Invited perspectives: How machine learning will change flood risk and impact assessment

Wagenaar, Dennis; Curran, Alex; Balbi, Mariano; Bhardwaj, Alok; Soden, Robert; Hartato, Emir; Mestav Sarica, Gizem; Ruangpan, Laddaporn; Molinario, Giuseppe; Lallemant, David

doi:https://doi.org/10.5194/nhess-20-1149-2020

Articles | Volume 20, issue 4

https://doi.org/10.5194/nhess-20-1149-2020

© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/nhess-20-1149-2020

© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 20, issue 4

Invited perspectives

|

29 Apr 2020

Invited perspectives |

| 29 Apr 2020

Invited perspectives: How machine learning will change flood risk and impact assessment

Dennis Wagenaar, Alex Curran, Mariano Balbi, Alok Bhardwaj, Robert Soden, Emir Hartato, Gizem Mestav Sarica, Laddaporn Ruangpan, Giuseppe Molinario, and David Lallemant

Download

Final revised paper (published on 29 Apr 2020)
Preprint (discussion started on 16 Oct 2019)

Interactive discussion

Status: closed

AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment

- Printer-friendly version

- Supplement

RC1: 'Comments on the manuscript "Invited perspectives: How machine learning will change flood risk and impact assessment", by Wagenaar et al.', Anonymous Referee #1, 26 Nov 2019
- AC1: 'Response to referee 1', Dennis Wagenaar, 14 Feb 2020
RC2: 'Review on "Invited perspectives: How machine learning will change flood risk and impact assessment"', Anonymous Referee #2, 03 Jan 2020
- AC2: 'Response to referee 2', Dennis Wagenaar, 14 Feb 2020

Peer-review completion

AR: Author's response | RR: Referee report | ED: Editor decision

ED: Reconsider after major revisions (further review by editor and referees) (15 Feb 2020) by Heidi Kreibich

AR by Dennis Wagenaar on behalf of the Authors (17 Feb 2020) Author's response Manuscript

ED: Referee Nomination & Report Request started (23 Feb 2020) by Heidi Kreibich

RR by Anonymous Referee #1 (09 Mar 2020)

RR by Fernando Nardi (15 Mar 2020)

Suggestions for revision or reasons for rejection

Review by Fernando Nardi (fernando.nardi@unistrapg.it)
See attached PDF for the general and specific comments, also hereafter provided.

I am glad that authors positively considered the comments and remarks received in the first round of review. The manuscript was accordingly edited and it is significantly improved.
However, I still think there are some missing points that relate to ML and flood-related DRR that authors caught but didn’t explore at full. In particular I’m sharing here two major comments that I’d like authors to check and eventually use for further improving the manuscript before publication. These two general comments relate to some aspects, of interest of ML for flood risk, that authors didn’t consider or explore at full and in particular 1) New data, crowdsourced data and citizen science; and 2) human behavior and the value of ML for trans-disciplinary studies involving social, behavioral sciences and risk awareness, communication strategies
General comments/remarks
1) New data, crowdsourced data and citizen science. Authors cite “new data”, but without providing a definition or a detailed conceptualization of what it is meant for new data in relation to the manuscript topic. Crowdsourced data and data/information from social network are also mentioned several times, but citizen science is never mentioned. There is an increasing trend, impact and production of citizen science project, social network based initiative for gathering, processing and modelling data of interest of flood risk management. The relationship and potential use of ML for empowering flood risk management by means of these new data is clear. I’d advise authors in this regard to check and cite these relevant works:
- Annis A. and Nardi F. (2019): Integrating VGI and 2D hydraulic models into a data assimilation framework for real time flood forecasting and mapping, Geo-spatial Information Science
- Mazzoleni, M., Verlaan, M., Alfonso, L., Monego, M., Norbiato, D., Ferri, M., & Solomatine, D. P. (2017). Can assimilation of crowdsourced data in hydrological modelling improve flood prediction?. Hydrology and Earth System Sciences, 21(2), 839-861.
- Assumpção, T. H., Popescu, I., Jonoski, A., & Solomatine, D. P. (2018). Citizen observations contributing to flood modelling: Opportunities and challenges. Hydrology and Earth System Sciences, 22(2), 1473-1489.
- Fohringer, J., Dransch, D., Kreibich, H., & Schröter, K. (2015). Social media as an information source for rapid flood inundation mapping. Natural Hazards and Earth System Sciences (NHESS), 15, 2725-2738.
2) Human behavior and the value of ML for trans-disciplinary studies involving social, behavioral sciences and risk awareness, communication strategies. An additional element that was never considered is the “human behavior” component that related to how citizens and in general the population behaves under flooding threats. The perception, the science literacy, and the way people respond before, during and after flooding events are all relevant factors that are missing in this work. This aspect goes beyond the development of more accurate or timelydata/information, either descriptive or predictive, and tackles a very important factor that is the way people act and react, the knowledge base and the general understanding of flood hazard. Moreover, communication and awareness are also not fully explored. I do believe ML will have a role in these components of the Disaster Management Cycle. ML may also play a crucial role in the need of developing transdisciplinary studies integrating earth/hydrological sciences with social, behavioral and communication sciences

As a result, as invited perspective, deriving from outcomes of the june 2019 UR workshop, keeping in mind this manuscript provides a sort of an extended meeting report , I agree that the manuscript can be published as it is. Nevertheless, I’d be glad if authors could share their ideas, thoughts and applied research references related to the two above mentioned points and provide a final version of the manuscript for publication

Specific comments
See attached commented PDF with further specific comments and remarks

Referee Report: PDF

Hide

ED: Publish subject to technical corrections (15 Mar 2020) by Heidi Kreibich

AR by Dennis Wagenaar on behalf of the Authors (30 Mar 2020) Author's response Manuscript

Short summary

This invited perspective paper addresses how machine learning may change flood risk and impact assessments. It goes through different modelling components and provides an analysis of how current assessments are done without machine learning, current applications of machine learning and potential future improvements. It is based on a 2-week-long intensive collaboration among experts from around the world during the Understanding Risk Field lab on urban flooding in June 2019.