# Exploring Precision Farming Scenarios Using Fuzzy Cognitive Maps

^{1}

^{2}

^{3}

^{*}

## Abstract

**:**

## 1. Introduction

## 2. Materials and Methods

#### 2.1. Traditional Fuzzy Cognitive Maps

_{i}to node C

_{j}means that C

_{i}happening causes C

_{j}to happen, and a negative edge from C

_{i}to C

_{j}means that C

_{i}happening prevents C

_{j}from happening.

_{i}and C

_{j}is associated with a weight w

_{ij}which varies in [−1, 1]. There are three different types of possible causalities between every pair of concepts C

_{i}and C

_{j}:

- ○
- w
_{ij}> 0, expresses positive causality, that is C_{i}causally increases C_{j}. - ○
- w
_{ij}< 0, expresses negative causality, that is C_{i}causally decreases C_{j}. - ○
- w
_{ij}= 0, designates no causality.

_{ij}), then the indirect effect of C

_{i}on C

_{j}on path (i, k

_{1},…,k

_{n}, j) is the product w

_{ik1}x w

_{k1k2}x …w

_{knj}, and the total effect is the sum of the path products. This weighting scheme removes indeterminacy from causal propagation and combination in FCMs [18]. Figure 2 shows an example of a Fuzzy Cognitive Map along with its weight matrix.

^{(t)}(state vector at time (t)), it can model the evolution of a scenario over time by evolving forward and letting concepts interact with one another. The next state of the system V

^{(t+1)}at time (t + 1) is produced by multiplying V

^{(t)}by the graph’s weight matrix E. In some FCM implementations, concepts are considered to have memory with a self-feedback link weight of 1. Thus, the next state value of each concept C

_{i}is elaborated, during simulation, by retrieving its value at the previous iteration, and adding it to the propagated weighted values of all factors C

_{j}that have a direct influence on the concept according to Equation (1).

_{i}

^{(k+1)}is the activation value of concept C

_{i}at iteration k+1, C

_{i}

^{(k)}is the value of node C

_{i}at iteration k, w

_{ji}is the weight of the cause–effect link between C

_{j}and C

_{i}, and F is a threshold function such as sigmoid, used for squashing the results of the sum between 0 and +1, or between −1 and +1 [45], as shown by Equation (2).

#### 2.2. FCM Approach Using Linguistic Fuzzy Influence

_{A}and μ

_{B}, respectively, is defined by a T-conorm mapping. One of the commonly used mappings is the maximum operator as shown by Equation (3):

_{A}(x) dx ÷ ∫μ

_{A}(x) dx

_{ij}) is calculated for the link between every pair of concepts C

_{i}and C

_{j}, prior to starting simulations. The simulation is then carried in a normal way, using the forward inference algorithm like in a conventional FCM model. To demonstrate how the linguistic terms are aggregated, let us consider the relation between K (Potassium) and Y (cotton yield), using expert knowledge from [28]:

- 1
^{st}Expert:“IF value of concept K is med THEN value of concept Y is medInfer: The influence from concept K towards concept Y is med” - 2
^{nd}Expert:“IF value of concept K is med THEN value of concept Y is highInfer: The influence from concept K towards concept Y is high” - 3
^{rd}Expert:“IF value of concept K is high THEN value of concept Y is very highInfer: The influence from concept K towards concept Y is very high”

_{K-Y}= 0.65. The process of aggregation and defuzzification is shown by Figure 3.

#### 2.3. Dynamic Rule-Based Fuzzy Cognitive Maps (DRBFCMs)

_{i}and C

_{j}is represented in the form of a FIS [31]. Each FIS is described using the Fuzzy Control Language (FCL) [55]. In FCL, a FIS is usually composed of one or more Function Blocks (FB). In DRBFCM models, each FB has one input fuzzy variable with membership functions to describe the threshold values of the cause concept, and an output fuzzy variable with membership functions to describe the threshold values of the effect concept, along with the defuzzification method. The FB is also made of one Rule Block (RB), composed of a set of rules, as well as the aggregation, activation and accumulation methods [31]. In DRBFCM, the rules have a single antecedent related to a concept’s state or variation, and a single consequent which is always a variation, representing a perturbation in the output concept. Since FCL supports only rules that map input concept states to output concepts states, the authors modified the FCL grammar to cope with rules describing concept variations [40]. The new FCL grammar has: (i) the “IN^ID” clause added to the subcondition, which is used to specify the causal variation, where “IN” is a keyword and “ID” denotes the cause variable; and (ii) the “ON^ID” clause added to the subconclusion to specify the effect variation, where “ON” is a keyword and “ID” denotes the effect variable:

- Type 1: IF C
_{i}is A THEN Variation is V_{i}ON C_{j} - Type 2: IF Variation is V
_{i}IN C_{i}THEN Variation is V_{j}ON C_{j}

#### 2.4. Simulation with SimulDRBFCM Models

_{j}of magnitude ΔC

_{j}, on the value of an output variable C

_{i}, with a potential change of magnitude ΔC

_{i}, we stimulate the system with magnitudes of change. Inference is carried according to an algorithm for combining effects on a given concept, and dealing with feedback using FCM inference. SimulDRBFCM inference computes the new system variables’ states using quantified perturbations produced by FCMs that are weighted using FISs. The inference algorithm is summarized in Figure 5.

^{(t+1)}is produced by multiplying the previous change (X

^{(t)}) by the graph’s weight adjacency matrix (E).

_{i}is calculated during simulation, by computing the propagated weighted estimated change of all concepts C

_{j}that have a direct influence on the concept C

_{i}, according to Equation (5).

## 3. Case Study: Cotton Yield SimulDRBFCM Model

#### 3.1. Cotton Yield Knowledge

#### 3.2. Cotton Yield Data

^{TM}yield monitor installed on a two row John Deere

^{TM}cotton picker [58]. After field harvesting was completed, a calibration procedure was performed to improve the yield estimation [59]. The FCM model has been developed based on a raster data GIS approach, i.e., the data was stored in a two-dimensional matrix that represents the spatial distribution of every factor in the field. Each cell of the matrix corresponds to an area of 10 × 10 m, which is the spatial resolution of the yield data model.

#### 3.3. Cotton Yield Simulations

_{K-Y}= 0.22 ± 1.88 × 10

^{−4}), compared to the traditional FCM (W

_{K-Y}= 0.6). The low influence produced by the SimulDRBFCM model seems to make sense as K produces a “high” variation when it is classified as “high” or “very high”. Nevertheless, by looking at the 360 cases of cotton yield data, K was classified as either “medium” or “low” all the time. Hence, SimulDRBFCM cotton yield model seems to generate weights that are coherent with the model structure and collected knowledge, and it produces weights that can be interpreted by tracing the rules that contributed to the results.

^{−1}under normal weather conditions. Table 5 shows some of the results under four scenarios, where the target acceptable decrease in cotton yield production was set to a maximum of 2.5%, 5%, 7.5%, and 10%. In a policy development exercise, the target could be set, for example, by a cost–benefit analysis or by stakeholders.

^{−2}, and the mean Phosphorus content is 15.83 ± 1.97. These values appear to be characteristic, independently from growing conditions, of a yield mean of 248.16 ± 9.91× 10

^{−1}, which corresponds nearly to the normal yield value in Greece.

^{−2}, and the mean Phosphorus content is 15.85 ± 2.62. These values appear to be characteristic of a yield mean of 245.59 ± 2.51.

^{−2}, and the mean Phosphorus content is 16.71 ± 5.65. These values appear to be characteristic of a yield mean of 243.51 ± 4.38.

^{−2}, and the mean Phosphorus content is 16.01 ± 5.87. These values appear to be characteristic of a yield mean of 240.80 ± 5.97.

#### 3.4. Statistical Significance Tests

^{2}, we cannot model the distribution of simulation scores with the normal distribution. However, we can use Student’s t-distribution [61], which approximates the normal distribution for a large simulation set n.

## 4. Conclusions and Future Work

## Author Contributions

## Conflicts of Interest

## References

- Biermacher, J.T.; Brorsen, B.W.; Epplin, F.M.; Solie, J.B.; Raun, W.R. The economic potential of precision nitrogen application with wheat based on plant sensing. Agric. Econ.
**2009**, 40, 397–407. [Google Scholar] [CrossRef] - Aubert, B.A.; Schroeder, A.; Grimaudo, J. IT as enabler of sustainable farming: An empirical analysis of farmers’ adoption decision of precision agriculture technology. Decis. Support Syst.
**2012**, 54, 510–520. [Google Scholar] [CrossRef] - Lencsés, E.; Takács, I.; Takács-György, K. Farmers’ Perception of Precision Farming Technology among Hungarian Farmers. Sustainability
**2014**, 6, 8452–8465. [Google Scholar] [CrossRef] - Seelan, S.K.; Laguette, S.; Casady, G.M.; Seielstad, G.A. Remote sensing applications for precision agriculture: A learning community approach. Remote Sens. Environ.
**2003**, 88, 157–169. [Google Scholar] [CrossRef] - Bongiovanni, R.; Lowenberg-Deboer, J. Precision Agriculture and Sustainability. Precis. Agric.
**2004**, 5, 359–387. [Google Scholar] [CrossRef] - Lowenberg-DeBoer, J.; Swinton, S. Economics of site-specific management in agronomic crops. In The State of site-Specific Management for Agriculture (Thestateofsites); American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America (ASA, CSSA, and SSSA): Madison, WI, USA, 1997; pp. 369–396. [Google Scholar]
- Batte, M.T. Factors influencing the profitability of precision farming systems. J. Soil Water Conserv.
**2000**, 55, 12–18. [Google Scholar] - Arnholt, M.; Batte, M.T.; Prochaska, S. Adoption and Use of Precision Farming Technologies: A Survey of Central Ohio Precision Farmers; Report Series: AEDE-RP-0011-01; Department of Agricultural, Environmental and Development Economics, The Ohio State University: Columbus, OH, USA, 2001. [Google Scholar]
- Rider, T.W.; Vogel, J.W.; Dille, J.A.; Dhuyvetter, K.C.; Kastens, T.L. An economic evaluation of site-specific herbicide application. Precis. Agric.
**2006**, 7, 379–392. [Google Scholar] [CrossRef] - Tey, Y.S.; Brindal, M. Factors influencing the adoption of precision agricultural technologies: A review for policy implications. Precis. Agric.
**2012**, 13, 713–730. [Google Scholar] [CrossRef] - Reichardt, M.; Jürgens, C.; Klöble, U.; Hüter, J.; Moser, K. Dissemination of precision farming in Germany: Acceptance, adoption, obstacles, knowledge transfer and training activities. Precis. Agric.
**2009**, 10, 525. [Google Scholar] [CrossRef] - Bramley, R.G.V. Lessons from nearly 20 years of Precision Agriculture research, development, and adoption as a guide to its appropriate application. Crop Pasture Sci.
**2009**, 60, 197–217. [Google Scholar] [CrossRef] - Torbett, J.C.; Roberts, R.K.; Larson, J.A.; English, B.C. Perceived importance of precision farming technologies in improving phosphorus and potassium efficiency in cotton production. Precis. Agric.
**2007**, 8, 127–137. [Google Scholar] [CrossRef] - Pierpaoli, E.; Carli, G.; Pignatti, E.; Canavari, M. Drivers of Precision Agriculture Technologies Adoption: A Literature Review. Procedia Technol.
**2013**, 8, 61–69. [Google Scholar] [CrossRef] - McBride, W.D.; Daberkow, S.G. Information and the adoption of precision farming technologies. J. Agribus.
**2003**, 21, 21–38. [Google Scholar] - Puerto, P.; Domingo, R.; Torres, R.; Pérez-Pastor, A.; García-Riquelme, M. Remote management of deficit irrigation in almond trees based on maximum daily trunk shrinkage. Water relations and yield. Agric. Water Manag.
**2013**, 126, 33–45. [Google Scholar] [CrossRef] - De la Rosa, J.M.; Conesa, M.R.; Domingo, R.; Aguayo, E.; Falagán, N.; Pérez-Pastor, A. Combined effects of deficit irrigation and crop level on early nectarine trees. Agric. Water Manag.
**2016**, 170, 120–132. [Google Scholar] [CrossRef] - Kosko, B. Fuzzy cognitive maps. Int. J. Man-Mach. Stud.
**1986**, 24, 65–75. [Google Scholar] [CrossRef] - Gray, S.A.; Gray, S.; De Kok, J.L.; Helfgott, A.E.; O’Dwyer, B.; Jordan, R.; Nyaki, A.; Analyzing, F. Using fuzzy cognitive mapping as a participatory approach to analyze change, preferred states, and perceived resilience of social-ecological systems. Ecol. Soc.
**2015**, 20, 11. [Google Scholar] [CrossRef] - Özesmi, U.; Özesmi, S.L. Ecological models based on people’s knowledge: A multi-step fuzzy cognitive mapping approach. Ecol. Model.
**2004**, 176, 43–64. [Google Scholar] [CrossRef] - Van Vliet, M.; Kok, K.; Veldkamp, T. Linking stakeholders and modellers in scenario studies: The use of Fuzzy Cognitive Maps as a communication and learning tool. Futures
**2010**, 42, 1–14. [Google Scholar] [CrossRef] - Kontogianni, A.D.; Papageorgiou, E.I.; Tourkolias, C. How do you perceive environmental change? Fuzzy Cognitive Mapping informing stakeholder analysis for environmental policy making and non-market valuation. Appl. Soft Comput.
**2012**, 12, 3725–3735. [Google Scholar] [CrossRef] - Samarasinghe, S.; Strickert, G. Mixed-method integration and advances in fuzzy cognitive maps for computational policy simulations for natural hazard mitigation. Environ. Model. Softw.
**2013**, 39, 188–200. [Google Scholar] [CrossRef] - Kafetzis, A.; McRoberts, N.; Mouratiadou, I. Using Fuzzy Cognitive Maps to Support the Analysis of Stakeholders’ Views of Water Resource Use and Water Quality Policy. In Fuzzy Cognitive Maps: Advances in Theory, Methodologies, Tools and Applications; Glykas, M., Ed.; Springer: Berlin/Heidelberg, Germany, 2010; pp. 383–402. [Google Scholar]
- Giordano, R.; Vurro, M. Fuzzy Cognitive Map to Support Conflict Analysis in Drought Management. In Fuzzy Cognitive Maps: Advances in Theory, Methodologies, Tools and Applications; Glykas, M., Ed.; Springer: Berlin/Heidelberg, Germany, 2010; pp. 403–425. [Google Scholar] [CrossRef]
- Papageorgiou, E.I.; Salmeron, J.L. A Review of Fuzzy Cognitive Maps Research During the Last Decade. IEEE Trans. Fuzzy Syst.
**2013**, 21, 66–79. [Google Scholar] [CrossRef] - Jetter, A.J.; Kok, K. Fuzzy Cognitive Maps for futures studies—A methodological assessment of concepts and methods. Futures
**2014**, 61, 45–57. [Google Scholar] [CrossRef] - Papageorgiou, E.I.; Markinos, A.; Gemptos, T. Application of fuzzy cognitive maps for cotton yield management in precision farming. Expert Syst. Appl.
**2009**, 36, 12399–12413. [Google Scholar] [CrossRef] - Papageorgiou, E.I.; Aggelopoulou, K.D.; Gemtos, T.A.; Nanos, G.D. Yield prediction in apples using Fuzzy Cognitive Map learning approach. Comput. Electron. Agric.
**2013**, 91, 19–29. [Google Scholar] [CrossRef] - Jayashree, L.S.; Palakkal, N.; Papageorgiou, E.I.; Papageorgiou, K. Application of fuzzy cognitive maps in precision agriculture: A case study on coconut yield management of southern India’s Malabar region. Neural Comput. Appl.
**2015**, 26, 1963–1978. [Google Scholar] [CrossRef] - Ross, T.J. Fuzzy Logic with Engineering Applications; Wiley: New York, NJ, USA, 2009. [Google Scholar]
- Srinivasan, S.P.; Malliga, P. A new approach of adaptive Neuro Fuzzy Inference System (ANFIS) modeling for yield prediction in the supply chain of Jatropha. In Proceedings of the 2010 IEEE 17th International Conference on Industrial Engineering and Engineering Management, Xiamen, China, 29–31 October 2010; pp. 1249–1253. [Google Scholar] [CrossRef]
- Naderloo, L.; Alimardani, R.; Omid, M.; Sarmadian, F.; Javadikia, P.; Torabi, M.Y.; Alimardani, F. Application of ANFIS to predict crop yield based on different energy inputs. Measurement
**2012**, 45, 1406–1413. [Google Scholar] [CrossRef] - Khashei-Siuki, A.; Kouchkzadeh, M.; Ghahraman, B. Predicting dryland wheat yield from meteorological data using expert system, Khorasan Province, Iran. J. Agric. Sci. Technol.
**2011**, 13, 627–640. [Google Scholar] - Hosseinpourtehrani, M.; Ghahraman, B. Optimal reservoir operation for irrigation of multiple crops using Fuzzy logic. Asian J. Appl. Sci.
**2011**, 4, 493–513. [Google Scholar] - Tremblay, N.; Bouroubi, M.Y.; Panneton, B.; Guillaume, S.; Vigneault, P.; Bélec, C. Development and validation of fuzzy logic inference to determine optimum rates of N for corn on the basis of field and crop features. Precis. Agric.
**2010**, 11, 621–635. [Google Scholar] [CrossRef] - Mazloumzadeh, S.M.; Shamsi, M.; Nezamabadi-Pour, H. Fuzzy logic to classify date palm trees based on some physical properties related to precision agriculture. Precis. Agric.
**2010**, 11, 258–273. [Google Scholar] [CrossRef] - Tagarakis, A.; Koundouras, S.; Papageorgiou, E.I.; Dikopoulou, Z.; Fountas, S.; Gemtos, T.A. A fuzzy inference system to model grape quality in vineyards. Precis. Agric.
**2014**, 15, 555–578. [Google Scholar] [CrossRef] - Carvalho, J.P.; Tomé, J.A.B. Fuzzy Mechanisms for Qualitative Causal Relations. In Views on Fuzzy Sets and Systems from Different Perspectives: Philosophy and Logic, Criticisms and Applications; Seising, R., Ed.; Springer: Berlin/Heidelberg, Germany, 2009; pp. 393–415. [Google Scholar] [CrossRef]
- Mourhir, A.; Rachidi, T.; Papageorgiou, E.I.; Karim, M.; Alaoui, F.S. A cognitive map framework to support integrated environmental assessment. Environ. Model. Softw.
**2016**, 77, 81–94. [Google Scholar] [CrossRef] - Zadeh, L.A. Fuzzy sets. Inf. Control
**1965**, 8, 338–353. [Google Scholar] [CrossRef] - Axelrod, R. Structure of Decision: The Cognitive Maps of Political Elites; Princeton University: Princeton, NJ, USA, 1976. [Google Scholar]
- Calais, G. Fuzzy cognitive maps theory: Implications for interdisciplinary reading: National implications. FOCUS Coll. Univ. Sch.
**2008**, 2, 1–16. [Google Scholar] - Kosko, B. Adaptive inference in fuzzy knowledge networks. In Proceedings of the First IEEE International Conference on Neural Networks, San Diego, CA, USA, 21–24 June 1987. [Google Scholar]
- Tsadiras, A.K. Comparing the inference capabilities of binary, trivalent and sigmoid fuzzy cognitive maps. Inf. Sci.
**2008**, 178, 3880–3894. [Google Scholar] [CrossRef] - Taber, R. Knowledge processing with Fuzzy Cognitive Maps. Expert Syst. Appl.
**1991**, 2, 83–87. [Google Scholar] [CrossRef] - Kosko, B. Fuzzy knowledge combination. Int. J. Intell. Syst.
**1986**, 1, 293–320. [Google Scholar] [CrossRef] - Kosko, B. Hidden patterns in combined and adaptive knowledge networks. Int. J. Approx. Reason.
**1988**, 2, 377–393. [Google Scholar] [CrossRef] - Adriaenssens, V.; Baets, B.D.; Goethals, P.L.M.; Pauw, N.D. Fuzzy rule-based models for decision support in ecosystem management. Sci. Total Environ.
**2004**, 319, 1–12. [Google Scholar] [CrossRef] - Kok, K. The potential of Fuzzy Cognitive Maps for semi-quantitative scenario development, with an example from Brazil. Glob. Environ. Chang.
**2009**, 19, 122–133. [Google Scholar] [CrossRef] - Helfgott, A.; Lord, S.; Bean, N.; Wildenberg, M.; Gray, S.; Gray, S.; Vervoort, J.; Kok, K.; Ingram, J. Working Paper 1: Clarifying Fuzziness: Fuzzy Cognitive Maps, Neural Networks and System Dynamics Models in Participatory Social and Environmental Decision-Aiding Processes; EU FP7 TRANSMANGO: Brussels, Belgium, 2015. [Google Scholar]
- Sterman, J.D. Business Dynamics: Systems Thinking and Modeling for a Complex World; Irwin/McGraw-Hill: Boston, MA, USA, 2000; Volume 19. [Google Scholar]
- Oja, E.; Ogawa, H.; Wangviwattana, J. Learning in nonlinear constrained Hebbian networks. In Artificial Neural Networks; Kohonen, T., Makisara, K., Simula, O., Kangas, J., Eds.; Elsevier: Amsterdam, The Netherlands, 1991; pp. 385–390. [Google Scholar]
- Papageorgiou, E.I.; Groumpos, P.P. A weight adaptation method for fuzzy cognitive map learning. Soft Comput.
**2005**, 9, 846–857. [Google Scholar] [CrossRef] - International Electrotechnical Commission (IEC). IEC 61131—Programmable Controllers-Part 7: Fuzzy Control Programming. International Electrotechnical Commission Technical Committee Industrial Process Measurement and Control; IEC: Geneva, Switzerland, 2000. [Google Scholar]
- Mamdani, E.H. Application of fuzzy algorithms for control of simple dynamic plant. Electr. Eng. Proc. IEEE
**1974**, 121, 1585–1588. [Google Scholar] [CrossRef] - Mourhir, A. Papageorgiou EI Empirical Comparison of Fuzzy Cognitive Maps and Dynamic Rule-based Fuzzy Cognitive Maps. In Proceedings of the Thirteenth International Conference on Autonomic and Autonomous Systems (ICAS 2017), Barcelona, Spain, 21–25 May 2017. [Google Scholar]
- Gemtos, A.; Markinos, A.; Toulios, L.; Pateras, D.; Zerva, G. Precision farming applications in cotton fields of Greece. In Proceedings of the 2004 CIGR International Conference, Beijing, China, 11–14 Ocrober 2004; pp. 10–14. [Google Scholar]
- Markinos, A.T.; Gemtos, T.A.; Pateras, D.; Toulios, L.; Zerva, G.; Papaeconomou, M. The influence of cotton variety in the calibration factor of a cotton yield monitor. Oper. Res.
**2005**, 5, 165–176. [Google Scholar] [CrossRef] - Borror, C.M. Statistical decision making. In The Certified Quality Engineer Handbook, 3rd ed.; Borror, C.M., Ed.; ASQ Quality Press: Milwaukee, WI, USA, 2009; pp. 418–472. [Google Scholar]
- Gosset, W.S. The probable error of a mean. Biometrika
**1908**, 6, 1–25. [Google Scholar] [CrossRef] - Krzywinski, M.; Altman, N. Points of significance: Significance, P values and t-tests. Nat. Methods
**2013**, 10, 1041–1042. [Google Scholar] [CrossRef] [PubMed] - Fay, M.P.; Proschan, M.A. Wilcoxon-Mann-Whitney or t-test? On assumptions for hypothesis tests and multiple interpretations of decision rules. Stat. Surv.
**2010**, 4, 1–39. [Google Scholar] [CrossRef] [PubMed]

**Figure 7.**With probability q = 0.95 (shaded area), the number of correct SimulDRBFCM simulations lies in an interval [$\overline{\mathrm{x}}-\mathrm{d},\text{}\overline{\mathrm{x}}+\mathrm{d}$] around the mean sentence score $\overline{\mathrm{x}}$.

**Figure 8.**Depiction and verification of the yield decrease percentage distribution produced on the set of the n = 2847 cases.

Concept | Description |
---|---|

C1: EC | Soil shallow electrical conductivity Veris (mS/m) |

C2: Mg | Magnesium (ppm) |

C3: Ca | The measured calcium in the soil in depth 0–30 cm (ppm) |

C4: Na | The measured Na (Sodium) in the soil in depth 0–30 cm (ppm) |

C5: K | The measured Potassium in the soil in depth 0–30 cm (ppm) |

C6: P | The measured Phosphorus in the soil in depth 0–30 cm (ppm) |

C7: N | The measured NO_{3} in the soil profile of 0–30 cm (ppm) |

C8: OM | The percent organic matter content in soil profile in depth 0–30 cm |

C9: Ph | The pH of the soil in depth 0–30 cm |

C10: S | The percent of the sand in the soil samples in depth 0–30 cm |

C11: Cl | The percent of the clay in samples in depth 0–30 cm |

C12: Y | Seed cotton yield from 1^{st} picking measured by yield monitor (t ha^{−1}) |

Concept | Membership Function | Concept | Membership Function |
---|---|---|---|

C1: (EC) | VL = TRAPE 0 0 7.5 15 | C7: (N) | VL = TRAPE 0 0 3 8 |

L = TRAPE 10 18 18 25 | L = TRAPE 5 8 8 17.5 | ||

M = TRAPE 25 28 28 35 | M = TRAPE 12 20 20 27.5 | ||

H = TRAPE 30 38 38 45 | H = TRAPE 22 32 32 45 | ||

VH = TRAPE 40 45 100 100 | VH = TRAPE 35 40 45 45 | ||

C2: (Mg) | VL = TRAPE 0 0 60 120 | C8: (OM) | L = TRAPE 0 0 0.6 1.1 M = TRAPE 0.5 1.5 1.5 2.5 H = TRAPE 1.8 2.1 3 3 |

L = TRAPE 60 140 140 240 | |||

M = TRAPE 160 290 290 360 | |||

H = TRAPE 300 500 500 1400 | |||

VH = TRAPE 700 950 1400 1400 | |||

C3: (Ca) | VL = TRAPE 0 0 455 1000 | C9: (Ph) | VL = TRAPE 0 0 4 5 |

L = TRAPE 545 1273 1273 2000 | L = TRAPE 4 5 5 6 | ||

M = TRAPE 1363 2455 2455 3000 | SL = TRAPE 5 6 6 7 | ||

H = TRAPE 2637 3909 3909 5000 | M = TRAPE 6 7 7 8 | ||

VH = TRAPE 4000 4380 5000 5000 | SH = TRAPE 7 8 8 9 | ||

C4: (Na) | VL = TRAPE 0 0 26 59 | C10: (S) | L = TRAPE 0 0 15 30 M = TRAPE 20 45 45 70 H = TRAPE 60 75 75 90 VH = TRAPE 80 90 100 100 |

L = TRAPE 32 70 70 123 | |||

M = TRAPE 80 140 140 200 | |||

H = TRAPE 156 250 250 600 | |||

VH = TRAPE 350 450 600 600 | |||

C5: (K) | VL = TRAPE 0 0 24 65 | C11: (Cl) | L = TRAPE 0 0 12.5 20 M = TRAPE 10 22.5 22.5 35 H = TRAPE 30 37.7 60 60 |

L = TRAPE 30 81 81 135 | |||

M = TRAPE 88 152 152 230 | |||

H = TRAPE 190 275 275 600 | |||

VH = TRAPE 300 470 600 600 | |||

C6: (P) | VL = TRAPE 0 0 5 10 | C12: (Y) | L = TRAPE 0 0 2 3 H = TRAPE 2 3 6 6 |

L = TRAPE 5 12.5 12.5 20 | |||

M = TRAPE 12.5 22 22 31.5 | |||

H = TRAPE 25 38 38 50 | |||

VH = TRAPE 40 45 50 50 |

Concept | If-Then Rules | Concept | If-Then Rules |
---|---|---|---|

C1:(EC) | IF EC IS VL THEN VAR IS PVL ON Y | C6:(P) | IF P IS VL THEN VAR IS PM ON Y IF P IS L THEN VAR IS PM ON Y IF P IS M THEN VAR IS PM ON Y IF P IS H THEN VAR IS PM ON Y IF P IS VH THEN VAR IS PM ON Y |

IF EC IS M THEN VAR IS PL ON Y | |||

IF EC IS H THEN VAR IS PM ON Y | |||

IF EC IS H THEN VAR IS PH ON Y | |||

IF EC IS VH THEN VAR IS PH ON Y | |||

IF EC IS L THEN VAR IS PVL ON Y | |||

C2:(Mg) | IF Mg IS VL THEN VAR IS NL ON Y IF Mg IS L THEN VAR IS NL ON Y IF Mg IS M THEN VAR IS NM ON Y IF Mg IS H THEN VAR IS NM ON Y IF Mg IS VH THEN VAR IS NM ON Y | C7:(N) | IF N IS VL THEN VAR IS PVL ON Y |

IF N IS L THEN VAR IS PL ON Y | |||

IF N IS M THEN VAR IS PL ON Y | |||

IF N IS H THEN VAR IS PL ON Y | |||

IF N IS VH THEN VAR IS PM ON Y | |||

IF N IS VH THEN VAR IS PL ON Y | |||

C3:(Ca) | IF Ca IS VL THEN VAR IS PM ON Y | C8:(OM) | IF OM IS L THEN VAR IS PL ON Y IF OM IS M THEN VAR IS PL ON Y IF OM IS H THEN VAR IS PM ON Y |

IF Ca IS L THEN VAR IS PL ON Y | |||

IF Ca IS M THEN VAR IS PM ON Y | |||

IF Ca IS H THEN VAR IS PM ON Y | |||

IF Ca IS VH THEN VAR IS PM ON Y | |||

IF Ca IS VH THEN VAR IS PL ON Y | |||

C4:(Na) | IF Na IS VL THEN VAR IS NVH ON Y IF Na IS VL THEN VAR IS NH ON Y IF Na IS L THEN VAR IS NM ON Y IF Na IS M THEN VAR IS NL ON Y IF Na IS H THEN VAR IS NH ON Y IF Na IS VH THEN VAR IS NH ON Y IF Na IS VH THEN VAR IS NVH ON Y | C9: (Ph) | IF Ph IS VL THEN VAR IS PVL ON Y |

IF Ph IS VL THEN VAR IS PL ON Y | |||

IF Ph IS L THEN VAR IS PVL ON Y | |||

IF Ph IS SL THEN VAR IS PVL ON Y | |||

IF Ph IS M THEN VAR IS PVL ON Y | |||

IF Ph IS SH THEN VAR IS PVL ON Y | |||

IF Ph IS H THEN VAR IS PVL ON Y | |||

IF Ph IS H THEN VAR IS PL ON Y | |||

IF Ph IS VH THEN VAR IS PVL ON Y | |||

IF Ph IS VH THEN VAR IS PM ON Y | |||

C5:(K) | IF K IS VL THEN VAR IS PVL ON Y | C10: (S) | IF S IS L THEN VAR IS NM ON Y |

IF K IS L THEN VAR IS PVL ON Y | IF S IS L THEN VAR IS NH ON Y | ||

IF K IS M THEN VAR IS PM ON Y | IF S IS M THEN VAR IS NM ON Y | ||

IF K IS H THEN VAR IS PM ON Y | IF S IS M THEN VAR IS NL ON Y | ||

IF K IS VH THEN VAR IS PM ON Y | IF S IS H THEN VAR IS NM ON Y | ||

IF K IS VH THEN VAR IS PH ON Y | IF S IS VH THEN VAR IS NH ON Y | ||

C11: (Cl) | IF Cl IS L THEN VAR IS PM ON Y | ||

IF Cl IS M THEN VAR IS PM ON Y | |||

IF Cl IS H THEN VAR IS PM ON Y |

Concept | Cotton Yield (Y) | |
---|---|---|

FCM | SimulDRBFCM ^{a} | |

EC | 0.25 | 0.22 ± 2.0488 × 10^{−4} |

Mg | −0.4 | −0.48 ± 4.27 × 10^{−2} |

Ca | 0.5 | 0.48 ± 3.15 × 10^{−2} |

Na | −0.7 | −0.7 ± 2.00 × 10^{−15} |

K | 0.6 | 0.22 ± 1.88 × 10^{−4} |

P | 0.5 | 0.49 ± 5.24 × 10^{−2} |

N | 0.4 | 0.35 ± 1.51 × 10^{−2} |

OM | 0.4 | 0.35 ± 8.23 × 10^{−3} |

pH | 0.1 | 0.26 ± 2.37 × 10^{−2} |

S | −0.6 | −0.5 ± 7.07 × 10^{−16} |

Cl | 0.5 | 0.5 ± 5.47 × 10^{−16} |

^{a}Mean values shown with standard deviation.

ΔK (%) | ΔP (%) | ΔN (%) | Average Yield Decrease (%) | High to Low Yield Decrease (%) |
---|---|---|---|---|

Maximum Yield Decrease: 2.5% | ||||

−10 | −10 | - | 2.55 | 1 |

Maximum Yield Decrease: 5% | ||||

- | −40 | - | 4.80 | 3 |

- | −10 | −40 | 5.06 | 3 |

- | - | −50 | 4.80 | 3 |

−10 | −20 | −10 | 4.80 | 3 |

−10 | −30 | - | 5.06 | 3 |

−20 | - | −20 | 4.80 | 3 |

−20 | −10 | −10 | 5.06 | 3 |

Maximum Yield Decrease: 7.5% | ||||

- | −50 | −10 | 7.19 | 6 |

- | −40 | −20 | 6.92 | 5 |

- | −10 | −60 | 7.19 | 6 |

- | - | −70 | 6.92 | 5 |

−10 | −30 | −20 | 7.19 | 6 |

−10 | −20 | −30 | 6.92 | 5 |

−20 | −10 | −30 | 7.19 | 6 |

−20 | - | −40 | 6.92 | 5 |

−30 | −20 | - | 6.92 | 5 |

−40 | −10 | - | 7.19 | 6 |

−40 | - | −10 | 6.92 | 5 |

Maximum Yield Decrease: 10% | ||||

- | −60 | −20 | 9.71 | 7 |

- | −20 | −70 | 9.71 | 7 |

- | −30 | −60 | 10.00 | 8 |

- | −70 | −10 | 10.00 | 8 |

−10 | - | −80 | 9.71 | 7 |

−10 | −40 | −30 | 9.71 | 7 |

−10 | −10 | −70 | 10.00 | 8 |

−10 | −50 | −20 | 10.00 | 8 |

−20 | −20 | −40 | 9.71 | 7 |

−20 | −30 | −30 | 10.00 | 8 |

−30 | - | −50 | 9.71 | 7 |

−30 | −40 | - | 9.71 | 7 |

−30 | −10 | −40 | 10.00 | 8 |

−40 | −20 | −10 | 9.71 | 7 |

−40 | −30 | - | 10.00 | 8 |

−50 | - | −20 | 9.71 | 7 |

−50 | −10 | −10 | 10.00 | 8 |

−60 | −80 | −90 | 10.00 | 8 |

Scenario | MIN | MAX | MEAN | SDV ^{a} |
---|---|---|---|---|

2.5% | ||||

K | 0.207 | 0.4 | 0.279586 | 0.0634 |

P | 11.832 | 20.01 | 15.83779 | 1.97 |

N | 6.24 | 16.9 | 11.02034 | 3.36 |

Y | 246.1704 | 249.721 | 248.1631 | 0.991 |

5% | ||||

K | 0.161 | 0.51 | 0.291374 | 0.0861 |

P | 10.353 | 23.72 | 15.85973 | 2.62 |

N | 4.68 | 16.9 | 9.949553 | 3.06 |

Y | 240.5466 | 249.8426 | 245.5901 | 2.51 |

7.5% | ||||

K | 0.138 | 0.57 | 0.294396 | 0.0845 |

P | 7.395 | 39.04 | 16.7129 | 5.65 |

N | 2.34 | 26.3 | 9.433609 | 3.52 |

Y | 233.8835 | 249.9708 | 243.5175 | 4.38 |

10% | ||||

K | 0.092 | 0.57 | 0.29462 | 0.0901 |

P | 4.437 | 39.04 | 16.01736 | 5.87 |

N | 0.78 | 26.3 | 9.088008 | 4.11 |

Y | 226.8011 | 249.9879 | 240.8015 | 5.97 |

^{a}SDV: Standard Deviation.

Paired Sample Names | Paired Differences | ||||||
---|---|---|---|---|---|---|---|

Mean | Std. Deviation | Std. Error Mean | 95% Confidence Interval of the Difference | t | |||

Lower | Upper | ||||||

Pair 1 | kdecr-yielddecr | −0.08693991 | 0.15064377 | 0.00282330 | −0.09247583 | −0.08140398 | −30.423 |

Pair 2 | pdecr-yielddecr | −0.13095115 | 0.18151794 | 0.00340193 | −0.13762165 | −0.12428064 | −38.64 |

Pair 3 | Ndecr-yielddecr | −0.19255686 | 0.22737567 | 0.00426138 | −0.20091556 | −0.18420415 | −45.56 |

Paired Sample Names | Paired Differences | ||||||
---|---|---|---|---|---|---|---|

Mean | Std. Deviation | Std. Error Mean | 97% Confidence Interval of the Difference | t | |||

Lower | Upper | ||||||

Pair 1 | kdecr-yielddecr | −0.08693991 | 0.15064377 | 0.00282330 | −0.09247583 | −0.08140398 | −30.755 |

Pair 2 | pdecr-yielddecr | −0.13095115 | 0.18151794 | 0.00340193 | −0.13762165 | −0.12428064 | −38.433 |

Pair 3 | ndecr-yielddecr | −0.19255686 | 0.22737567 | 0.00426138 | −0.20091556 | −0.18420415 | −45.167 |

Paired Sample Names | Paired Differences | ||||||
---|---|---|---|---|---|---|---|

Mean | Std. Deviation | Std. Error Mean | 99% Confidence Interval of the Difference | t | |||

Lower | Upper | ||||||

Pair 1 | kdecr-yielddecr | −0.08693991 | 0.15064377 | 0.00282330 | −0.09247583 | −0.08140398 | −30.794 |

Pair 2 | pdecr-yielddecr | −0.13095115 | 0.18151794 | 0.00340193 | −0.13762165 | −0.12428064 | −38.493 |

Pair 3 | ndecr-yielddecr | −0.19255686 | 0.22737567 | 0.00426138 | −0.20091556 | −0.18420415 | −45.187 |

Match | N | Mean Rank | Sum of Ranks | |
---|---|---|---|---|

Yield-reduction | 1 | 17,624 | 11,102.63 | 1.96×10^{8} |

2 | 3071 | 6017.30 | 18,479,117.00 | |

Total | 20,695 | |||

K-reduction | 1 | 17,624 | 10,501.16 | 1.85×10^{8} |

2 | 3071 | 9469.03 | 29,079,397.00 | |

Total | 20,695 | |||

P-reduction | 1 | 17,624 | 10,273.80 | 1.81×10^{8} |

2 | 3071 | 10,773.84 | 33,086,475.00 | |

Total | 20,695 | |||

N-reduction | 1 | 17,624 | 10,348.00 | 1.82×10^{8} |

2 | 3,071 | 10,348.00 | 31,778,708.00 | |

Total | 20,695 |

Yield-reduction | K-reduction | P-reduction | N-reduction | |
---|---|---|---|---|

Mann–Whitney U | 1.376×10^{7} | 2.436×10^{7} | 2.575×10^{7} | 2.706×10^{7} |

Wilcoxon W | 1.848×10^{7} | 2.908×10^{7} | 1.811××10^{8} | 3.178×10^{7} |

Z | −43.531 | −8.863 | −4.83 | 0.000 |

Asymp. Sig. (two-tailed) | 0.000 | 0.000 | 0.000 | 1.000 |

^{a}Grouping Variable: Match.

© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Mourhir, A.; Papageorgiou, E.I.; Kokkinos, K.; Rachidi, T. Exploring Precision Farming Scenarios Using Fuzzy Cognitive Maps. *Sustainability* **2017**, *9*, 1241.
https://doi.org/10.3390/su9071241

**AMA Style**

Mourhir A, Papageorgiou EI, Kokkinos K, Rachidi T. Exploring Precision Farming Scenarios Using Fuzzy Cognitive Maps. *Sustainability*. 2017; 9(7):1241.
https://doi.org/10.3390/su9071241

**Chicago/Turabian Style**

Mourhir, Asmaa, Elpiniki I. Papageorgiou, Konstantinos Kokkinos, and Tajjeeddine Rachidi. 2017. "Exploring Precision Farming Scenarios Using Fuzzy Cognitive Maps" *Sustainability* 9, no. 7: 1241.
https://doi.org/10.3390/su9071241