Mapping soil organic matter based on land degradation spectral response units using Hyperion images

Abstract Soil degradation is the core of land degradation. The lost of organic material is a signature of the quality degradation and productivity decline of soils. Therefore, it is of importance to develop a relatively rapid and economic method for mapping soil organic matter (SOM) at regional scale using remote sensing images. The study was to investigate the upscaling method of estimation model for mapping SOM using hyperspectral remote sensing images and multi-source data through segmentation of land degradation spectral response units (DSRUs). The estimation model of SOM based on DSRUs was developed to map SOM using Hyperion images of Hengshan County in the northern Shaanxi Province of China. Results showed that the DSRU estimation model, based on the relationship between the SOM content and features of spectral and regional variables of DSRUs, was valid to estimate the content of SOM. The determinative coefficient ( R 2 ) of the model increased from 0.562 at the scale level of 25, to 0.721 at the scale level of 50, and to 0.722 at the scale level of 100. The environmental variables of elevation, slope and vegetation index (MCARI) were included in the DSRU models, making it advantageous over the in situ models. The mapping content of SOM based on DSRU estimation model using Hyperion image agreed with that by the field survey and by the Kridge interpolation. Our research concluded that the DSRU model provides a relatively rapid, cost-effective, and accurate way to map the content of SOM for the land management of large area.