R Taghizadeh-Mehrjardi1; F Sarmadian; A. A Zolfaghari; A. Jafari
Abstract
Introduction: Cation exchange capacity (CEC) has long been input parameter of many environmental models (Manrique et al., 1991). Added to this, CEC data can give more clear and complete interpretation of soil, plant nutrition process and consequently fertilizer and soil amendment requirements. Laboratory ...
Read More
Introduction: Cation exchange capacity (CEC) has long been input parameter of many environmental models (Manrique et al., 1991). Added to this, CEC data can give more clear and complete interpretation of soil, plant nutrition process and consequently fertilizer and soil amendment requirements. Laboratory analysis is the most accurate method for direct measurement of CEC. However, direct measurement of CEC is difficult, particularly in the soils of arid and semi-arid regions of Iran, due to large amounts of calcium carbonate that makes measuring expensive, laborious, and time-consuming (Amini et al., 2005). It can be an appropriate approach to predict CEC from readily available properties via developing nonparametric or parametric methods (Minasny et al., 1999). Therefore, the objectives of this study were to compare and apply different data mining approches including multi-linear regression (MLR), multi-nonlinear regression (MNR), cascade neural network (CNN), two radial base functions (RBF), multi-layer perceptron neural network (MLP), and adaptive neuro-fuzzy inference system (ANFIS) to estimate cation exchange capacity in different soils of Iran. Materials and Methods: For this purpose, 1770 soil samples were selected from different sites in Iran from which 356 samples were used as the testing data, and the remaining 1414 soils were employed as the training. The soil samples were dried, crushed and passed through a 2 mm sieve to prepare for physical and chemical analyses. The percentages of sand (50 -2000 mμ), silt (2-50 mμ) and clay (<2μm) were determined using the hydrometer method according to USDA soil textural classification system. The soil organic carbon was determined using Walkly-Black method and the CEC was measured by the standard method. Then the data mining techniques (i.e. MLR, MNR, CNN, RBF, MLP, ANFIS) were applied to predict CEC from readily available data (i.e. soil organic carbon and clay percentages). Finally, to compare efficiencies of these techniques, different error criteria including root mean square error (RMSE), mean error (ME), coefficient of determination (R2) and relative improvement (RI) were applied. In the present research, an effort was made to calculate the uncertainty of pedotransfer functions using Monte Carlo technique. Results and Discussion: Statistical analyses indicated the soil organic matter and soil texture have the highest variation. For example, variation of SOM has ranged from 0.01 to 2.94. Investigation of correlation coefficients shows that CEC is more related to the parameters, clay and soil organic matter content. Thus, the parameters, clay, silt, sand and organic carbon content were the input independent variables (readily available properties), and the CEC was an output dependent variable in this study. Root mean square error (RMSE) of linear and nonlinear regression was 4.74 and 4.71 meq 100g-1, respectively. This indicates that both methods are able to properly and equally predict CEC. Nonlinear recession equation increased the accuracy of prediction by 0.6 %. Results show that nonparametric artificial neural networks do not increase the accuracy of prediction CEC, significantly. The best result of neural networks was obtained using MLP. Nonparametric regression tree accuracy was slightly better than artificial neural network methods (4.53 and 4.61 meq 100g-1, respectively). The best method for prediction of CEC was ANFIS (RMSE=4.02 meq 100g-1). The accuracy of prediction using this method was 15 % more than linear regression. Moreover, the ANFIS model on the partitioned data by fuzzy k-means cloud enhances the prediction accuracy up to 26%. Monte Carlo results indicate the highest and lowest uncertainty belongs to MLR and ANFIS models, respectively. Conclusion: In the present research, different data mining techniques were applied to predict CEC in various ranges of soils. The data base related to 1770 soil samples was gathered from all over Iran. Results of the comparison indicate the highest prediction accuracy belongs to ANFIS model. Moreover, partitioning the data base to four groups enhances the accuracy of models. This result confirms that pedotransfer functions are more reliable only on the range of existing data. Overall, our efforts resulted only in R2 of 0.58. This means that soil organic matter and clay percentage could only model the 58% CEC variation. This suggests we should incorporate more input data including kind of clay mineral, percentage of calcium carbonate, gypsum, and etc.
L Naderloo; R alimardani; M Omid; F Sarmadian; H Javadikia; M. Y Torabi
Abstract
Introduction: Social, technical and economic factors in addition to environmental, soil and climate factors affect crop yield and cultivation. This study was implemented to know the impact of age, experience and literacy level of farmers as social factors and access to water supply, roads, silo, labor, ...
Read More
Introduction: Social, technical and economic factors in addition to environmental, soil and climate factors affect crop yield and cultivation. This study was implemented to know the impact of age, experience and literacy level of farmers as social factors and access to water supply, roads, silo, labor, tractors and machinery and conservation tillage as technical-mechanization factors on crop yield. Fuzzy rule-based inference system converts the complex decision-making problems to the smaller criteria and makes easier the multi-criteria evaluation process. So we decided to use fuzzy approach to modeling the social and technical-mechanization indices. The main disadvantage of fuzzy systems is their inability to learn. So, the optimization of fuzzy systems is the most important step in its implementation. Genetic algorithm (GA) approach is used as a complementation of fuzzy model to optimize fuzzy rules. One method to optimize the fuzzy rules is Pittsburgh method in GA. In this method, one gene is used for every rule and the gene value finds out the rule.The kind of membership function will have a great impact on the result. The kinds of membership function for fuzzy sets involve triangular, trapezoidal, generalized bell, Gaussian, Gaussian combination, Sigmoidal, product of two sigmoidal, difference between two sigmoidal, Π, Z and S shapes. The objectives of this study are: 1- providing two fuzzy models for the social and technical-mechanization indices for wheat production 2- optimizing the fuzzy rules and the type of membership function for the fuzzy set. Materials and Methods: Fuzzy toolbox of MATLAB software ver. 7.8.0 (R2009a) was used to design fuzzy model. Fuzzy inference system (FIS) used in this study was Mamdani type that is based on if-then rules. The age, experience and literacy level of farmers were selected as input data for fuzzy social model. Access to water supply, roads, silage, labor, tractors and machinery and conservation tillage equipment were selected as input data for fuzzy technical-mechanization model. Mamdani fuzzy inference system was used to design models. Fuzzy rules were written by a mechanization expert knowledge. To correct written rules, the method of Pittsburgh in GA was used to optimize the fuzzy rules for all FISs. Then, a program was written in MATLAB software to get the best combination of membership functions to achieve the best result. The program tested 24 kinds of combined membership functions for medial and side fuzzy sets of input variables. The result was the best when the relationship between obtained index and crop yield had the highest value of the correlation coefficient (R2), minimum value of mean square error (MSE) and mean absolute error (MAE). So the fuzzy-GA model will produce the social and technical-mechanization indices while the fuzzy rules of model have been optimized and the best combination of membership functions has been selected. Results and Discussion: The coefficients of determination were obtained 0.11 for fuzzy social model and 0.51 for technical-mechanization model before optimization of fuzzy rules. The error of fitness function decreased with rising generation numbers of GA until the best answer was obtained. After optimization of fuzzy rules by genetic algorithm, these values increased to 0.50 and 0.71 for the fuzzy social and technical-mechanization models, respectively. This result showed that optimizing the fuzzy rules had a significant impact on results of models. After implementation of the written program, to select the best type of membership functions for fuzzy input variables, coefficient of determination varied from 0.14 to 0.51 and 0.1 to 0.73 for the fuzzy social and technical- mechanization models, respectively. This result showed that the effect of social factors on wheat yield was less than technical-mechanization factors and yield can be predicted by technical-mechanization factors with more accuracy than social factors. In the social model for input of experience, the lowest MSE and the highest R2 belong to a FIS with three fuzzy sets and S, Π and Z-shaped membership functions for the right, medial and left fuzzy sets, respectively. In the technical model for input of road availability, the lowest MSE and the highest R2 belong to a FIS with three fuzzy sets and s, trapezoid and z- shaped membership functions for the right, medial and left fuzzy sets, respectively. These results showed that the type of membership functions for fuzzy sets had considerable importance for the accuracy of the model. Conclusion: It can be concluded that the accuracy of the fuzzy model with optimized rules by GA and the best type of membership function for fuzzy sets are considerable. Effect of technical-mechanization factors on wheat yield was more than social factors. This result also showed the strength of fuzzy–GA method in modeling of such issues.
Soil Genesis and Classification
R. Taghizadeh-Mehrjardi; F. Sarmadian; M. Omid; N. Toomanian; M.J. Rousta; M.H. Rahimian
Volume 37, Issue 2 , March 2015, , Pages 101-115
Abstract
In recent years, there has been a great development in the digital soil mapping which has led to production of maps for countries and the continents. Although many studies have been conducted all over the world, few Iranian soil scientists have shown interests in digital mapping. Therefore, in the present ...
Read More
In recent years, there has been a great development in the digital soil mapping which has led to production of maps for countries and the continents. Although many studies have been conducted all over the world, few Iranian soil scientists have shown interests in digital mapping. Therefore, in the present research, different data mining techniques (i.e. regression logistic, artificial neural network, genetic algorithm, decision tree and discriminant analysis) were applied to spatial prediction of great group soils in the area covering of 72000 ha in Ardakan. In this area, by using the conditioned Latin hypercube sampling method, location of 187 soil profiles was selected, which was then described, sampled, analyzed and allocated in taxonomic classes according to soil taxonomy of America. Auxiliary data used in this study to represent predictive soil forming factors were terrain attributes, Landsat 7 ETM+ data and a geomorphologic surfaces map. Results showed that decision tree model had the highest accuracy while it could increase the accuracy of prediction up to 44% in comparison with discriminant analysis technique. Results also indicated using the taxonomic distances led to improving the overall accuracy of decision tree up to 3%. Results confirmed capability of decision tree, artificial neural networks, genetic algorithm, logistic regression, and discriminant analysis with 70%, 65%, 65%, 55%, and 47% accuracy, respectively. Moreover, results showed that decision tree model could predict soil classes in sub-great group with the overall accuracy of 84.2%.