Approach To Extracting Farmland Soil Organic Matter Content Based On Hyperion Data

The soil is one of the elements of the geographical environment,which exchangesmatter and energy with the surrounding environment. One part of the matter andenergy that enter the interior of the soil convert into energy to maintain soil itselfactivities,the other register as soil fertility characteristics.The soil fertility is an ability of soil which provides and coordinates nutrition andenvironment conditions for plant growth.The four prime factors affect soil fertility arewater, fertilizer, gas and heat, the specific indexes including the soil texture,structure,the moisture content, the permeability, the organic matter, PH and so on. AlthoughSoil organic matter(SOM) content shares little of soil, but it plays an important role inthe environmental protection and sustainable development of agriculture.This paper combined the content of SOM with the remote sensing image,andtried to inverse SOM content within the study area by the Hyperion image.Thismethod gived a reference for the dynamic monitoring of land quality and opened anew way for the rapid determination of soil fertility.the following content is tointroduce the main achieved in the paper:(1)The preprocessing method of Hyperion images was introduced in detail.Andthe DN value of Hyperion images were converted into reflectance;(2)The paper used the continuum removal method and wavelet pocket analysis todeal with the spectral data of simple points,and introduced the wavelet pocket analysisinto the processing of enhanced feature characteristic spectrum.(3)The correlation analysis between the SOM content and the original reflectancedata, the continuum removal reflectance data and the high frequence and lowfrequence from wavelet pocket analysis shows:First,the maximum correlationcoefficient of the original reflectance was-0.893,they had negative correlation.Second,the continuum removal reflectance had the maximum positive correlation coefficientat band153,the value was0.59,the maximum negative correlation coefficient at band54,the value was-0.793.Then,the high frequence of the original reflectance,maximum positive correlation coefficient0.814,at band35;on the contrary,-0.776,at51band,the maximum negative correlation coefficient of low frequence of the original reflectance was-0.897.Finally, the high frequence of the continuum removalreflectance got the maximum negative correlation coefficient at band39,the value was-0.761.(4)According to the correlation analysis results,this paper used the originalspectrum,continuum removal spectrum and wavelet pocket analysis spectrum as theindependent variable,the SOM concent as dependent variable to do regressionanalysis.The regression models showed that continuum removal spectrum was betterthan the original reflectance,and wavelet pocket analysis was the best.This resultstelled that the wavelet pocket analysis could improve the accuracy of the inversion ofthe SOM’s high spectral characteristics.Compared with the results of previous studies,we found that the correlationcoefficient were smaller,the reasons are as follows:(1)The acquisition time of remote sensing images and the soil samples got in thefield were different.The impurity in the soil might affect the determination of SOMand extraction of simples’reflectance.(2)The soil samples were tested about the organic content after dried,crushed,sueved impurity,but the reflectance extracted from remote sensing image ison the exposed soil.So the differences of soil texture influenced the relationshipbetween the SOM content and reflectance data.