The spatio-temporal variation of the 2014 Tokwe-Mukosi floods: a GIS and remote sensing based approach

Abstract

Floods are natural hazards that have always created an unexpected threat to human life and property. When severe floods occur in areas occupied by humans, they can create natural disasters that involve the loss of human life, loss of property and may result in serious disruption to the ongoing activities of urban and rural communities. The unanticipated and incessant heavy rains in 2014 that pounded theTokwe-Mukosidam catchment from late January and March led to flooding of the dam basin affecting villagers in the area. The affected villages were mainly those that had not headed to the call by government to relocate to identified land. This created an emergency situation due to increase in volume of water in the dam reservoir. The increase in volume contributed to flooding of the dam basin, marooning the houses and fields in the upstream and destroying infrastructure. Earth observation techniques and GIS can contribute towards more accurate flood hazard mapping that can be used to assess damage to residential properties, infrastructure and agricultural crops as well as provide information for emergency relief work. The objective of this study was to use remote sensing and Geographic Information Systems to determine the spatial variation of flood hazard in the Tokwe-Mukosi Dam area during and after the 2014 flooding using the (recent launched 2013) 30m Landsat 8 OLI based images via GloVis website (http://glovis.usgs.gov/).A segment map digitizing technique in ILWIS was used to separate the water and non-water areas. Areas occupied by water in the different dates were also determined through histogram analysis. The digitized segment maps was exported as a shapefile in ArcGIS and then over-layed with other important data such as soil type and land cover maps. The land surface characteristics (such as elevation, land cover, and distance from stream network, rivers and road network) were also related with the presence and absence data (flood condition) using a spatial logistic regression in SPSS v17.0. Results indicate that the water inundated area increased by more than 40% from January 2013 (just before the floods) to February 2014 (During the floods) and April 2014 (after the floods). Factors such as elevation, distance from river network had a signification relationship (p<0.05) with flood hazard. Thus the use of remote sensing and statistical techniques can be a valuable tool for flood plain zoning and disaster preparedness for the settlements in the Tokwe-Mukosi Dam area. It can also help to prevent undesirable side effects of the developments and can assist in implementing effective mitigation measures.