A Data-Driven Method for Identifying Drought-Induced Crack-Prone Levees Based on Decision Trees

Round 1

Reviewer 1 Report

This paper studies the proneness of levees to crack due to drought periods. The method applied is statistical and considers several variable that may be relevant for the crack appearance prognostics. The multidimensional aspect of this forecasting problem is managed by using decision trees. The method is applied to a region near Rotterdam, The Netherlands. Crisp results are obtained. Levee’s cracks are related to the flexibility and precipitation deficit.

General comments

The Cramer V formula does not produce negative values. However you say “Negative values of Cramer’s V were not considered due to the domain of this statistical metric (values below 0 have no meaning).” Please check the convenience of this sentence. Consider eliminating it.

 

as a peat width of at least 31 centimeters may still induce cracking. Shure it is Centimeters?

 

Writing

Graph in Figure 7B: Cannot see Rank 3 because of the white color.

 

“…for which the negatives were generated in differently.”  The ‘ín’ is extra.

 

Figure 10 is great. Just fix the superposition problem in horizontal axis values for subsidence rate.

 

Final comments:

The manuscript is very well written and organized. Explaining these multidimensional problems is never easy. The authors however present a clear a nice explanation of their sophisticated method. From the point of view of the statistics, the details a cared without going into excessively theoretical considerations. I have no suggestion to improve this manuscript. It is near as good as it can be. In my view is ready for publication.

Author Response

Dear reviewer,

please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The study in the present form is not suitable for publication as the MCC of 0.31 or 0.51 could not be considered a perfect classifier. The authors are suggested to improve their model via hybridization techniques or using other models such as random forest.

Minor points:

• The given title could be improved. I am not sure the term "drought-induced" is suitable for this study.
•  Page 1 line 24: replace “drought periods” with “dry periods”
• Page 3 line 76: correct your method. It is not the random forest. In line 213, you said “model tree classification algorithm”
• Apart from MCC, it is also needed to calculate classic performance measures such as total accuracy, precision, and recall.

Author Response

Dear reviewer, 

please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Dear authors,

your paper "A data-driven method for identifying drought-induced crack
prone levees based on decision trees" is a very nice scientific contribution and nicely written paper. Your organization of paper in seven structures is good and logical. I have some minor comments:

Introduction - authors stated that they explore the potential of a data-driven approach that allows us to understand not only the drivers of cracking on clay and peat levees but also their spatial distribution. The introduction part is good structured and described; 

Section 2 outlines the literature study on proposing factors that contribute to the cracking mechanism. It is written in clear, logical and scientific language. 

In section 3, you present the methodology used to collect data, develop a machine learning (random forest) method to identify vulnerable levees
and to generate hazard maps.

In Section 4, the authors elaborated on a case study used to demonstrate the method. It is written in clear, logical and scientific language.

In section 5, the authors then present the results. It is written in clear, logical and scientific language.  In a very clear way, the authors presented the results;

In section 6, the authors discussed the results. I will suggest you improve comparations of your results with some other similar studies;

Conclusion - I will only recommend you add scientific and social implications to your research;

Thank you for this very nice and important research in this field.

Kind regards

Author Response

Dear reviewer, 

please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Accepted.