The Retrieval Of Soil Organic Matter Content Based On ASTER Images

Soil is one of the important environmental factors of the terrestrial ecosystems and is the natural resources for the subsistence of the mankind. Since the mid-20th century, because of the increasingly prominent conflict between population, resources and environment around the world, land is becoming the centralized manifestation for a variety of resources and environmental issues. At present, understanding the status of land use and monitoring the dynamic changes of soil components attract many countries attentions.Northeast Plain with rich soil, rivers, lakes and vegetation and other natural resources, is an important marketable grain base in our country. However, in recent years, owing to the influence of human life and natural factors, this area's ecological environment is degenerating, leading to a series of land degradation problems such as soil erosion, salinization and desertification. Therefore, dynamic monitoring the soil physicochemical properties of the Northeast region can provide scientific and reasonable information and data for the government and agricultural production to ensure the sustainable development of ecological agriculture.The traditional routine measurement of soil component content by chemical analysis method with long time, high cost, high precision and sparse measuring point, can not meet the demand of precision agriculture on the spatial-temporal variation conditions of soil components. So how we can obtain the required soil information in a short time is particularly important. Since the 1960s, with the emergence and development of remote sensing technology, the scholars of all nations have introduced hyper spectral remote sensing technology into land-quality monitoring. It opens up a new frontier of Soil Science.Soil spectral reflectance characteristics are concentrated reflection of soil physicochemical characteristics and the physical basis of the soil remote sensing. The soil spectrum study is from the qualitative analysis of spectrum curves in the early days to quantitative inversion of soil component content .Now accepted impact factors of the soil spectral features are water content, organic matter content, iron oxides content, mechanical composition and the parent material etc.Soil organic matter (SOM) which is the main source of plant essential nutrients is the important source of various nutrients in soil and an important indicator of soil fertility. SOM plays a prominent role on promoting the formation of soil structure, improving soil physical properties, and increasing the ability of maintaining soil fertility. Due to the special spectral response characteristic of the soil organic matter from visible light to near-infrared area, the Visible and near-IR wave bands have become the most important bands for organic matter classification. The unique spectral characteristics of SOM inaugurate a new road for the determination of soil fertility rapidly.Author established the organic matter content and soil reflectance equation by studying the relationship between soil reflectance spectral characteristics of ASTER remote sensing images and SOM content and tried to inverse SOM content within the study area by the ASTER images.ASTER images were processed with the conversion of DN value into reflectance. According to geography information of samples transformed projection, the reflectance spectra of samples were accessed directly from remote sensing.According to Visible and near-IR wave, Mid IR and thermal infrared bands distribution of ASTER remote sensing images, the spectrums of the sampling points were acquired from the RS images directly. This not only required the sampling points'geographic information, but also needed convert the original RS image DN value into reflectance. At the same time, the surface temperature was obtained through the TES iterative algorithm. Then according to the different temperature of various surface features, the soil interference information was removed, which prepared well for the follow-up soil classification.And then in this paper, author analyzed spectral reflectance curves characteristic on the actual spectra of the study area's sampling points, and maked continual removal of the spectrums of sampling points to acquire absorption depth. The relationship between soil spectral signatures (absorption feature) and diagnostic bands of organic matter were analyzed in the ASTER images. By comparison, identifying characteristics absorption bands of soil organic matter were the second and third bands.Based on the analysis results above, in the multiple regression analysis, taking the organic matter content of sample points as dependent variable, while taking the spectral reflectance value and its mathematics formation such as reciprocal, logarithm, first-order differentiation as independent variables, which were used to do correlation analysis and establish the multiple regression analysis. It was found that organic matter content had high correlation with the second, third, eighth and ninth band. Based on the stability problem of a model, forecast capacity and the number of selected parameters, we could choose the optimal regression equation which was OM=4.921-1.581b3+8.418b8-32.731b9+8.117b3′for the inversion of organic matter content.With the optimal model, we classified soil organic matter content in the study area as five categories. Through observing the sort results, we found soil organic matter content was high around phaeozem and chernozem. region The organic matter content was characterized by 2% to 3% within the research scope. This conformed to the statistical result of the organic matter contents around Changchun.In this paper, doing multiple analysis between the spectra of samples and organic matter content,we tried to assess soil organic matter content in survey region by quantitative inversion and make the spatial distribution of organic matter content displayed intuitively through remote sensing images.