Study On The Fraction Cover Extraction Of Land Cover Based On Spectral Indices In Karst Environment

Karst rocky desertification is similar to the special landscape of desertification, the fractional cover of exposed rock and vegetation are the main performance characteristics of the ground and also key indicators of desertification to access desertification. Land degradation is a dynamic process which is a comprehensive reflection of various surface coverage elements included vegetation, bedrock, soil and other land cover elements. The phenological characteristics of vegetation, deciduous vegetation and its litter of non-photosynthetic capacity of vegetation cover, soil cover, etc. They also has important characterizing rocky desertification information role. Therefore, the evaluation of desertification requires taking into account the variety of surface feature coverage. The fractional cover of vegetation can be estimated approximately from remote sensing images through vegetation indices. Previous researches have proposed a color index, form index and intensity index for mapping land degradation However; the vegetation indices cannot be easily applicable to all land cover types. However, there is no direct indication of spectral indices to the development or degree of information of rocky desertification.Spectral research is the foundation of remote sensing theory research and applied research. In this study, field spectral reflectance measurements were used to establish spectral library of vegetation and other typical land cover in karst regions. Fractional cover of photosynthetic vegetation (PV), non-photosynthetic vegetation (NPV), bare soil (Soil), and exposed bedrock (Rock) were selected as the main symptoms of the ground desertification and key ecological evaluation indicators; the spectral characteristics of the vegetation and other typical land cover are mainly analyzed. According to spectrum curve itself, combined with in situ spectral reflectance, spectral features difference among land cover (PV, NPV, Soil, Rock) in karst are revealed. The spectral reflectance of PV shows characteristic absorption features in the VNIR, which can be recognized with other land cover. As for bare soil and exposed bedrock, the short wavelength infrared (SWIR,2000-2400 nm) spectral range allows to differentiate between bare soil and exposed bedrock. The short wavelength infrared (SWIR, 2000-2400) was the best option for characterization of PV, NPV, Bare soil and exposed bedrock. By analyzing the unique spectral characteristics of PV, NPV, bare soil, and exposed bedrock new spectral indices (karst rocky desertification spectral indices,KSI) are developed to estimating the fractional cover of surface symptoms of karst rocky desertification; the relationship between spectral features of main land cover types and their responding fractional cover is explored. Compare with linear spectral unmixing(LUS), The results showed that proposed spectral indices (KSI) much higher correlation coefficients with factional cover than does spectral reflectance. Vegetation indices(eg.NDVI) have good linear relation with the fractional cover of photosynthetic vegetation.The proposed spectral indices have higher correlation coefficients with fractional cover of non-photosynthetic vegetation (NPV) and bare soil(Soil),while those KSI have lower correlation coefficient with the factional cover of exposed bedrock. With remote sensing technology, especially the development of hyperspectral remote sensing technology, spectral indices are explored preliminary for information extraction and ecological monitoring in karst regions.