Identifying Suitable Watersheds across Nigeria Using Biophysical Parameters and Machine Learning Algorithms for Agri–Planning

Round 1

Reviewer 1 Report

-> Introduction: please explain more about watershed prioritizing and highlight your contribution in this paper -> Methods: How do you define the levels (low/Moderate/High priority) for different datasets. I think it is not very clear for me to define the levels. -> Why there exists "Very low priority" in Fig. 5(a)? and the legend color is as the same as "low priority" -> Equation (1) and (2) should be more  standardized. -> why do you select this spatial model to integrate thematic layers? -> minor: Page 3 line99-100 32oC and 24oC --> 32℃ and 24℃; Please explain what is LGP at your first time mentioned it (Page 5, line135);  ->  Some references could be considered to cite if you think they are suitable for your paper.

Pandey, M., & Sharma, P. K. (2017, July). Remote sensing and GIS based watershed prioritization. In 2017 IEEE International Geoscience and Remote Sensing Symposium (IGARSS) (pp. 6182-6185). IEEE.

Zheng, J., Fu, H., Li, W., Wu, W., Zhao, Y., Dong, R., & Yu, L. (2020). Cross-regional oil palm tree counting and detection via a multi-level attention domain adaptation network. ISPRS Journal of Photogrammetry and Remote Sensing, 167, 154-177. Said, S., Siddique, R., & Shakeel, M. (2018). Morphometric analysis and sub-watersheds prioritization of Nagmati River watershed, Kutch District, Gujarat using GIS based approach. Journal of Water and Land Development.

Author Response

1. Introduction: please explain more about watershed prioritizing and highlight your contribution in this paper

Our response:

Thanks for the comments. Added in the introduction.

Line no 70-78;

“The purpose of identifying priority watersheds is to identify focus watersheds for complete restoration activities, this can address the critical needs in that watershed and allow for interventions.  This way of identification is easy to adapt and useful for decision-makers, as it combines all the necessary information and also allows for comparison among watersheds within the same cluster type. This approach produces a significant list of prioritized watersheds that allows the users in developing a summary of watersheds of interest, by spatially locating a watershed and obtaining relevant information about its vulnerability. The process can help in locating multiple watersheds’ with regard to prioritizing watershed protection and restoration.”

2. Methods: How do you define the levels (low/Moderate/High priority) for different datasets? I think it is not very clear for me to define the levels.

Our response:

Thanks for the comment. The levels are defined based on the expert’s opinion.

 Line no: 223-224

“Based on experts/scientist’s knowledge and published papers (Moore, Grayson et al. 1991, Murthy 2000, Moran, Peters-Lidard et al. 2004, Murthy and Mamo 2009, Gumma and Pavelic 2013), weights were allotted to the above layers.”

3. Why does there exists "Very low priority" in Fig. 5(a)? And the legend colour is the same as "low priority"

Our response:

Thanks for noticing. We have corrected it.

Line no: 178, Figure 5a

4. Equations (1) and (2) should be more standardized.

Our response:

Thanks for the suggestion. We have standardized it with arranging subscripts

Line 237, 245. Tsw = Tr * W ; Pm=ΣTsw

 

 

5. Why do you select this spatial model to integrate thematic layers?

Our response:

Thanks for the comment. For large area, by utilizing the public datasets, this type of spatial model can help in prioritising any Land use with fast and efficient manner.

6. Minor: Page 3 line99-100 32oC and 24oC --> 32℃ and 24℃;

Our response:

Thanks for noticing. Corrected the superscripts.

Line no: 101, 102; 32℃ and 24℃;

7. Please explain what is LGP at your first time mentioned it (Page 5, line135); 

Our response:

Thanks for noticing. We have added “Length of Growing Periods (LGP)”. Line no 140

8. Some references could be considered to cite if you think they are suitable for your paper.

Pandey, M., & Sharma, P. K. (2017, July). Remote sensing and GIS-based watershed prioritization. In the 2017 IEEE International Geoscience and Remote Sensing Symposium (IGARSS) (pp. 6182-6185). IEEE.

Zheng, J., Fu, H., Li, W., Wu, W., Zhao, Y., Dong, R., & Yu, L. (2020). Cross-regional oil palm tree counting and detection via a multi-level attention domain adaptation network. ISPRS Journal of Photogrammetry and Remote Sensing, 167, 154-177. Said, S., Siddique, R., & Shakeel, M. (2018). Morphometric analysis and sub-watersheds prioritization of Nagmati River watershed, Kutch District, Gujarat using GIS based approach. Journal of Water and Land Development.

Our response:

Thanks for the suggestion. We have added. Line no 70.

Reviewer 2 Report

 

 

·         In this study, the authors try to prioritize the watershed based on biophysical parameters through the Machin learning algorithm for Agri–planning in Nigeria

 ·         Citation to be placed as per the journal format

·         Page4 Line no 122: check the statement “LULC map from sentinel–2 10m data using NDVI time-series data for the entire year”

 

 ·         The time period of the study (which year has been selected for this evaluation) is missing

 

 ·         In the title, the Machin learning algorithm was focused/highlighted, but it hasn’t been reflected throughout the study. And nowhere discussed about ML algorithms and their working procedures. How and what are the parameters were used through this.

 

 ·         The authors stated that in the abstract as “modelled in the Google Earth Engine (GEE) platform” and in the methodology section “, other thematic spatial layers were acquired from the public domain and downloaded using Google Earth Engine. (Page4 Line no 124)” Both the statements are different meanings.

 

 ·         Also, the authors haven’t described about the GEE platform and its data preparation and modelled procedures.

 

 ·         Which ranking method was used in this study. (Multi-criteria evaluation like AHP, FAHP, TOPSIS etc.). The weighting and ranking approach is not straightforward.

 

 ·         Page 6, line no 166: check the statement “Soils (FAO) and Soil parameters were important parameters”

 

 ·         There are several studies performed in various parts of the globe, hence how this study has differed from other studies. I feel its looks like the same methods and datasets.

 

 ·         Table 4: abbreviated word of the objective to be given (Irr/Ws/HP) Irrigation, Water supply, Hydroelectrical power generation (This is my observation).

·         For validation, why authors have considered only DAMs. What about other planning and land-use practices?

 

 ·         The result and discussion section are very general. Need to be appropriately added.

 ·         The recommendation and suggestion to be site-specific

 ·         Units and other terms should be appropriately placed (Superscript and appropriate signs).

 Authors could check the following studies for a better understanding

1. https://doi-org.libproxy.viko.lt/10.1007/s12517-021-08105-z 

2. Prioritization of watersheds using multi-criteria evaluation through fuzzy analytical hierarchy process Agric Eng Int: CIGR Journal, 15 (1) (2013), pp. 11-18

 

 

Author Response

In this study, the authors try to prioritize the watershed based on biophysical parameters through the Machin learning algorithm for Agri–planning in Nigeria

Our response:

Thanks for your time in reviewing the paper. We have responded for the comments.

1. Citation to be placed as per the journal format

Our response:

Thanks for the comment. We have revised the references with MDPI format. Line no 440-560

2. Page 4 Line no 122: check the statement “LULC map from sentinel–2 10m data using NDVI time-series data for the entire year”

Our response:

Thanks for the comment. We have rewritten the phrase. “The primary data for thematic layers like LULC map of year 2014 was prepared from the MODIS 250m satellite imagery using NDVI time series data” (line 125 -126)

3. The time period of the study (which year has been selected for this evaluation) is missing

Our response:

Thanks for the comment. The time period is year 2014, since the basic layer of the study is LULC map – 2014. For other parameters, we have used available datasets and considered average values for parameters like Rainfall, temperature and others.

4. In the title, the Machine learning algorithm was focused/highlighted, but it hasn’t been reflected throughout the study. And nowhere discussed about ML algorithms and their working procedures. How and what are the parameters were used through this.

Our response:

Thanks for noticing, we have include section 2.6, describing the GEE operations and machine learning algorithms. Line no 249-282

5. The authors stated that in the abstract as “modelled in the Google Earth Engine (GEE) platform” and in the methodology section “, other thematic spatial layers were acquired from the public domain and downloaded using Google Earth Engine. (Page4 Line no 124)” Both the statements are different meanings.

Our response:

Thanks for noticing. We have used public datasets and ingested the non-available datasets, followed by the model in the Google earth engine. We have deleted the word “downloaded”. Line no: 127-128; Line 252-253

6. Also, the authors haven’t described about the GEE platform and its data preparation and modelled procedures.

Our response:

Thanks for noticing, we have included section 2.6, describing the GEE operations and machine learning algorithms. Line no 249-282

7. Which ranking method was used in this study? (Multi-criteria evaluation like AHP, FAHP, TOPSIS etc.). The weighting and ranking approach is not straightforward.

Our response:

Thanks for the comment. Since we have selected a large study area, and available coarse resolution data, we have assigned weights ranging from 3 to 1. Given weights in every section.

8. Page 6, line no 166: check the statement “Soils (FAO) and Soil parameters were important parameters”

Our response:

Thanks for noticing. Revised the sentence “Soil parameters (Soil Texture and Soil Depth) plays a vital role in watershed prioritization as of their role in runoff “; Line 169-171

9. There are several studies performed in various parts of the globe, hence how this study has differed from other studies. I feel its looks like the same methods and datasets.

 Our response:

Thanks for your comment. This is the first study on Nigeria’s extent and the whole process was modelled in Google Earth Engine and also conducted validation at country level. There are many studies but only a few validated the study.

10. Table 4: abbreviated word of the objective to be given (Irr/Ws/HP) Irrigation, Water supply, Hydro electrical power generation (This is my observation).

Our response:

Thanks for noticing. We appreciate your observation. Water Supply is correct. We have corrected it. Line 343

11. For validation, why authors have considered only DAMs. What about other planning and land-use practices?

Our response:

Thanks for the comment. Since the study area is large and concentrated on watersheds, we got only DAMs with the purpose, for validation. There is very less availability of secondary data. Line 342-343

12. The result and discussion section are very general. Need to be appropriately added.

 Our response:

Thanks for the comment, we have added our results in order along with validation and discussed more in discussion section 4.0. Added in line 327-332.

“Considering only precipitation, it is observed that only 98 watersheds have high suitable rainfall conditions which is crucial layer for agri planning and about 159 watersheds are in moderate priority. For maximum and minimum temperature, almost all watersheds have moderate and high suitable conditions, whereas soil conditions are also in favourable conditions. This shows the importance of watersheds in this country.”

13. The recommendation and suggestion to be site-specific

 Our response:

Thanks for the suggestion. Our future studies will focus on the site-specific using very high-resolution data.

14. Units and other terms should be appropriately placed (Superscript and appropriate signs).

Our response:

Thanks for the suggestion. We will revise them as advised. “32oC and 24oC --> 32℃ and 24℃; also subscripts in formulae” Line no: 101, 102; Line 235, 248.

15. Authors could check the following studies for a better understanding
16. https://doi-org.libproxy.viko.lt/10.1007/s12517-021-08105-z 
17. Prioritization of watersheds using multi-criteria evaluation through fuzzy analytical hierarchy process Agric Eng Int: CIGR Journal, 15 (1) (2013), pp. 11-18

Our response:

Thanks for your suggestions. This will help us to improve better.

Reviewer 3 Report

Why do "Machine Learning Algorithms" exist in the Title?

Keywords must be relevant for database search, and different than those already appearing in the title. The function of keywords is to supplement the information given in the title. Words in the title are automatically included in indexes, and keywords serve as additional pointers. Authors should put good attention to the keywords they provide.

Line 14: what do you mean by "viz."

Figure 1: This map does not reflect the information about the study area in Section 2.1. The study area map should show more information on the study area.

Figure2: show the role of GEE in the methodology.

Show the validation of the Land Use and Land Cover map (Section 2.3.1).

Provide the related Reference of Soils (Section 2.3.3) and Length of Growing Period data (Section 2.3.5).

Section 2.3 "Criteria and Determining Factors" why do you use 3 classes? it is more generalized, and more discussion and references are required.

Section 3.3. "Validation of priority watersheds" is accepted but it is not enough.

Move Table 4 to the supplementary data section.

Please show a table with all the information about the data. Variable, year, type, resolution, and source. Please place this table in the supplementary material.

Author Response

1. Keywords must be relevant for database search, and different from those already appearing in the title. The function of keywords is to supplement the information given in the title. Words in the title are automatically included in indexes, and keywords serve as additional pointers. Authors should put good attention to the keywords they provide.

Our response:

Thanks for the suggestion. Updated the keywords as advised. “water; agriculture; dry land; Google Earth Engine” Line no: 23

2. Line 14: what do you mean by "viz."

Our response:

Thanks for your comment. Viz. is the shortcut form of “such as”, we have deleted it and written as such as. Line no: 15

3. Figure 1: This map does not reflect the information about the study area in Section 2.1. The study area map should show more information on the study area.

Our response:

Thanks for the comment. We have added agro ecological zones to the study area for more information. (Figure 1); Line no 112, Figure 1

4. Figure2: show the role of GEE in the methodology.

Our response:

Thanks for the comment. The whole process is carried in GEE. Included logo in methodology (line 134) (Figure 2) and explained in section 2.6

5. Show the validation of the Land Use and Land Cover map (Section 2.3.1).

Our response:

Thanks for the comment. We have used the published LULC and cited. “https://doi.org/10.1016/j.isprsjprs.2017.01.019”.Cited in Line no 150

6. Provide the related Reference of Soils (Section 2.3.3) and Length of Growing Period data (Section 2.3.5).

Our response:

Thanks for the comment. We have cited. And shown in annexure table as you suggested. Also cited in respective sections. Line no: 600, Table x in annexure.

7. Section 2.3 "Criteria and Determining Factors" why do you use 3 classes? it is more generalized, and more discussion and references are required.

Our response:

Thanks for the comment. Since the data used was coarse, we have classified it into three classes for clear understanding.

8. Section 3.3. "Validation of priority watersheds" is accepted but it is not enough.

Our response:

Thanks for the comment. Since the study area is large and concentrated on watersheds, we got only DAMs are the better locations with the purpose for validation. There is very less availability of secondary data.

9. Move Table 4 to the supplementary data section.

Our response:

Thanks for the comment. Since the table itself represents the validation, we opted for placing the table in the main article.

10. Please show a table with all the information about the data. Variable, year, type, resolution, and source. Please place this table in the supplementary material.

Our response:

Thanks for your suggestion. Added in Line no: 600, Table x

Variable	Year	Type	Resolution	Source
Maximum Temperature	2010–2018	Raster	~4 km	TerraClimate
Minimum Temperature	2010–2018	Raster	~4 km	TerraClimate
Precipitation	2010–2018	Raster	~4 km	TerraClimate
Slope	2014	Raster	90 m	SRTM
Soil	1994	Vector	1:5000000	FAO [29]
LULC	2014	Raster	250m	LULC [28]
LGP	2011	Raster	8km	LGP [31]

 

 

 

Round 2

Reviewer 1 Report

The authors have addressed all my issues well. I recommend this article to publish in this journal with minor revision. Yet, just two more small suggestions. -> Section 2.7 Watershed delineation should be described in more details.  -> I think these two references are still deserved to cite in your paper, which you may miss in last review round.  Zheng, J., Fu, H., Li, W., Wu, W., Zhao, Y., Dong, R., & Yu, L. (2020). Cross-regional oil palm tree counting and detection via a multi-level attention domain adaptation network. ISPRS Journal of Photogrammetry and Remote Sensing, 167, 154-177. Said, S., Siddique, R., & Shakeel, M. (2018). Morphometric analysis and sub-watersheds prioritization of Nagmati River watershed, Kutch District, Gujarat using GIS based approach. Journal of Water and Land Development.

Author Response

1. The authors have addressed all my issues well. I recommend this article to publish in this journal with minor revision.

 Our response:  Thanks for reviewing our paper. We appreciate your recommendations. Your comments and responses have improved our paper better.

2. Yet, just two more small suggestions.

• Section 2.7 Watershed delineation should be described in more details. 

Our response: Thanks for your comment. We have included in line: 285-291

“The process starts with filling the sinks by comparing the values of neighbouring cells. The filled sinks help in the generation of flow direction by finding the steepest descent of every cell. Then, flow accumulation is calculated using flow direction by counting the number of cells that are flowing to a particular cell. A set of thresholds for flow accumulation and flow direction generates the stream network. The generation of pour points at the 6th stream order for the entire study area helps in the generation of watersheds (Figure 9(a)).”

3. I think these two references are still deserved to cite in your paper, which you may miss in last review round. 

Zheng, J., Fu, H., Li, W., Wu, W., Zhao, Y., Dong, R., & Yu, L. (2020). Cross-regional oil palm tree counting and detection via a multi-level attention domain adaptation network. ISPRS Journal of Photogrammetry and Remote Sensing, 167, 154-177.

Said, S., Siddique, R., & Shakeel, M. (2018). Morphometric analysis and sub-watersheds prioritization of Nagmati River watershed, Kutch District, Gujarat using GIS based approach. Journal of Water and Land Development.

Our response: Thanks for your suggestions. We have cited above papers in line 69.

Author Response File: Author Response.docx

Reviewer 2 Report

Dear Authors.

I think the core concept will divert the study. The authors may change the content to suitability watersheds instead of prioritisation. In general, priority is given to the weaker area to be given more concentration in terms of agricultural planning and management.

In this study, table 1 shows the priority level for thematic layers, actually, this is the thematic layer's suitability condition for agriculture. So, this study focused on finding the suitability of watersheds. So, it makes confusion for the readers.

The ranking part is not satisfactory, because the normal scaling method is only fit for the suitability, for prioritization needs to go for fuzzy logic, AHP & TOPSIS based weighting and ranking. Nowadays, many techniques and ranking methods are widely used. So, I suggest instead of priority authors try to change as suitability watershed.  For the ranking purpose size/area of the basin is not n issue. Anyhow you are assigning the ranking for the thematic layers, it is suitable for the micro watershed to basin scale.

Page 15 line 326 : the authors stated that  “Considering only precipitation, it is observed that only 98 watersheds have high suitable rainfall conditions”. It's not suitable, those watersheds need to give high priority.  (That’s what I would suggest authors need to re-consider the concept)

Page 7 Lines 362 to 366 : Discussion section updated track change content not properly appeared.

Page 18Line 408: spell check

Author Response

I think the core concept will divert the study. The authors may change the content to suitability watersheds instead of prioritisation. In general, priority is given to the weaker area to be given more concentration in terms of agricultural planning and management.

In this study, table 1 shows the priority level for thematic layers, actually, this is the thematic layer's suitability condition for agriculture. So, this study focused on finding the suitability of watersheds. So, it makes confusion for the readers.

The ranking part is not satisfactory, because the normal scaling method is only fit for the suitability, for prioritization needs to go for fuzzy logic, AHP & TOPSIS based weighting and ranking. Nowadays, many techniques and ranking methods are widely used. So, I suggest instead of priority authors try to change as suitability watershed.  For the ranking purpose size/area of the basin is not n issue. Anyhow you are assigning the ranking for the thematic layers, it is suitable for the micro watershed to basin scale.

Page 15 line 326 : the authors stated that  “Considering only precipitation, it is observed that only 98 watersheds have high suitable rainfall conditions”. It's not suitable, those watersheds need to give high priority.  (That’s what I would suggest authors need to re-consider the concept)

Our response: Thanks for reviewing the paper. We sincerely appreciate your recommendations. We are changing our title to “Identifying suitable watersheds across Nigeria using biophysical parameters and Machine Learning Algorithms for Agri–planning”. Inserted in line 2, 13-14, and made few changes in abstract, conclusions and methodology flowchart as per title.

Page 7 Lines 362 to 366 : Discussion section updated track change content not properly appeared.

Our response: Thanks for your comment. We have revised that content in line 358-362.

“In past studies, watershed prioritization has been carried out using quantitative analysis, statistical methods, fuzzy and AHP techniques [37-39], morphometric analysis [40], delineation of groundwater potential zones [41], prioritization of sub-watersheds [42,43], prioritization of semi-arid agricultural watersheds [44], spatial assessment of soil erosion risk [45,46] and many other parameters.”

Page 18 Line 408: spell check

Our response: Thanks for your comment. We have corrected it. In line 401-403

“The Identification and delineation of such watersheds areas helps in better agricultural development planning as well as implementing appropriate interventions.”

 

Author Response File: Author Response.docx

Reviewer 3 Report

The authors have improved the manuscript, it can be accepted for publication.

Author Response

The authors have improved the manuscript, it can be accepted for publication.

Our response: Thanks for reviewing our paper. We appreciate your recommendations. Your comments and responses have improved our paper better.

Author Response File: Author Response.docx