autoRA: An Algorithm to Automatically Delineate Reference Areas—A Case Study to Map Soil Classes in Bahia, Brazil

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The article entitled autoRA: An algorithm to automatically delineate reference areas. A case study to map soil classes in Bahia-Brazil submitted by Rodrigues et al. presents a significant and novel method in Pedometrics and Digital Soil Mapping supported with the submitted patents. The grammar and english language does not require corrections.

Abstract and Introduction were prepared very well. For me they can be published in the present form. The methodology does not require changes in the structure. The concept and methods has been well and clearly presented. They are a few corrections that I suggest to do:

p5, line 197: Random Forest is not a soil model but an algorithm of machine learning that is used not only for DSM purposes. Hence, the text in brackets requires correstion, for exaple (e.g. using Random Forest algorithm).

p6, lines 204-205. The first sentence is a repetition of the first sentence in the next subchapter. I subbest to delete it from here. It will not change the understanding of this subchapter.

p14, lines 419-420: Maybe both Figure 5 and 6, not Figure 5.

p22: line 562: Instead of Figure 13, it should be Fig. 14, isn't it?

p23: line 590: After "McBratney’s concept in the Homosoils framework" insert the reference

p25: line 628 and 644: Figure 16 should be cited before Figure 17.

p27: line 675: Table 3, not Table 2

Figure 18: Maybe you can change the colours of the graph for RA autoRA 10-50% as a gradient of one colour?

I miss the paragraph about the filling of research gap. Could you prepare an additional one by the end of discussion about similar studies and why your research is important due to the obtained results? Do you think that your studies help somehow in the environmental planning and protection in Brazil? You mention this in the Conclusion, but maybe you can extend this and implement to the Discussion part of the manuscript.

All the references were cited correctly.

Please, check all the Figures and Tables references in the manuscript.

To conclude, I can see a multiple use of the presented method. The studies were prepared on a very high scientific level and can be used also as a toll for the pedodiversity analyses and delineation of the soil districts. I read the article with great pleasure, seeing the combination of classical soil science with advanced mathematical models. There are not so many manuscripts that requires just a few corrections. For this reasons I propose a very minor revision.

Author Response

Dear Reviewer

 

Please see the attached answers to your comments on the manuscript.

 

My Best Regards,

Hugo

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The authors present a paper that aims to introduce autoRA (automatic Reference Area), a novel algorithm that locates and delineates RA based on covariate maps commonly applied in DSM workflows using Gower's Dissimilarity Index. The idea of ​​the authors is interesting and of international relevance, as it addresses the traditional problem of spatial variability of soils. Here are some suggestions to back up my idea and help the author improve the quality of the paper:

I'm not sure that the benchmark map concept is appropriate, although it would be useful.

The Gower’ Dissimilarity Index map (Figure 13) represents the spatial variation in dissimilarity across the RI, with values. Explained this way it is bland. Please rewrite.

Line 109. In this study, we present we used autora. Incorrect, modify accordingly.

Line 205. The methodology comprised three main stages: 1) delineation of Reference Areas (RAs), 2) soil sampling simulation, and 3) soil classification modeling (Figure 1). But later it is indicated: Before initiating fieldwork, the specialist compiled spatial covariates influencing soil formation in the area. These covariates included geology, geo morphology, and pedology maps from the Brazilian Institute of Geography and Statistics 237 (IBGE) at a 1:250,000 scale [42]. Shouldn't this come first?

Line 251. In each layer (covariate maps of each soil formation factor), the experts sought to identify small regions (polygons) that aggregated all or most of the diversity of categories (categorical data) or the range of values (continuous data). Thus, several diversity polygons were generated for each layer. Then, all of these polygons were superimposed, and a final polygon was drawn that encompassed them in a single polygon. Please explain better.

Line 240. It provides high-resolution gridded climate data with monthly averages from 1970 to 2000. It is not understood why that strip. Please explain.

Line 282. Characterization of Mapping Units. It appears very well described. As is also the case: Line 393 to 539. Soil Landscape Relationship and Spatial Distribution

The figure 1 should improve its quality.

As far as I know, Benchmark Soils is well known. Benchmark MU map is not so well known, or at least it is less used. It would be useful to go into more depth on the term, with specific references.

The discussion is very long and somewhat speculative, please reduce.

In short, this article, which investigates one of the "workhorses" of edaphology, namely soil mapping, is potentially of great interest. And its results are probably highly valid, but they are written in an unclear, I would even say convoluted, way. Therefore, I consider the article to be very meritorious, but part of it should be rewritten.

Finally I wonder: could this algorithm that automatically delineates reference areas be easily used by other researchers in other areas of the world?

Comments on the Quality of English Language

I am not an expert in the English language, so I should not give an opinion.

Author Response

Dear Reviewer

 

Please see the attached document with the answers to your comments on the manuscript.

 

My Best Regards,

Hugo

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

I recommend publication, my comments were taken into consideration.

Comments on the Quality of English Language

I´m not an expert in the English language, but I think it's okay.