Evaluation of submerged mangrove recognition index using multi-tidal remote sensing data

摘要

For effective mangrove forest mapping, it is valuable to develop vegetation indices from remote-sensing imagery that can characterize the unique characteristics of mangrove forests and differentiate them from other land cover types (especially other vegetation types). In addition to diverse range of vegetation indices derived from single-phase, remote-sensing imagery that has been applied to mangrove forest classifications, recently a submerged mangrove recognition index (SMRI for short) that considers multi-tidal, high-resolution, remote-sensing imagery, and which is based on the differential spectral signature of mangrove forests under high and low tides, was proposed for use in mangrove forest classifications (Xia et al., 2018). However, to date SMRI has not been compared with existing vegetation indices that are often applied in mangrove forest classifications based on remote-sensing imagery in detail. In this study, the SMRI values obtained from medium- and high-resolution images (i.e., Landsat 8 OIL/TIRS and GF-1 respectively) are compared with four vegetation indices widely used in mangrove forest classifications (i.e., the normalized difference vegetation index, ratio vegetation index, enhanced vegetation index, and soil adjusted vegetation index). One more vegetation index, which was only available for remote-sensing imagery with visible bands, a short-wave infrared band, and a mid-wave infrared band, i.e., Landsat 8 OIL/TIRS images, was also compared with the SMRI obtained from the medium-resolution images. The results from experiments with medium- and high-resolution images of Yulin City, Guangxi Zhuang Autonomous Region of China show that the SMRI can distinguish submerged mangrove forests more effectively than the compared vegetation indices, especially in areas between high- and low-tide levels. Furthermore, the SMRI results obtained from high-resolution images perform better than those obtained from medium-resolution images.