Construction of 3D landscape indexes based on oblique photogrammetry and its application for islands

Abstract

Landscape indexes are quantitative indicators that reflect the composition and spatial configuration of landscape structures. However, the current two-dimensional (2D) spatial analysis methods lack accuracy in assessing patch characteristics due to the lack of three-dimensional (3D) information. Following the “Patch-Class-Landscape” framework, eight 3D landscape indexes were built to quantitatively describe spatial landscape features and two sensitivity indexes were developed to identify the differences between 2D and 3D structures. Based on two types of oblique photogrammetry data, validation and comparison studies were conducted for Tianheng Island and Sichang Island. The results found the following: (1) At the patch level, the 3D shape index (TPSI) of vegetation was generally higher than that of buildings, with an R2 of 0.634, and the classification index (TCI) showed remarkable performance in identifying patch type. The patch type was likely to be building or vegetation when TCI approached 33, respectively, with a classification accuracy of 90% after verification. (2) At the class level, the 3D percentage of landscape (TPLAND) of grassland and arbor types on the two islands were quite different, reflecting significant differences in the dimensionality of the vegetation landscapes, as influenced by different climatic zones. Moreover, the 3D landscape shape index (TLSI) and other shape-related indexes had higher exponential sensitivity coefficient (ESC) values, due to the higher amount of 3D shape information they carry. (3) At the landscape level, the two 3D Shannon indexes (TSHDI and TSHEI) did not significantly change compared with their 2D counterparts, implying that these two indexes, as larger-scale landscape indicators, had lower sensitivity when extra-dimensional information was added. Overall, the 3D landscape indexes can better present 3D information at different landscape levels. As a potential and effective assessment tool and it will be applied to improve existing spatial planning and landscape management.