Mapping the distribution and abundance of common needs Phragmites australis in the Nylsvley Wetland, South Africa using SPOT and Landsat imagery

Abstract

Extensive research on the biology and ecology of Phragmites australis has been done since the 1960s. This has been carried out to manage and monitor the distribution of P. australis. Phragmites australis is one of the most invasive plants in wetlands that thrive successfully as compared to the native species. Phragmites australis alters hydrology and wildlife habitat, increases fire potential, and shades native species. In this study, the distribution and abundance of P. australis was mapped and analysed. To understand the distribution and abundance of the common reed P. australis invasions, research in a spatial context at several scales is required. In this study, Landsat 4-5 Thematic Mapper (TM), Landsat 8 Operational Land Imager (OLI), and SPOT 6 were used to map the distribution of P. australis in Nylsvley Wetland. Five sampling sites were selected in Nylsvley Wetland and reference data was collected to aid the classification process. Nylsvley Wetland is considered one of the largest floodplain systems in South Africa, with the Nyl River flowing through the central and North eastern parts of the Nylsvley Nature Reserve. The surface area of P. australis will be estimated using Garmin® Etrex 62 Global Positioning System (GPS) within the selected sites. Images from year 2011, 2013, 2015 and 2017 for SPOT and Landsat were selected for further P. australis classification. Supervised classification was used to classify the satellite images into different classes. From the classification images it was observed that P. australis covers a small area of the study site relative to other identified land cover types and were mostly distributed along the river system. However, SPOT images showed an increasimg trend in P. australis cover which was not evident on Landsat images. Accuracy assessment was performed to compare the performance of SPOT and Landsat. The results showed that average overall accuracy was 71.50% and 61.62% for SPOT and Landsat, respectively. Correlation between the classified image is shown by the overall kappa coefficient average of 0.5648 and 0.37 for SPOT and Landsat, respectively.