Research On Evaluation Methods For Soil Fertility

Soil fertility, which is the basis for sustainable development of soil, is the direct supplier of agricultural products. It directly affects the yield and quality of agricultural products. Soil fertility evaluation can provide a theoretical basis for agricultural production, therefore more and more evaluation methods are applied to the soil fertility, such as principal component analysis, grey correlation method, expert analysis, fuzzy comprehensive evaluation and information entropy method. But most studies of soil fertility evaluation focused on using a single method to evaluate soil fertility, lack of comparison and analysis. So this paper study three evaluation methods on soil fertility, and compare the results of these methods, and analyze the reasons for inconsistent evaluation results. On this basis, this paper proposes combination evaluation method application on soil fertility, which combines results of three single evaluation methods to obtain a more accurate result. The main contributions are as follows:(1) Using single evaluation methods to evaluate soil fertilityIn this study, principal component analysis, Fuzzy Comprehensive Evaluation and Grey correlation method are applied to soil fertility evaluation. This study uses the measurement data from Long-term positioning experiments as the data sets and the total size of data set is 211. Principal component analysis is a classic feature extraction algorithm, and it can effectively extract the most major information to get the low dimensional represent of the original complex data. This study uses principal components which are extracted by PCA to provide a comprehensive evaluation of the data set and gets the fertility hierarchies from the evaluation results. Fuzzy Comprehensive Evaluation first use analytic hierarchy process to get the weights of the evaluation factors, and builds the fuzzy comprehensive evaluation matrix by membership function, and then get the evaluation results. Grey correlation method is applied to soil fertility evaluation. It uses the optimal value of the evaluation factors as a reference sequence, and calculates the degree of association between evaluation objects and the reference sequence, and then ranks the results.(2) Analyzing and comparing the difference of the results of three evaluation methods This study analyzed the differences in evaluation results of three reasons. The realization of the principle of PCA were simplified reasoning, arrive at the expression of PCA formula weight, compared with the other two methods, PCA and their differences are mainly due to the different weights. Grey relational analysis and fuzzy comprehensive evaluation of the weights are different, but the normalized handle different, leading to inconsistent results of three methods.(3) Researches on Combination evaluation method based on entropy in the application of soil fertility evaluationIn order to avoid the limitation of single evaluation methods, this study proposes combination evaluation method based on entropy to evaluate those results from the three single evaluation methods. Compared with the expert evaluation, Accuracy rate of the combination evaluation method can achieve 91.47%, which is 7% higher than average accuracy rate of single evaluation methods. Therefore, in the evaluation of soil fertility, the combined evaluation method gets better performance.