Multitemporal classification of TerraSAR-X data for wetland vegetation mapping

摘要: This paper is concerned with wetland vegetation mapping using multitemporal syn- thetic aperture radar imagery. Although wetlands play a key role in controlling flooding and nonpoint source pollution, sequestering carbon providing an abundance of ecological ser- vices, knowledge the flora fauna these environments patchy, understanding their functioning still insufficient for reliable functional assessment on areas larger than few hectares. The aim this to evaluate TerraSAR-X imagery precisely map distribution formations considering flood duration. A series six dual-polarization images (HH-VV) was acquired 2012 during dry wet seasons. One polarimetric parameter, Shannon entropy (SE), two intensity parameters (σ° HH σ° VV), which vary status roughness, were first extracted. These then classified support vector machine techniques based specific kernel adapted comparison time-series data, K-nearest neighbors, decision tree (DT) algorithms. results show that can be identified very accurately (kappa index ¼ 0.85) from classification SE temporal profiles derived images. They also reveal importance use param- eters instead backscattering coefficients alone (HH or VV) combined VV). © 2014 Society Photo-Optical Instrumentation Engineers (SPIE) (DOI: 10.1117/1.JRS.8.083648)