Hyperspectral Prediction Model of Soil Organic Matter in Black Soil Region Based on Fuzzy Recognition

Abstract

Land is an important basic resource that must be relied upon, irreplaceable and non-renewable for human survival and development. The limited quantity and the scarcity of supply determine that we must make full use of land resources. The use of hyperspectral remote sensing to establish accurate soil organic matter prediction models is an inevitable requirement for rapid determination of soil organic matter, precision agriculture, and soil carbon pool estimation. Accurate understanding of soil distribution information is an important basis for soil mapping and establishment of soil databases, and has important guiding significance for land management and land use planning in China. Based on the above background, the purpose of this paper is to study the hyperspectral prediction model of soil organic matter in black soil region based on fuzzy recognition, and to establish a high-spectral estimation model of soil organic matter by linear regression and fuzzy recognition method. In this paper, the fuzzy closeness inversion model of soil organic matter content is established, and compared with the results of the linear regression model; the accuracy of the linear model is basically consistent and low. The accuracy of the fuzzy closeness inversion model is higher, and the lowest is also It reached 89.71%, and the rest were more than 90%. In addition, based on the diversity of spectral inversion factors and the dynamics of organic matter content, an interval-valued fuzzy inversion model was proposed. The prediction result is accurate for 5 samples. There is an error, and the sample deviation with error is 3.22%. This shows that the fuzzy recognition method can effectively solve the problems of nonlinearity and ambiguity, and it is feasible to use the fuzzy recognition method for soil organic matter content prediction.