Analysis of Four Delineation Methods to Identify Potential Management Zones in a Commercial Potato Field in Eastern Canada

Round 1

Reviewer 1 Report

minor revision

Please add a flowchart in the first section of the method.
Line 112: add a relevant reference
Fig 1: please add the location of Canada.
Line 173: did you change the number of clusters?
Line 200: this section is unclear
Please clarify: did you perform the ECa for all studied areas or just in the sampling points? If you have ECa data in the sampling points, what is the benefit of using ECa?

 

Author Response

Reviewer 1:

 Please add a flowchart in the first section of the method.

Answer: A flowchart is added (line 188)

 Line 112: add a relevant reference

Answer:

Similar sampling grids sizes were chosen by Perron et al. 2018 and Cambouris et al. 2006 in their studies in potato fields. These two references have been added in the text. (From line 114 to line 116)

 Fig 1: please add the location of Canada.

Answer:

Thanks for your question. Figure 1.a has been removed due to poor image quality and imprecise field location. The field positioning under study could be determined by the geographic coordinates mentioned in the paper. (Line 107)

 Line 173: did you change the number of clusters?

Answer:

For the four delimitation methods including the Fuzzy K-means method, we worked with the same data set (Kriged values of ECa) and for each of these methods, we have delimited the field according to the same following hierarchy: 2MZs, 3MZs, 4MZs and 5MZs.

 Line 200: this section is unclear. Please clarify: did you perform the ECa for all studied areas or just in the sampling points? If you have ECa data in the sampling points, what is the benefit of using ECa?

Answer:

We used the kriged values of ECa in the whole field to delimit the MZs. We have added a section on kriging in the text to clarify the use of the data (From line 159 to line 168).

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors set out to test four methods to delineate an potato field in PEI for agricultural management practices.

The study has several shortcomings which need to be addressed before publication.

L55 'most' widely L62 'most used' please give scientific evidence for these statement. Please review the whole paper on such usage

L56ff - ECa has also been proven not to work at all in specific situations. Please mention this as well. see review articles on ECa

L100 rewrite-  the reviewer doubts that the field site produces 1/4 of canadas potato production

L107 - what type of irrigation and how much - please be precise. Please comment and revise.

The reviewer misses from an PA approach the measured detailed yields spatially detailed. WHy are they are not reported, what is the climate, etc. etc. the standard staff. 

L111 - how has the sampling design been developed, have a distance of at least 30m to the field boundaries excluded from the sampling (known artifacts due to compaction of turning agricultural equipment).Please comment and revise.

L113 - the study should be rejected for the purpose that only 0-15cm was sampled and the comparison/development was done on EC0-100cm. Thats a major flaw. What about the subsoil properties ? They are known to quite heavily influence the ECa values/yields/etc. Please comment and revise. We have done similar work 20 years ago.

Fig 1 - location of field site is not visible in GE image - what is the purpose of showing a bad quality (which year?) satellite image. Why not an NDVI image of the plot during May for example ? Please comment and revise.

L116 please specify ISO method numbers for the deployed soil analytical methods. Please comment and revise.

L125 - how were the 1 of 4 samples for pipette analysis been selected ? which methods - how were the statistical bias avoided ? Again, how were site locations close to field boundaries excluded ? Please comment and revise.

L128 ff - why is the spatial support to RTK and ECa different (e.g. number of samples per ha)? Have both measurments been done at the same time. At which time have the measurments been performed ? Whats the distance in along track and across track for elevation and ECa. How have the authors take the different spatial support into account for further processing? Please comment and revise.

L128 ff From Fig3 it becomes clear that the authors have worked with kriged data. This method is nowhere described in the paper. I'm almost at the stage to reject the paper.  The reviewer wonders how the EC and elevation data have been processed - by wizardry ?  Keep in mind - science is about reproducibility. I wonder why non of the coauthors has addressed that issue.

There is enough literature out which have worked on ECa and geostatistical interpolation (kriging is not neccessarily the best), see Precision Agriculture conference proceeding, geostatistics, ISATIS,...). The authors need to rework that missing part. If not added, the paper need to be rejected. FULL STOP

L145 and other. Good practice is to add software version numbers to a paper.

L144 Did the authors work on the single transect measurments or on interpolated values  ? Please add a DOT after "To do so, all the 143 ECa measurements from within 5-m radius around each soil sample were averaged to  represent co-located measurements" and rewrite the following part

L172 - table 1 - where does the word for efficency, user friendliness etc come from - Is that the authors own elaboration ? The reviewer does not agree with the statements in there. Please add refernces or sources for these ratings. or remove

L173 - good old Budi - i wait for his judgment on your table 1

L180 - what are the default parameters, default cell size etc ? Non of the work here is reproducible. below par documented  science. Rewrite. 

L190 - same applies

L201 - for ecognition (btw, there are nowadays also open source variants around - so need to pay the hefty licence fee) and other methods please clarify which input data have been used. The reviewer assumes that the same dataset has been used for all 4 methods ? or ? please revise and comment

L218 - what are the default parameter ?

L227 which parameter ? no one can reproduce this paper for the moment.

L226-228 According to the reviewer the authors have made a major design flaw in their experiment here. While in OIB they have aggregated the MZ to larger areas the same approach has not been applied to the first three methods. (majority filtering 2x, threshold of small areas to be excluded, standard gis operations ) The authors need to correct that before the paper can be accepted, otherwise its scientifically flawed and need to be rejected.

L276 - has this been also found in this study ? Without scientific proof in the study results the authors should not neglect the 0-30 results. Otherwise the reader ones why ?

L281 - where is the testing/selection of number of managment zones in the objective(s) ? Please revise and comment. 

Fig2 - why have 4 different plots - it would be more advisable to see all of them in one plot to see the differences in variance reduction in detail. Its also apparent that OIB hs the worst reduction in variance reduction - discuss. Redo

L281 - why only 5 MZ in the variance reduction ? even if the choosen applications later can be restricted to 2-3 MZ - here we expect to see a much longer tail. Budi had a paper on that if the reviewer not mistaken where there was an optimal cut off value. This should be taken into account and not choose arbitary values based on agronomy equipment. That goes back to the field site description - missing tools (width of planter, plow), tillage practice).

L302 have these significant difference also observed in yields ? what about elevation - why is elevation missing from discussion and table 4?

L357 need to be rewritten after applying similar approaches to all four methods.

L407 - what is fast and what is slow. be precise and give numbers and don't judge. You are scientist.

 

 

 

 

 

 

 

 

 

 

Author Response

Please see the attached Author's responses

Author Response File: Author Response.pdf