Study On The Inversion Of Soil Organic Matter Content Based On Hyper-spectrum

The content of soil organic matter is an important index to evaluate soil fertility, and the spatialdistribution of soil organic matter can be used to study soil carbon and nitrogen cycling, soil qualityevaluation and the grain yield.The traditional method to obtain the soil organic matter content is usually achieved by collectingsoil sample data in the field and measuring that with chemical method indoor. Although this method hasthe characteristics of high accuracy, it needs heavy workload, high cost and long cycle time, whichmakes it hard to obtain the soil organic matter content area and spatial distribution of soil organic matterin a relatively short period. Because of its low spectral resolution, optical remote sensing technologiesare still not good for soil organic matter content in parcel level or regional level. While with highspectral resolution, hyper-spectral remote sensing can be used to examine the slight variation of soilorganic matter content accurately and promptly, and meet the requirements of the development ofmodern precision agriculture.This study takes soil in northeast China as the research object.First of all, under the condition ofindoor, using the ASD Field Spec4spectrometer measure soil samples’ spectral, set up soil spectralvariables relation with soil organic matter content of the multiple regression model; Then, in thecommon sensitive bands, the Hyperion hyperspectral image is used, and through the band calculationmodule of ENVI software using the programming, the prediction model is applied to the Hyperionimage to obtain the organic matter content distribution map of study area. Conclusions of the study areas follows:1） Hyperspectral derivative technology can obtain nanometer soil spectral reflectance andaccomplish the mathematical transformation of the spectral reflectance. It also can be used to expandspectral characteristic differences between the samples and increase the subtle differences of the soilattribute information. Apart from that, this method also can be applied to the soil surface (0-20cm)organic matter content monitoring.2）In the experiment of processing the data of characteristic bands for492nm,663nm,1221nm,1317nm,1835nm and2130nm, the coefficient R2is up to0.909using the independent variableregression model established with the first-order differential reflectivity to inverse the soil organicmatter content. By comparing the predicted and measured values, we can conclude that the RMSE is theleast, which illustrate that this method can be used to predict the soil organic matter content.3）In Band14, Band31, Band108, Band117, Band198band, we can obtain image-sensitive bandreflectivity order differential with the pre-treated Hyperion hyperspectral image processed using theprogram. Using the computing modules of the ENVI software, mapping of soil organic matter in theregion is accomplished, the precision is up to0.737.