Building spectral catalogue for salt marsh vegetation, hyperspectral and multispectral remote sensing

Abstract

Salt marshes are coastal wetlands dominated by the grasslands and they are one among the most productive coastal ecosystems which provide several benefits and multiple ecosystem services. Their spatial distribution is at risk due to wide range of human activities and sea level rise. Remote sensing technique can be applied to assess their spatial coverage in quick time. Field spectrometry, an accessory tool of remote sensing, measures the reflectance using spectroradiometer to acquire continuous spectrum from a target. Developing spectral catalogue for salt marsh vegetation is essential to discriminate the vegetation into species level. In the present study, species canopy spectral signatures for seven salt marsh species were measured using spectroradiometer. Several post-processing steps were conducted for noise removal and the spectral catalogue was developed. In addition, Landuse/Landcover (LULC) spectral signatures for salt marsh vegetation were acquired from Multispectral data ofSentinel-2B satellite in near real-time using field ground truth coordinates. The species canopy and LULC spectral signatures were applied in the MSI data to perform species level classification of salt marsh vegetation. The results were significant, and indicated an overall accuracy of 65.8%, and 73.55% for species canopy and LULC spectral signatures respectively. The spectral outcome of this study can be utilized directly for rapid classification and to assess the change in the dynamics of salt marsh ecosystems for the areas with low accessibility. Further the mapping results can be used as baseline information for change detection analysis which would support coastal resources management, and help in strengthening the regulatory framework for protection and conservation purposes.