Determining wetland spatial extent and seasonal variations of the inundated area using multispectral remote sensing

Abstract

Wetlands can only be well managed if their spatial location and extent are accurately documented, which presents a problem as wetland type and morphology are highly variable. Current efforts to delineate wetland extent are varied, resulting in a host of inconsistent and incomparable inventories. This study, done in the Witbank Dam Catchment in Mpumalanga Province of South Africa, explores a remote-sensing technique to delineate wetland extent and assesses the seasonal variations of the inundated area. The objective was to monitor the spatio-temporal changes of wetlands over time through remote sensing and GIS for effective wetland management. Multispectral satellite images, together with a digital elevation model (DEM), were used to delineate wetland extent. The seasonal variations of the inundated area were assessed through an analysis of monthly water indices derived from the normalised difference water index (NDWI). Landsat images and DEM were used to delineate wetland extent and MODIS images were used to assess seasonal variation of the inundated area. A time-series trend analysis on the delineated wetlands shows a declining tendency from 2000 to 2015, which could worsen in the coming few years if no remedial action is taken. Wetland area declined by 19% in the study area over the period under review. An analysis of NDWI indices on the wetland area showed that wetland inundated area is highly variable, exhibiting an increasing variability over time. An overlay of wetland area on cultivated land showed that 21% of the wetland area is subjected to cultivation which is a major contributing factor to wetland degradation.