Next Article in Journal
Organic Pollutants Removal from Olive Mill Wastewater Using Electrocoagulation Process via Central Composite Design (CCD)
Next Article in Special Issue
Synthetic Musk Fragrances in Water Systems and Their Impact on Microbial Communities
Previous Article in Journal
Comparing Experiences of Constitutional Reforms to Enshrine the Right to Water in Brazil, Colombia, and Peru: Opportunities and Limitations
Previous Article in Special Issue
Water Resource Risk Assessment Based on Non-Point Source Pollution
 
 
Article
Peer-Review Record

Assessing Surface Water Flood Risks in Urban Areas Using Machine Learning

Water 2021, 13(24), 3520; https://doi.org/10.3390/w13243520
by Zhufeng Li 1, Haixing Liu 2,*, Chunbo Luo 1 and Guangtao Fu 1
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Reviewer 3: Anonymous
Water 2021, 13(24), 3520; https://doi.org/10.3390/w13243520
Submission received: 10 November 2021 / Revised: 29 November 2021 / Accepted: 7 December 2021 / Published: 9 December 2021
(This article belongs to the Special Issue Environmental Risk Management)

Round 1

Reviewer 1 Report

Maybe the introduction and the conclusions could be extended.

Author Response

Please see the attached file for the response.

Author Response File: Author Response.pdf

Reviewer 2 Report

Please see the attached file for comments.

 

Comments for author File: Comments.pdf

Author Response

Please see the attached file for the response.

Author Response File: Author Response.pdf

Reviewer 3 Report

  1. This research uses machine learning to discover patterns directly from data. Although there is no need to define rules, as far as the two major challenges of machine learning are concerned, how to choose input data to reach high-quality models directly, and how to use deep learning algorithms to enhance features on large data sets, should clearly explain the operation process.
  2. This research is based on a data-driven model. The reason for using the 11 characteristics of the basin to construct the model should not be repeated. If you want to explain, you should clearly explain the physical meaning of the relationship between the 11 characteristics and the flood. For example, the relationship between “distance to road” and the flood is not explained.
  3. Various existing machine learning technologies have their own advantages and disadvantages, and they cannot be compared unless they are used for specific purposes. Authors should review the literature to find the best method without comparing methods.
  4. Events with a low return period can provide more observational information, and the map data of its flooding potential is more reliable, but the accuracy of the CNNs model shows that the accuracy of 100-year flood events is higher than that of 30-year flood events. This is unreasonable with the facts (Figure 3).
  5. It is not appropriate to discuss the accuracy of the model with an event recurrence period of 1000 years. The design strength of general flood control structures will not adopt such a large incident return period.
  6. Flooding can be divided into inner water and external water. This research focuses on regional drainage, and does not involve river bank overflow flooding. It should be simulated by the design return period of various flooding prevention measures in the study area. Authors should conduct innovative research on research gaps or practical needs.

Author Response

Please see the attached file for the response.

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

My conclusion is that the author considered the reviewer’s suggestions and suggested that the manuscript can be accepted for publication.

Back to TopTop