Optimal and robust vegetation mapping in complex environments using multiple satellite imagery: Application to mangroves in Southeast Asia

• DOI: 10.1016/j.jag.2021.102320
• Publication Date: 2021-03-13T23:02:54Z
• Authors: Han Xiao; Fenzhen Su; Dongjie Fu; Vincent Lyne; Gaohuan Liu; Tingting Pan; Jiakun Teng
• Keywords: Land Cover

A band selection model was described for efficient and accurate remotely-sensed vegetation mapping in cloudy mixed-vegetation areas, demonstrated with an application on mangroves Southeast Asia (SE Asia). We show how to use multi-source satellite imagery Cloud Computing Platforms improve computational efficiency complex environments. key element of the method relies upon field surveys establish a detailed sample database that includes easily-confused land cover. The Maximal Separability Information (MSI) developed select bands target cover classification from multiple based two principles: 1. maximize separability other cover; 2. prioritize information combinations. Application MSI map SE using three optical SAR data systems (Landsat OLI, Sentinel-2 Sentinel-1) showed: is better at classifying mangrove than Landsat Sentinel-1; SWIR, NIR Red (with SWIR particular) are effective separating vegetation. MSI-mapped showed lower computation cost compared all individual satellites, higher accuracy (above 90%) when applied Asia. It robust tolerating smaller sizes, thereby demonstrating feasibility substantial improvements large-scale