Construction of three-dimensional remote sensing ecological index (TRSEI) based on stereopair images: A case study of Miaodao Archipelago in China

High -intensity human activities make the sustainability of the island ecosystem face unpredictable risks. Aiming at the problem that most ecological evaluation indexes are based on a two-dimensional level, this study constructs a digital surface model based on stereo image pairs and proposes a set of remote sensing ecological indexes for islands based on three-dimensional information. (1) From 2000 to 2016, Miaodao Archipelago's overall ecological quality has been significantly improved. The area with poor ecological quality has decreased by 49 %; (2) Based on TRSEI contribution analysis, NDVI (+) WET (+) NDBBI (-) LST (-) regularity was presented. With height and volume variables, TRSEI response to LST was indirectly improved (Pearson correlation coefficient = -0.844); (3) TRSEI was positively correlated with CONTAG and vegetation area and negatively correlated with SHDI and LSI, indicating that for relatively small island ecosystems, the decrease of vegetation landscape contiguity and the increase of landscape diversity and patch shape complexity would aggravate landscape fragmentation and increase interaction with surrounding patches, resulting in the decline of island ecological quality; (4) Due to the high intensity of human activities and urbanization construction, islands I1 and I2 got lower TRSEI scores, and the areas with poor ecological quality mainly located in harbors and residential areas. In contrast, I3 and I5, with less population and low human activities, have good and the above areas accounting for 67.8 % and 57.7 %, respectively. Compared with RSEI, TRSEI incorporates 3D information and has a more extensive dynamic range, providing a solution for the ecological assessment of islands and coastal regions.