The impact of DEM data source on prediction of flood, sea- level rise and erosion risk

Coveney, S. and Fotheringham, S. (2011) The impact of DEM data source on prediction of flood, sea- level rise and erosion risk. International Journal of Geographical Information Science, 25(7), pp. 1191-1211. (doi: 10.1080/13658816.2010.545064)

Abstract

DEM and DSM data are used widely in GIS for the spatial prediction of sea-level rise, and for the assessment of coastal flooding and erosion risk. The selection of a suitable DEM/DSM typically involves consideration of elevation accuracy, spatial-resolution, data coverage and data availability. Synoptic studies often employ photogrammetric derived DEM data, and DSMs derived from Laser Scanning are now becoming the standard where larger scale base data are required. However, the problems associated with using different types of DEM/DSM for sea-level rise prediction can sometimes be overlooked. DEM/DSM users typically make reference to ± DEM elevation error statistics in order to choose a suitable base model. However, these statistics can fail to give a reliable impression of the spatial distribution of elevation error, and they can also overstate the accuracies that are achievable outside ground truth areas. These issues introduce uncertainty into spatial predictions of flood risk, potentially invalidating prediction results. This study uses external validation and a comparative analysis of sea-level rise prediction performance to highlight important issues associated with the use a range of DEM/DSM types. The elevation accuracies that can be achieved within a high-resolution small-scale photogrammetric DEM, GPS-derived DEMs at two different scales, and a very high-resolution local-scale Terrestrial Laser Scanning DSM are quantified using comprehensive external validation in a large coastal saltmarsh area in southwestern Ireland. The errors within each model are discussed in general terms, and in relation to the suitability of each model for the spatial prediction of sea-level rise risk. The performance of each model for the spatial prediction of sea-level risk is subsequently evaluated, and the potential for each dataset to be used to predict potential saltmarsh loss is also examined. Results and conclusions to be drawn from this analysis are likely to be of interest to anyone considering the potential of a range of DEM sources for the spatial prediction of sea-level rise risk. The magnitude of the errors detected in the photogrammetric DEM are found to be substantially beyond quoted error, demonstrating the degree to which quoted DEM accuracy can understate actual DEM error, and highlighting the risks associated with considering spatial-resolution as an analogue for DEM quality. The superior performance of the much lower resolution GPS-derived DEMs for the spatial prediction of sea-level rise risk confirms this, and demonstrates the practical potential of GPS-derived DEMs for sea-level rise modelling at local to regional scales. The presence of very dense ground vegetation in the test study area highlights the important distinction that needs to be drawn between Laser Scan acquisition accuracy and DSM accuracy. The potential for this problem to affect any laser scanning dataset is briefly discussed and is overcome in this study using the GPS data to define ground level in the Laser Scan data. Conclusions regarding the performance of each dataset for the prediction of habitat loss demonstrate the extent to which high-resolution survey data facilitates more meaningful prediction than can be achieved using high-resolution interpolated data.

• Item Type: Articles
• Status: Published
• Refereed: Yes
• Glasgow Author(s) Enlighten ID: Coveney, Dr Seamus
• Authors: Coveney, S., and Fotheringham, S.
• College/School: College of Science and Engineering > School of Geographical and Earth Sciences > Geography
• Journal Name: International Journal of Geographical Information Science
• ISSN: 1365-8816
• ISSN (Online): 1365-8824