CPDOS: A Web-Based AI Platform to Optimize Crop Planting Density
Round 1
Reviewer 1 Report
The manuscript “CPDOS: A Web Platform to Optimize Crop Planting Density Based on an Intelligent Algorithm” summarizes six different renowned crop planting density optimization models and creatively introduces a genetic algorithm, BP neural network and polynomial regression for planting density optimization of different crops. Artificial intelligence is the simulation of human intelligence processes by machines, especially computer system to get objective oriented results. Artificial Intelligence in agriculture not only helping farmers to automate their farming but also shifts to precise cultivation for higher and sustainable crop production. Planting density optimization is for important and considered as a major factor ensuring higher crop production. The MS is interesting for the readers. Some suggestions are given below to improve the readability of MS.
· Title
Title “CPDOS: A Web Platform to Optimize Crop Planting Density Based on an Intelligent Algorithm” can be changed as “CPDOS: A Web based AI Platform to Optimize Crop Planting Density”. This is just suggestion, but the authors can keep the same title as they already given.
· Abstract
Please write down some output of the Intelligent Algorithm for Optimization of Crop Planting Density, it will increase readers’ interest.
· Introduction
Introduction is looking fine, change the citation style to normal text as the citations are given in the from of superscript. Another suggestion is to reduce the introduction by keeping most important and relevant citation and text.
· M & M
The M&M is well written and explained the model very well, I have one concern about the tables mentioned in M&M as supplementary files, it is quite for the readers to check the supplementary material, any I feel there are so many things are given M&M that’s why these are given as supplementary material.
· Results
Results are looking fine to me.
· Discussion
Discussion section is lacking the citation. Try to add some citation in the discussion which can boost your arguments and make your comments valuable.
· Conclusion
Conclusion is not given in the MS. I suggest adding a conclusion section to summarize your study and arguments. This will keep the interest of reader and easy to get take home message.
All the best!
-
Author Response
Response to Reviewer 1 Comments
Point 1: Title “CPDOS: A Web Platform to Optimize Crop Planting Density Based on an Intelligent Algorithm” can be changed as “CPDOS: A Web based AI Platform to Optimize Crop Planting Density”. This is just suggestion, but the authors can keep the same title as they already given.
Response 1: Thanks for review’s comments, we have revised in Title part of manuscript as “CPDOS: A Web based AI Platform to Optimize Crop Planting Density”. We also believe that this modification can better highlight the core and eye-catching aspects of the article.
Point 2: Please write down some output of the Intelligent Algorithm for Optimization of Crop Planting Density, it will increase readers’ interest.
Response 2: Thank you for your comments，We have added some output in the abstract section of the manuscript and have reedited the abstract.（L26-34）
Point 3: Introduction is looking fine, change the citation style to normal text as the citations are given in the from of superscript. Another suggestion is to reduce the introduction by keeping most important and relevant citation and text.
Response 3: Thank you for your comments，We have changed the citation style to normal text in the from of superscript. We have reduced introduction by retaining the most important and relevant citations and texts.
Point 4: The M&M is well written and explained the model very well, I have one concern about the tables mentioned in M&M as supplementary files, it is quite for the readers to check the supplementary material, any I feel there are so many things are given M&M that’s why these are given as supplementary material.
Response 4: Thank you for your comments, We have added the tables in some supplementary files to the main materials and methods of the article.(Table1-8)
Point 5: Discussion section is lacking the citation. Try to add some citation in the discussion which can boost your arguments and make your comments valuable.
Response 5: Thank you for your comments, We have added some citation in the discussion.( Line828 ,Line852, Line860, Line867,)
Point 6: Conclusion is not given in the MS. I suggest adding a conclusion section to summarize your study and arguments. This will keep the interest of reader and easy to get take home message.
Response 6: Thank you for your comments, We have added a conclusion in the article.( Line885 )
Reviewer 2 Report
CPDOS: A Web Platform to Optimize Crop Planting Density Based on an Intelligent Algorithm
Securing an appropriate amount of food is the basic function of agriculture, which is carried out in the conditions of decreasing agricultural land area. It is expected that in 2050 it will amount to only 15 a. per person. Hence the constant efforts of farmers and scientists dealing with agronomy to optimize the conditions for growth, development and yielding of plants. In the case of cereals and potatoes, the size of the plant yield is determined by two features, i.e. the number of grains in the ear and the weight of 1000 grains, or in the case of potatoes, the number of tubers per bush and the average weight of the tuber, in the case of oilseeds (rapeseed) and legumes (soybean, field bean, pea) ) are the 3 primary characteristics (number of pods, number of seeds in a pod, weight of 1000 seeds). Adding another feature - plant density per 1 m2 allows you to calculate the yield per area unit, and this feature has a decisive influence on the size of the yield of plants and may also be negatively correlated with other elements of the yield structure. Plant density can be determined directly on the ground or indirectly from aerial, satellite or low-altitude (drone) photos, or using vegetation indices such as LAI, NDVI, NDRE, EVI and others. What can and is used in yield forecasting. Therefore, the work presented by the authors is very valuable, especially since the calculation platform is available for registration and use.
The authors conducted a thorough review of the literature on the methods of determining the graded factor, e.g. sowing density or incremental doses of N per single plant yield and yield per area unit. When discussing deterministic models, it was necessary to mention others widely described in the world literature (WOFOST, DAISY, APSIM, EPIC, CropSyst, STICS).
The authors selected 6 methods (equations) for determining the effect of plant density and fertilization dose on the yield of a single plant and proposed the use of neural networks to improve the estimation of the values of parameters (a,b,c) appearing in these equations. Moreover, in each case, they determined their biological interpretation. The root mean square error (RMSE) and coefficient of determination were used to evaluate the final results (estimated - observed). In the text of the article, please also include the relative root mean squared error (RRMSE). In the work, the authors use the term "optimal yield density model" or should it be understood as the dependence of yield on plant density(?). Maybe the word “density” is not needed in this case, because it is about the quantity of the yields, not its density.
Complete the conclusion or summary.
I believe, that the work is valuable. After introducing corrections, the work can be accepted for printing
Author Response
Response to Reviewer 2 Comments
Point 1: In the work, the authors use the term "optimal yield density model" or should it be understood as the dependence of yield on plant density(?). Maybe the word “density” is not needed in this case, because it is about the quantity of the yields, not its density.
Response 1: Thanks for review’s comments, CPDOS is based on six classic yield density models. We summarized the relationship between crop planting density and yield and organized six crop yield density models. CPDOS is built based on these six yield density models, which can be driven by planting data of different crop types to use the system. The system will use EGA genetic algorithm and BP neural network to select the most suitable yield density model and optimize crop planting density. CPDOS has three main modules: optimal planting density optimization, optimal planting density range prediction, and optimal planting density fertilization ratio optimization.
Point 2: Complete the conclusion or summary.
Response 2: Thank you for your comment. We have added conclusions to the article and reedited the abstract.（L885）
Reviewer 3 Report
This paper provided detailed equations and architect of the CPDOS system, and showed some results from the CPDOS. Overall, the paper is dis-organized and I do not see scientific merits in this manuscript.
There is a mis-alignment between Introduction, Methods and Results section. The objective statement (L150-151) does not say much about the specific goal that the authors were trying to achieve. The methodology section provides details about equations within the CPDOS. I do not see anything about how the results are generated in the Method section.
The results & graphics are poorly explained. Example, for figure 4, which crop is it generated for? What is the unit for yield, kernels per plant or kg/ha? What is the unit for planting density?
My other comments are listed below.
-
What does EGA, and BP stands for (L13 & 15)?
-
The Introduction section largely focuses on row crops (e.g. corn). However, the results present optimal yield and density for other crops (e.g. potatoes, sorghum, pepper, soybean, etc.). The authors need to include some background research information on these other crops in their study.
-
The authors seem to imply the CPDOS is a potato-based model (L164-166), however, the results are for corn, soybean, peppers, etc.
-
L150-151: I always appreciate when the authors explicitly state the objective of the study. However, this statement is vague. What’s the objective of this study? To provide an overview of the CPDOS and to validate the CPDOS in optimizing density for crop yield for smallholder fields in China?
-
L151-162: This is really lengthy. I do not believe these descriptions of the approach belongs to Introduction section. Rather, the authors should consider simplifying these texts to 2-4 lines to provide a brief overview of the methodologies in the Section 2 Materials and Method before section 2.1.
-
The Materials and Method section is really lengthy. I am not sure if such lengthy discussion is necessary without knowing the objective of this paper. Additionally, there are minimal references to the equations in this section. What’s the base of these approaches?
-
The authors need to use a metric unit (e.g. ha), not “mu”. The authors used “mu” in multiple places, e.g. plants/mu in L916 & l947, and ton/mu L947.
-
The authors used abbreviations without stating its full phrase in multiple places. Example, RGB (L142), GBP (L168), and MSE (L497) in addition to the two un-explained abbreviation in Abstract.
-
The authors mentioned “Huidong” in multiple places (e.g. L166, 833, 841) without any reference. This is not acceptable for scientific articles. Is there a published paper that the author is referring to?
-
Example, L833 claimed “Huidong believes that the yield…”. Which paper is it referring to?
-
Figure 7 is not reference. What is the “Y”, yield or density? What’s the unit?
Author Response
Response to Reviewer 3 Comments
Point 1: There is a mis-alignment between Introduction, Methods and Results section.
Response 1: Thanks for review’s comments, We have reedited the Introduction(L45-154), Materials and Methods(L155-720), and Results(L721-823) sections of the article, and added the Conclusion(L885-933) section to make the expression clearer.
Point 2: The objective statement (L150-151) does not say much about the specific goal that the authors were trying to achieve.
Response 2: Thank you for your comment. We have reedited the Introduction section of the article to clarify our research purpose(L141-L154). The research purpose of this article is to rely on modern computing technology to optimize planting density and promote excellent research results, providing important reference and basis for the development of intelligent crop planting plans. This article first selected six classical yield density models were selected and a genetic algorithm was used to estimate and optimize the model parameters , selected as the best model to match the current dataset. Secondly, after formulating the optimal yield density model and considering the production costs, the quantitative relationship between economic yield and planting density was obtained, and the optimal planting density range was calculated. Finally, based on the above, a crop planting density optimization system was developed and implemented. The system has three core modules; a crop yield density model optimization module, an optimal planting density calculation module, and a fertilization planting density optimization module. Using this system, users can better analyze existing data such as yield density, and realize planting density optimization.
Point 3: The methodology section provides details about equations within the CPDOS. I do not see anything about how the results are generated in the Method section.
Response 3: Thank you for your comment. We have reedited the Materials and Methods section of the article to clarify how the results are generated in the Method section .Due to the excessive length, we have elaborated on the process of obtaining results in the supplementary materials.
Point 4: The results & graphics are poorly explained. Example, for figure 4, which crop is it generated for? What is the unit for yield, kernels per plant or kg/ha? What is the unit for planting density?
Response 4: Thank you for your comment. We have annotated the coordinates in all the figures in the article with units(Figure 4 L758; Figure 5 L759; Figure 6 L800; Figure 7 L815). Figure 4 shows the curve of the optimal yield density model based on the data in Table 1(Potato). (L758)The left vertical axis is Yield per plant（tons / acre）, and the right vertical axis is Yield pre unit area（tons / acre）, and the horizontal axis is planting density（tons / acre）.
Point 5: What does EGA, and BP stands for (L13 & 15)?
Response 5: Thank you for your comment. We have added full names to these abbreviations for the convenience of readers' reading and understanding. EGA（Evolutionary Genetic Algorithm）is called evolutionary genetic algorithm, which is an optimization algorithm based on biological evolution process. It treats feasible solutions as individuals and seeks the optimal solution by simulating the evolution process.The role of EGA in the system is to estimate parameters and select the optimal yield density model. (L17-L18)
BP（ Back Propagation ） neural network is a multi-layer feedforward network trained by error backpropagation (referred to as error backpropagation). Its algorithm is called the BP algorithm, and its basic idea is the gradient descent method. It uses gradient search technology to minimize the mean square error between the actual output value and the expected output value of the network. The role of the BP neural network in CPDOS is to perform regression analysis on planting density and fertilization data using polynomial regression, and optimize it using genetic algorithms to find the combination of planting density and fertilization that maximizes yield.(L25)
Point 6: The Introduction section largely focuses on row crops (e.g. corn). However, the results present optimal yield and density for other crops (e.g. potatoes, sorghum, pepper, soybean, etc.). The authors need to include some background research information on these other crops in their study.
Response 6: Thank you for your comment. We mainly focused on the corn and soybean data studied by Mo Huidong, and also tested the optimal yield and density of other crops (such as potatoes, sorghum, chili peppers, soybeans, etc.), highlighting the strong applicability of the CPDOS system. We have added some background research information about these other crops in the supplementary materials.
Point 7: The authors seem to imply the CPDOS is a potato-based model (L164-166), however, the results are for corn, soybean, peppers, etc.
Response 7: Thank you for your comment. CPDOS can select the optimal yield density model from six crop yield density models based on MSE for the same crop variety at the same location, or optimize for different crop varieties at different locations to select the optimal yield density model from the six models. The experimental data for the parameters estimation and crop yield density model selection in this study were extracted from Huidong's potato yield density data , as presented in Table 1 (L183-L190). The optimum planting density range experimental data of the crop yield density model in this study was selected from the maize yield density data of Mo Huidong , as presented in Table 2 (L193-L199) . In this study, the experimental data of Xuguang's soybean was selected to optimize the optimal planting density and fertilization ratio when the crop yield was the largest, as shown in Table 3 (L202-L205) . Due to the similar patterns reflected on a type of crop, we also simulated and optimized yield density for multiple crops, demonstrating the universality of CPDOS. This study mainly focuses on Huidong's potato yield density data , Huidong's maize yield density data , and Xuguang's soybean yield density data , and the subsequent emergence of crop data such as sorghum, chili peppers, and soybeans is to verify the universality of the CPDOS system(L207-L210).
Point 8: L150-151: I always appreciate when the authors explicitly state the objective of the study. However, this statement is vague. What’s the objective of this study? To provide an overview of the CPDOS and to validate the CPDOS in optimizing density for crop yield for smallholder fields in China?
Response 8: Thank you for your comment. We have made changes to the issue of unclear expression of the research purpose of the article（L138-L141）. The research purpose of this article is to rely on modern computing technology to optimize planting density and promote excellent research results, providing important reference and basis for the development of intelligent crop planting plans. This article first selected six classical yield density models were selected and a genetic algorithm was used to estimate and optimize the model parameters , selected as the best model to match the current dataset. Secondly, after formulating the optimal yield density model and considering the production costs, the quantitative relationship between economic yield and planting density was obtained, and the optimal planting density range was calculated. Finally, based on the above, a crop planting density optimization system was developed and implemented. The system has three core modules; a crop yield density model optimization module, an optimal planting density calculation module, and a fertilization planting density optimization module. Using this system, users can better analyze existing data such as yield density, and realize planting density optimization.
Point 9: L151-162: This is really lengthy. I do not believe these descriptions of the approach belongs to Introduction section. Rather, the authors should consider simplifying these texts to 2-4 lines to provide a brief overview of the methodologies in the Section 2 Materials and Method before section 2.1.
Response 9: Thank you for your comment. We have reedited Introduction section to make it more concise and clear in expression.（L42-L151）
Point 10: The Materials and Method section is really lengthy. I am not sure if such lengthy discussion is necessary without knowing the objective of this paper. Additionally, there are minimal references to the equations in this section. What’s the base of these approaches?
Response 10: Thank you for your comment. We have reedited the materials and methods section to make it more concise and clear. We have provided detailed information on the basis of the equation in this section in the supplementary materials.
Point 11: The authors need to use a metric unit (e.g. ha), not “mu”. The authors used “mu” in multiple places, e.g. plants/mu in L916 & l947, and ton/mu L947.
Response 11: Thank you for your comment. We have changed the 'mu' used in multiple places to 'acre' or ''.
Point 12: The authors used abbreviations without stating its full phrase in multiple places. Example, RGB (L142), GBP (L168), and MSE (L497) in addition to the two un-explained abbreviation in Abstract.
Response 12: Thank you for your comment. We have provided explanations and explanations for the abbreviations of the entire text, making it clear to readers. RGB, also known as RedGreenBlue, is a color mode, which is the most commonly used color space in color images. It consists of three primary colors: red, green, and blue. Each color is represented by a numerical value between 0 and 255, which can be combined to represent different colors (L119). GBP is the British pound (Pound), which is the name of the national currency and monetary unit of the United Kingdom (L186). MSE (Mean square error) is a measure that reflects the degree of difference between the estimator and the estimated quantity (L389).
Point 13: The authors mentioned “Huidong” in multiple places (e.g. L166, 833, 841) without any reference. This is not acceptable for scientific articles. Is there a published paper that the author is referring to?Example, L833 claimed “Huidong believes that the yield…”. Which paper is it referring to?
Response 13: Thank you for your comment. “Huidong” is the abbreviation of Mr. “Mo Huidong”. The crop yield density data and six basic yield density models (Table 1 and 2) used in this article are derived from two papers by Mr.“Mo Huidong”( L186-L202 ) . To avoid interference with readers, we have changed Mr. Mo Huidong's abbreviation“Huidong”to“Mo Huidong”(L184,730,749,768,766,775,830).
Point 14: Figure 7 is not reference. What is the “Y”, yield or density? What’s the unit?
Response 14: Thank you for your comment. The horizontal axis in Figure 7 represents the predicted value of CPDOS for your crop yield, while the vertical axis represents the actual value of the crop yield. The graph shows a fit curve, and the closer the points in the graph are to the diagonal, the higher the accuracy of CPDOS.We have annotated the units in the figure， Real value Y : ; Forecast value Y : .(L815)
Round 2
Reviewer 1 Report
ok
ok
Author Response
Thanks for review’s comments. We have reedited and polished the article, such as abstract, introduction, conclusion, etc. These have been annotated in the article.
Reviewer 3 Report
The key referenced papers should be in English language and published in major international journals.
Author Response
Thanks for review’s comments, We have reedited the article according to your request. These have been annotated in the article. The research on classic yield density models has not been absent in previous international English literature, such as the six classic yield density models summarized by Mr. Mo Huidong based on this article 【Bunting, E. (1973). Plant density and yield of grain maize in England. The Journal of Agricultural Science, 81(3), 455-463. doi:10.1017/S0021859600086512】. However, there has been less and less research on classical yield density models since then, which has affected the optimization and development of these models. The research purpose of this study is to use genetic algorithms to estimate and optimize the parameters of six classic yield density models, and use modern computing techniques to optimize planting density, thereby promoting noteworthy research results. These results can provide important reference and basis for developing intelligent crop planting plans.
