Next Article in Journal
ENSO Impacts on Jamaican Rainfall Patterns: Insights from CHIRPS High-Resolution Data for Disaster Risk Management
Previous Article in Journal
The Lac Fallère Area as an Example of the Interplay between Deep-Seated Gravitational Slope Deformation and Glacial Shaping (Aosta Valley, NW Italy)
 
 
Article
Peer-Review Record

Machine Learning-Based Flood Forecasting System for Window Cliffs State Natural Area, Tennessee

GeoHazards 2024, 5(1), 64-90; https://doi.org/10.3390/geohazards5010004
by George K. Darkwah 1, Alfred Kalyanapu 2,* and Collins Owusu 3
Reviewer 1: Anonymous
Reviewer 2:
GeoHazards 2024, 5(1), 64-90; https://doi.org/10.3390/geohazards5010004
Submission received: 5 December 2023 / Revised: 15 January 2024 / Accepted: 23 January 2024 / Published: 25 January 2024

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

1. It is recommended to use the International System of Units instead of Anglo-Saxon units.

2. Relationship (1): "All input data were linearly normalized between 0 and 1 using Equation 1." In fact, the X-norm values are between Fmin and Fmax.

3. Precipitation is the trigger of floods. It is not clear why the first simulations were made without including rainfall data.

4. Model training should be done considering a full hydrological year but characterized by exceptional water levels to infer water depths for a variety of situations, including the worst events.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript implies machine learning techniques to forecast the floods using the time-series data. Followings are a few of my comments to improve the performance further:

1. The introduction can benefit from the intuition of the research and contributions needs to be clearly listed towards the end.

2. A detailed literature review section is missing. There is abundant of literature available in regards to the use of ML and AI for flood forecasting. I would suggest authors to review the benchmark literature and identify the existing research gaps to give the reader idea of the state of the art in this domain. A subjective critical analysis must be performed on the literature based on some assessment criteria. 

3. A more detailed EDA on the dataset would be helpful for reader to understand the challenge and nature of data being used for training.

4. Discussions should be kept seperate from conclusions. Conclusion should be a standalone section.

5. Discussions should discuss subjectively about why models performed better or worst using examples. 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

Authors addressed the comments to my satisfaction from first round of review 

Back to TopTop