the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 25 Sep 2023
Possible role of anthropogenic climate change in the record-breaking 2020 Lake Victoria levels and floods
Abstract. Heavy rainfall in East Africa between late 2019 and mid 2020 caused devastating floods and landslides throughout the region. These rains drove the levels of Lake Victoria to a record-breaking maximum in the second half of May 2020. The combination of high lake levels, consequent shoreline flooding, and flooding of tributary rivers caused hundreds of casualties and damage to housing, agriculture and infrastructure in the riparian countries of Uganda, Kenya and Tanzania. Media and government reports linked the heavy precipitation and floods to anthropogenic climate change, but a formal scientific attribution study has not been carried out so far. In this study, we characterise the spatial extent and impacts of the floods in the Lake Victoria basin, and then investigate to what extent human-induced climate change influenced the probability and magnitude of the record-breaking lake levels and associated flooding, by applying a multi-model extreme event attribution methodology. Using remote sensing-based flood mapping tools, we find that more than 29 thousand people living within a 50 km radius of the lake shorelines were affected by floods between April and July 2020. Precipitation in the basin was the highest recorded in at least three decades, causing lake levels to rise by 1.21 m between late 2019 and mid 2020. The flood, defined as a 6-month rise in lake levels as extreme as that observed in the lead-up to May 2020, is estimated to be a 63-year event in the current climate. Based on observations and climate model simulations, the best estimate is that the event has become more likely by a factor of 1.8 in the current climate compared to a pre-industrial climate, and that in the absence of anthropogenic climate change an event with the same return period would have led lake levels to rise by 7 cm less than observed. Nonetheless, uncertainties in the attribution statement are relatively large due to large natural variability, and include the possibility of no observed attributable change in the probability of the event (probability ratio, 95 % confidence interval 0.8–15.8) or in the magnitude of lake level rise during an event with the same return period (magnitude change, 95 % confidence interval 0–14 cm). In addition to anthropogenic climate change, other possible drivers of the floods and their impacts include human land and water management, the exposure and vulnerability of settlements and economic activities located in flood-prone areas, and modes of climate variability that modulate seasonal precipitation. The attribution statement could be strengthened by using a larger number of climate model simulations, as well as by quantitatively accounting for non-meteorological drivers of the flood and potential unforced modes of climate variability. By disentangling the role of anthropogenic climate change and natural variability in the high-impact 2020 floods in the Lake Victoria basin, this paper contributes to a better understanding of changing hydrometeorological extremes in East Africa and the African Great Lakes region.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(15675 KB)
-
Supplement
(141 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(15675 KB) - Metadata XML
-
Supplement
(141 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-1827', Anonymous Referee #1, 22 Oct 2023
General remarks
Pietrouiusti et al. investigated the effect of anthropogenic climate change on the flood event of Victoria lake in 2020, using long-term in situ data, hydrological simulation and a large ensemble climate experiments with and without anthropogenic climate change. Authors found that the occurrence of the flood event has increased by a factor of 1.8 compared to a pre-industrial climate, and the anthropogenic climate change with the same return period would have led lake levels to rise by 7cm less without considering the anthropogenic climate change.The background, purpose, and results of the study are clearly and well presented. In exchange for the very thorough explanations, there are a few redundant sections and low readability are the only weaknesses. However, these text would help researchers who do attribution for other hydrologic variables or in other places using similar experimental data, so if there is no limit to the number of words in the journal, I would not ask for a drastic reduction.
-The future projection in 1.2 appears to be unnecessary. The description of future projections does not necessarily match climate change impacts on past events with projected changes in a warmer world in the future; differences in ENSO and IOD trends, changes in evapotranspiration, etc. will also change, so we felt that the description in the text, although background, could be minimized.
- I also do not know the intent of presenting Table 4. The results can vary depending on the number of hist-nat samples for the total sample, and it is not clear what the message is beyond what is obtained in Table 3.Changes in lake levels due to climate change tend to be smoothed out, possibly due to increased runoff, which increases in response to inflows, or human activity. Therefore, I found it difficult to adequately detect climate change signals. For example, if there is a large year-to-year variation in accumulated precipitation over the lake, it is possible that a more robust climate change signal could be detected if the accumulated precipitation is attributed. I thought it would be a good idea to add an addendum to the Discussion about this.
No Major concerns raised
Specific points
・ P.3, L75~: The main objective of this study is to investigate the effects of past climate change on the probability and magnitude of the past flood event in the lake Victoria. Since target period of future projection in precipitation and hydrometeorological extremes by CMIP5 and CMIP5, witten in 1.2, is not consistent with the target period of this study,
・ P.4, L85: Specify the time period of the average of precipitation.
・ P.19, L375-418: Since this paper already has many Appendixes, but 3.3 is not the main result either, I thought it would be better to put only the essence in Chapter 2 (Methods) of the main text and move the rest to the Appendix.
・ P.21, L424: The difference between model and observation around 2005-2015 appears to be due to the fact that there is less of a decrease in water levels between 2006 and 2007 in the model; the decrease in water levels since 2003 stops around 2005 in the model, but continues to fall through 2007 in the observation. The shape of the trend is similar to that of the rapid increase in water levels and the interannual variability of water levels since 2007. Is the difference in the drop in water level from 2003-2005 due to outflow?
・ P.25, Fig.11: I cannot see the upper side of the width of the preindustrial uncertainty. Could you please shade it or write it in with a dotted line, etc.?
・ P.29, Fig.12: I can't understand why you need the purple bar. What is the intention of averaging models and observations? Are you assuming that the observations are an ensemble of pasts that were not obtained by the model?Citation: https://doi.org/10.5194/egusphere-2023-1827-RC1 - AC1: 'Reply to RC1', Rosa Pietroiusti, 05 Dec 2023
-
RC2: 'Comment on egusphere-2023-1827', Anonymous Referee #2, 25 Oct 2023
General Comments:
Pietroiusti et al. examines the influence of anthropogenic climate change on the flooding of Lake Victoria in 2020, by applying a water-balance model in combination with a very well-established attribution methodology. This work advances the attribution field by using a new metric (rate of change of lake levels) more closely linked to impacts, compared with most studies that use meteorological variables without hydrological modelling. It also addresses the lack of attribution studies in the region examined. The authors find that anthropogenic climate change has likely increased the magnitude of the lake levels by 7cm (CI: 0-14 cm) and that the event was 1.8 (CI: 0.8-15.8) times more likely, based off a synthesis of results from the observations and climate models used. This work also gives an estimate of the impacts of the wider regional flooding event in 2020 using a remote sensing analysis. The methodology, analysis and calculations are well described. The detail provided is important and could be very useful for others working in the field. Some of it could be moved from the paper into the supplementary material- if required.
Science comments:
Line 50: Is this the first attribution study looking at rainfall or flooding in the Eastern Africa region? If so, this might be worth mentioning in the introduction, and if not - it would be good to know what other studies have found.
The result for the change in magnitude of the lake levels due to anthropogenic climate change seems more meaningful (in both how useful it is and its confidence intervals), compared to the risk ratio. In the conclusion the magnitude change only gets one line compared to the risk ratio which is discussed a lot more. It would probably be better the other way round. It might be worth comparing the anthropogenic magnitude change to the overall lake level rise, and the rise due to anomalous precipitation over the 180 days.
The main metric used (delta L/delta t for delta t=180 days) is well justified taking into account the influence of both rainfall and evaporation over the 180-days, with sensitivity tests carried out on the length of time chosen. This metric however, will naturally cover time periods where the rainfall/evaporation ratio is relatively high. Any change in evaporation due to climate change during the drier part of the year (relatively), which will likely influence the lake-level before the 180-day extreme lake-level rise, may be missed. This doesn't change, the anthropogenic influence on rate of change of lake level rise seen in 2020- but is probably worth including in the limitations. However, in general, the limitations and assumptions have been covered and addressed in great detail.
Technical/typos:
Line 444: Typo section title 3.4.2- ‘modelling’
Line 682-683: This line or similar would be good to have in the introduction- to give some context.
Citation: https://doi.org/10.5194/egusphere-2023-1827-RC2 - AC1: 'Reply to RC1', Rosa Pietroiusti, 05 Dec 2023
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-1827', Anonymous Referee #1, 22 Oct 2023
General remarks
Pietrouiusti et al. investigated the effect of anthropogenic climate change on the flood event of Victoria lake in 2020, using long-term in situ data, hydrological simulation and a large ensemble climate experiments with and without anthropogenic climate change. Authors found that the occurrence of the flood event has increased by a factor of 1.8 compared to a pre-industrial climate, and the anthropogenic climate change with the same return period would have led lake levels to rise by 7cm less without considering the anthropogenic climate change.The background, purpose, and results of the study are clearly and well presented. In exchange for the very thorough explanations, there are a few redundant sections and low readability are the only weaknesses. However, these text would help researchers who do attribution for other hydrologic variables or in other places using similar experimental data, so if there is no limit to the number of words in the journal, I would not ask for a drastic reduction.
-The future projection in 1.2 appears to be unnecessary. The description of future projections does not necessarily match climate change impacts on past events with projected changes in a warmer world in the future; differences in ENSO and IOD trends, changes in evapotranspiration, etc. will also change, so we felt that the description in the text, although background, could be minimized.
- I also do not know the intent of presenting Table 4. The results can vary depending on the number of hist-nat samples for the total sample, and it is not clear what the message is beyond what is obtained in Table 3.Changes in lake levels due to climate change tend to be smoothed out, possibly due to increased runoff, which increases in response to inflows, or human activity. Therefore, I found it difficult to adequately detect climate change signals. For example, if there is a large year-to-year variation in accumulated precipitation over the lake, it is possible that a more robust climate change signal could be detected if the accumulated precipitation is attributed. I thought it would be a good idea to add an addendum to the Discussion about this.
No Major concerns raised
Specific points
・ P.3, L75~: The main objective of this study is to investigate the effects of past climate change on the probability and magnitude of the past flood event in the lake Victoria. Since target period of future projection in precipitation and hydrometeorological extremes by CMIP5 and CMIP5, witten in 1.2, is not consistent with the target period of this study,
・ P.4, L85: Specify the time period of the average of precipitation.
・ P.19, L375-418: Since this paper already has many Appendixes, but 3.3 is not the main result either, I thought it would be better to put only the essence in Chapter 2 (Methods) of the main text and move the rest to the Appendix.
・ P.21, L424: The difference between model and observation around 2005-2015 appears to be due to the fact that there is less of a decrease in water levels between 2006 and 2007 in the model; the decrease in water levels since 2003 stops around 2005 in the model, but continues to fall through 2007 in the observation. The shape of the trend is similar to that of the rapid increase in water levels and the interannual variability of water levels since 2007. Is the difference in the drop in water level from 2003-2005 due to outflow?
・ P.25, Fig.11: I cannot see the upper side of the width of the preindustrial uncertainty. Could you please shade it or write it in with a dotted line, etc.?
・ P.29, Fig.12: I can't understand why you need the purple bar. What is the intention of averaging models and observations? Are you assuming that the observations are an ensemble of pasts that were not obtained by the model?Citation: https://doi.org/10.5194/egusphere-2023-1827-RC1 - AC1: 'Reply to RC1', Rosa Pietroiusti, 05 Dec 2023
-
RC2: 'Comment on egusphere-2023-1827', Anonymous Referee #2, 25 Oct 2023
General Comments:
Pietroiusti et al. examines the influence of anthropogenic climate change on the flooding of Lake Victoria in 2020, by applying a water-balance model in combination with a very well-established attribution methodology. This work advances the attribution field by using a new metric (rate of change of lake levels) more closely linked to impacts, compared with most studies that use meteorological variables without hydrological modelling. It also addresses the lack of attribution studies in the region examined. The authors find that anthropogenic climate change has likely increased the magnitude of the lake levels by 7cm (CI: 0-14 cm) and that the event was 1.8 (CI: 0.8-15.8) times more likely, based off a synthesis of results from the observations and climate models used. This work also gives an estimate of the impacts of the wider regional flooding event in 2020 using a remote sensing analysis. The methodology, analysis and calculations are well described. The detail provided is important and could be very useful for others working in the field. Some of it could be moved from the paper into the supplementary material- if required.
Science comments:
Line 50: Is this the first attribution study looking at rainfall or flooding in the Eastern Africa region? If so, this might be worth mentioning in the introduction, and if not - it would be good to know what other studies have found.
The result for the change in magnitude of the lake levels due to anthropogenic climate change seems more meaningful (in both how useful it is and its confidence intervals), compared to the risk ratio. In the conclusion the magnitude change only gets one line compared to the risk ratio which is discussed a lot more. It would probably be better the other way round. It might be worth comparing the anthropogenic magnitude change to the overall lake level rise, and the rise due to anomalous precipitation over the 180 days.
The main metric used (delta L/delta t for delta t=180 days) is well justified taking into account the influence of both rainfall and evaporation over the 180-days, with sensitivity tests carried out on the length of time chosen. This metric however, will naturally cover time periods where the rainfall/evaporation ratio is relatively high. Any change in evaporation due to climate change during the drier part of the year (relatively), which will likely influence the lake-level before the 180-day extreme lake-level rise, may be missed. This doesn't change, the anthropogenic influence on rate of change of lake level rise seen in 2020- but is probably worth including in the limitations. However, in general, the limitations and assumptions have been covered and addressed in great detail.
Technical/typos:
Line 444: Typo section title 3.4.2- ‘modelling’
Line 682-683: This line or similar would be good to have in the introduction- to give some context.
Citation: https://doi.org/10.5194/egusphere-2023-1827-RC2 - AC1: 'Reply to RC1', Rosa Pietroiusti, 05 Dec 2023
Peer review completion
Journal article(s) based on this preprint
Data sets
Input data for Pietroiusti et al. 2023 ESD Rosa Pietroiusti https://zenodo.org/record/8233523
Model code and software
GitHub repository with code Rosa Pietroiusti https://github.com/VUB-HYDR/2023_Pietroiusti_etal_ESD
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 428 | 118 | 22 | 568 | 38 | 21 | 17 |
- HTML: 428
- PDF: 118
- XML: 22
- Total: 568
- Supplement: 38
- BibTeX: 21
- EndNote: 17
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Inne Vanderkelen
Friederike E. L. Otto
Clair Barnes
Lucy Temple
Mary Akurut
Philippe Bally
Nicole P. M. van Lipzig
Wim Thiery
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(15675 KB) - Metadata XML
-
Supplement
(141 KB) - BibTeX
- EndNote
- Final revised paper