HYDRODYNAMICS FLOOD MODELLING IN BAROTSE FLOODPALIN, ZAMBIA: EFFECTIVENESS OF DIGITAL ELEVATION MODELS

Abstract

The extensive Barotse floodplain in western Zambia is among the most flood prone area. This study focused on hydrodynamic flood modeling to simulate flood inundation characteristics that will benefit provision of recession farming, canal management, and land use planning in the Barotse floodplain. For this study the one-dimensional hydrodynamic model (HEC-RAS 4.1.), HEC-GeoRAS10 (GIS extension), ARC-GIS 10, and Google Earth are used. The SRTM 30- m, SRTM 90-m, and ASTER 30-m DEMs are first compared to select a better DEM for model simulation. This was done by a visual comparison and through objective functions using SRTM 30-m as reference. The mean elevation deviation of SRTM 90-m ranges from 0 to 1m and ASTER 30-m ranges from -9 to 3m. SRTM 30-m better captured the channel geometry and the heterogeneity of the floodplain to a certain details. ASTER 30-m as resulted in a somewhat exaggerated elevation of river channel banks. Additionally taken the topographic survey of Mouyowamo canal elevation as reference and the result obtained that RMSE of SRTM 30m and SRTM 90m is 2.99 and ASTER 30m is 7.28. The floodplain terrain slope shows the largest variation in ASTER 30-m and the least variation for the course DEM SRTM 90-m. None of the DEM captured the canals in the study area. Topographic survey and Google Earth provided important information to correct the canal cross-sections. The results show that all the three DEM sources have major limitation for the study area. SRTM 30-m DEM was used for modeling flood inundation as it has better accuracy. The inundation depth, velocity and extent map were developed using HEC-RAS flood model, HEC-GeoRAS and ARC-GIS were integrated to preprocess HEC-RAS inputs and post process the model outputs. The simulated inundated extent of Mongu district (10,870km 2 ) was 6.07%, 6.39% and 6.59% for 5, 10 and 20 year return periods respectively. Also, the results showed that the Lealui villages (targeted area) are not inundated with these return period floods. This study was a first attempt to model the flood dynamics of the complex Brotse floodplain. To be more conclusive to use the HEC-HMS model, the inputs of particular floodplain and channel geometry of selected areas should be surveyed in the future.