Abstract
Three internationally‐available elevation models were assessed for suitability for use in the prediction of flood inundation. The elevation data were selected because they represent data which may be available in data‐poor developing countries. These were a contour data set, a remotely sensed dataset (interferometric SAR) and a dataset which may be obtained rapidly through a differential global positioning system (DGPS) survey. The latter dataset was known to have much less uncertainty than the contour dataset and was, therefore, accepted as a benchmark against which to test the other two. Each dataset was used to predict flood inundation for an event in the United Kingdom in 1998 using the two‐dimensional model LISFLOOD‐FP. The contour dataset was different in spatial character (overly smooth) to the DGPS dataset and resulted in substantial differences in the timing and extent of flood inundation. The interferometric SAR dataset was also different in spatial character (overly rough) to the DGPS dataset although the differences in the timing and extent of flooding were not as great as for the contour dataset. However, results demonstrate potential problems with the use of satellite remotely sensed topographic data in flood hazard assessment over small areas.