Dominant wetland vegetation species discrimination and quantification using in situ hyperspectral data

Abstract

Wetland vegetation is an important bio-indicator of wetland physical or chemical degradation. Wetland vegetation continually faces threats from natural disturbance and unsustainable human activities. Traditionally, routine field observations are used to monitor wetlands. However, these methods are expensive and require a lot of human resources, as destructive sampling is required. Remote sensing offers non-destructive and real-time information useful for sustainable and effective management of wetland vegetation. The aim of this study was to explore the potential of hyperspectral remote sensing for wetland vegetation discrimination at the species level. In particular, the study focuses on enhancing or improving class separability among wetland vegetation species using hyperspectral data. In situ hyperspectral measurements were conducted on four dominant grass species in a wetland in Harare. One-way analysis of variance demonstrated significant statistical differences between hydrophytic vegetation species. Vegetation indices performed better compared to red-edge algorithms at discriminating wetland vegetation, with overall accuracy of 82% and 60%, respectively.