Evaluation of the different spectral indices to map biocrust using spectral information

Abstract

Biological soil crusts (BSC) are complex communities formed by a close association of soil particles and cyanobacteria, algae, microfungi, lichens or bryophytes that live within or immediately on top of the uppermost millimeters of the soil surface. These communities cover non vegetated areas in most of the arid and semiarid ecosystems, and modify numerous soil surface properties and ecosystem processes. Given the importance of BSC in ecosystem functioning, accurate and spatially explicit information on the distribution of BSC is mandatory. With this objective, considerable effort has been devoted to identify and map BSC using remote sensing data, and some spectral indices have been developed. These indexes use the spectral differences among BSC and other surface components like vegetation or bare soil to identify the areas dominated by BSC. Our main objective is to test the feasibility of the previous indices published in the literature for mapping different types of BSC in a complex study area, where these index have not been developed, at different spatial scales. Our results showed the low capability of indexes based on multiespectral images to identify areas covered by BSCat field and image spatial scales. Hyperspetral indices, on the other hand, showed better results than those obtained with multispectral indices, with an accuracy around 71% because they analyzed specific absorption features related with photosynthetic pigments like chlorophyll and carotenoids. We can conclude that the spatial heterogeneity of the area and the spectral similarities among BSC, green and dry vegetation or bare soil makes it difficult to correctly distinguish BSC in arid and semiarid ecosystems using only multispectral information, whereas hyperspectral images offer an important tool to map different types of BSC and to discriminate among these and other surface components.