate change scenarios for the entire region of China at high resolution. In this study, we use the Joint UK Land Environment Simulator (JULES), specifically tailored for simulating hydrological processes in China at a 0.25-degree resolution. Downscaled and bias-corrected forcing data from Global Climate Models (GCMs), using the bias-correction and spatial disaggregation (BCSD) method, were used to drive the JULES model to project future hydrological processes under medium (SSP245) and high (SSP585) emission scenarios. The results indicate that annual runoff in China is projected to increase significantly under the high emission scenario, notably in the eastern and southern basins. Wetter summers and drier winters are expected in the south, while the opposite trend is expected in the north. Wetter conditions in the near future and drier summers in the far future are expected in northern China. Shifts from drier to wetter conditions are projected in the southeast and southwest areas, while the middle Yangtze River basin may experience the opposite trend. The flood risk is expected to increase in spring, summer, and autumn, along with heightened drought risk in winter, summer, and autumn. Southern China would face greater flood risk, while the central Yangtze River basin would face intensified drought risk, especially in the far future. These findings underscore the influence of different emission scenarios on flood and drought risks, emphasizing the need for proactive measures to enhance climate adaptation in the future.
- Preprint
(5523 KB) - Metadata XML
-
Supplement
(4369 KB) - BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on hess-2024-166', Anonymous Referee #1, 01 Jul 2024
-
AC1: 'Reply on RC1', Danyang Gao, 27 Aug 2024
Dear Reviewer,
Thank you very much for your detailed and insightful comments on the manuscript. We have prepared a point-by-point response to all your comments. Please kindly refer to the supplement document attached, which contains the detailed response and supporting references.
Thank you very much!
Best regards,
Danyang Gao
-
AC1: 'Reply on RC1', Danyang Gao, 27 Aug 2024
-
RC2: 'Comment on hess-2024-166', Anonymous Referee #2, 01 Jul 2024
This study applies ‘the Joint UK Land Environment Simulator (JULES)’ to simulate and project runoff in China at a high resolution; and further analyze the flood and drought risks. The authors claim that 1) annual runoff in China is projected to increase significantly, notably in eastern and southern basins; 2) northern China is expected to have wetter conditions in the near future and drier summers in the far future; 3) southern China is projected to face greater flood risks; 4) central Yangtze River basin can face intensified drought risk. Overall, the idea of using the JULES to project runoff in China is kind of good; However, the whole story is really boring and unclear, and there are some technique problems exist. Overall, I consider this manuscript cannot be published at such a good journal at this stage.
Â
Â
Major concerns:
- In abstract, I cannot see any logic in reporting the results but only redundancy. The authors only say e.g., annual runoff increase; southern China runoff increase…; all of them is only qualitative description without a single robust number to show the results; besides, no mechanisms at all to reveal why the runoff is projected to change like that. Because of increasing precipitation? Evapotranspiration? This makes the whole story really boring! And I cannot even see a clear storyline to show how the spatial patterns of runoff changes in China!
- Line 116: a reference period from 1959-2014 were chosen to generate two CDFs. This reference period contains 65 years in total and involves significant warming and is not stationary!! How can the authors choose such a long time period as a reference period? The typical time period only involves 20-30 years to ensure the stationarity of the data!
- The authors only use six CMIP6 GCMs to downscale and simulate and project river runoff. I fully cannot understand why they need to downscale themselves? The ISIMIP 3b already provide high-resolution downscale climate output (https://www.isimip.org/newsletter/isimip3a3b-protocol-published/), they also use ERA5 as the reference dataset. And I consider their bias correction methodology is even better than the authors did. If the authors use more than 15 models to specifically consider the uncertainty and associated source, then it is fine; But the authors only use 6 models while ISIMIP3b provide 5 model outputs with really good bias correction, then why the authors perform bias correction themselves? If they do, I need to see the superiority of their bias correction results comparing to the ISIMIP 3b. (The ISIMIP 3b is a publicly accessible dataset and without comparison, I would like to trust them more comparing the authors did).
- In Lines 157-159: there is a sharp decrease with regard to the hydrological model performance. Is it because of over-fitting? Or what reasons cause this? Why the authors do-not use cross-validation (e.g., Arsenault et al., 2018) to calibrate the hydrological models?
Arsenault, R., Brissette, F., & Martel, J. L. (2018). The hazards of split-sample validation in hydrological model calibration. Journal of hydrology, 566, 346-362.
- In Lines 160-162: ‘Most stations with good performance are large rivers, indicating that the model simulates better in large rivers. This is really unreasonable! Large rivers involve reservoirs and hydraulic engineering, land cover and land type changes, which can evidently affect runoff regimes and is more difficult to simulate, irrespective of model resolution. This is why most previous studies focus on catchment scale and especially small catchments. The JULES model also does not consider land use and land cover change, nor the impacts from reservoirs, how can the performance be better in large catchments than small ones?
- Most importantly! All calibration and validation of runoff are based on a monthly scale, how can the authors then use the calibrated models to simulate daily runoff (e.g., Figs. 7-8)?! I cannot trust the results in this case. If the authors need to analyze the daily runoff variations, they need to train and validate the models at the daily scale.
- In 278-285: again, I consider the authors really donot know basic concepts of hydrology. 1) if using monthly data to calibrate the hydrological model, then the calibrated models cannot be used to simulate daily runoff! Nor can they be used to simulate and project floods which are measured at the daily scale! 2) the low flows, e.g., 10th percentile runoff, cannot be directly used to indicate droughts! The drought episode is typically defined as the abnormally runoff deficits persist for a long time! Not the change of low flow!
Â
Â
Specific concerns:
- Figs 2-3, please change the unit of m3/s to mm by dividing the discharge area.
- In all figures, the labels and characters are really too small.
Citation: https://doi.org/10.5194/hess-2024-166-RC2 -
AC2: 'Reply on RC2', Danyang Gao, 27 Aug 2024
Dear Reviewer,
Thank you very much for your detailed and insightful comments on the manuscript. We have prepared a point-by-point response to all your comments. Please kindly refer to the supplement document attached, which contains the detailed response and supporting references.
Thank you very much!
Best regards,
Danyang Gao
Status: closed
-
RC1: 'Comment on hess-2024-166', Anonymous Referee #1, 01 Jul 2024
-
AC1: 'Reply on RC1', Danyang Gao, 27 Aug 2024
Dear Reviewer,
Thank you very much for your detailed and insightful comments on the manuscript. We have prepared a point-by-point response to all your comments. Please kindly refer to the supplement document attached, which contains the detailed response and supporting references.
Thank you very much!
Best regards,
Danyang Gao
-
AC1: 'Reply on RC1', Danyang Gao, 27 Aug 2024
-
RC2: 'Comment on hess-2024-166', Anonymous Referee #2, 01 Jul 2024
This study applies ‘the Joint UK Land Environment Simulator (JULES)’ to simulate and project runoff in China at a high resolution; and further analyze the flood and drought risks. The authors claim that 1) annual runoff in China is projected to increase significantly, notably in eastern and southern basins; 2) northern China is expected to have wetter conditions in the near future and drier summers in the far future; 3) southern China is projected to face greater flood risks; 4) central Yangtze River basin can face intensified drought risk. Overall, the idea of using the JULES to project runoff in China is kind of good; However, the whole story is really boring and unclear, and there are some technique problems exist. Overall, I consider this manuscript cannot be published at such a good journal at this stage.
Â
Â
Major concerns:
- In abstract, I cannot see any logic in reporting the results but only redundancy. The authors only say e.g., annual runoff increase; southern China runoff increase…; all of them is only qualitative description without a single robust number to show the results; besides, no mechanisms at all to reveal why the runoff is projected to change like that. Because of increasing precipitation? Evapotranspiration? This makes the whole story really boring! And I cannot even see a clear storyline to show how the spatial patterns of runoff changes in China!
- Line 116: a reference period from 1959-2014 were chosen to generate two CDFs. This reference period contains 65 years in total and involves significant warming and is not stationary!! How can the authors choose such a long time period as a reference period? The typical time period only involves 20-30 years to ensure the stationarity of the data!
- The authors only use six CMIP6 GCMs to downscale and simulate and project river runoff. I fully cannot understand why they need to downscale themselves? The ISIMIP 3b already provide high-resolution downscale climate output (https://www.isimip.org/newsletter/isimip3a3b-protocol-published/), they also use ERA5 as the reference dataset. And I consider their bias correction methodology is even better than the authors did. If the authors use more than 15 models to specifically consider the uncertainty and associated source, then it is fine; But the authors only use 6 models while ISIMIP3b provide 5 model outputs with really good bias correction, then why the authors perform bias correction themselves? If they do, I need to see the superiority of their bias correction results comparing to the ISIMIP 3b. (The ISIMIP 3b is a publicly accessible dataset and without comparison, I would like to trust them more comparing the authors did).
- In Lines 157-159: there is a sharp decrease with regard to the hydrological model performance. Is it because of over-fitting? Or what reasons cause this? Why the authors do-not use cross-validation (e.g., Arsenault et al., 2018) to calibrate the hydrological models?
Arsenault, R., Brissette, F., & Martel, J. L. (2018). The hazards of split-sample validation in hydrological model calibration. Journal of hydrology, 566, 346-362.
- In Lines 160-162: ‘Most stations with good performance are large rivers, indicating that the model simulates better in large rivers. This is really unreasonable! Large rivers involve reservoirs and hydraulic engineering, land cover and land type changes, which can evidently affect runoff regimes and is more difficult to simulate, irrespective of model resolution. This is why most previous studies focus on catchment scale and especially small catchments. The JULES model also does not consider land use and land cover change, nor the impacts from reservoirs, how can the performance be better in large catchments than small ones?
- Most importantly! All calibration and validation of runoff are based on a monthly scale, how can the authors then use the calibrated models to simulate daily runoff (e.g., Figs. 7-8)?! I cannot trust the results in this case. If the authors need to analyze the daily runoff variations, they need to train and validate the models at the daily scale.
- In 278-285: again, I consider the authors really donot know basic concepts of hydrology. 1) if using monthly data to calibrate the hydrological model, then the calibrated models cannot be used to simulate daily runoff! Nor can they be used to simulate and project floods which are measured at the daily scale! 2) the low flows, e.g., 10th percentile runoff, cannot be directly used to indicate droughts! The drought episode is typically defined as the abnormally runoff deficits persist for a long time! Not the change of low flow!
Â
Â
Specific concerns:
- Figs 2-3, please change the unit of m3/s to mm by dividing the discharge area.
- In all figures, the labels and characters are really too small.
Citation: https://doi.org/10.5194/hess-2024-166-RC2 -
AC2: 'Reply on RC2', Danyang Gao, 27 Aug 2024
Dear Reviewer,
Thank you very much for your detailed and insightful comments on the manuscript. We have prepared a point-by-point response to all your comments. Please kindly refer to the supplement document attached, which contains the detailed response and supporting references.
Thank you very much!
Best regards,
Danyang Gao
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
356 | 113 | 25 | 494 | 77 | 14 | 15 |
- HTML: 356
- PDF: 113
- XML: 25
- Total: 494
- Supplement: 77
- BibTeX: 14
- EndNote: 15
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1