Abstract
Abstract. Severe flood events turned out to be the most devastating catastrophes for Europe's population, economy and environment during the past decades. The total loss caused by the August 2002 flood is estimated to be 10 billion Euros for Germany alone. Due to their capability to present a synoptic view of the spatial extent of floods, remote sensing technology, and especially synthetic aperture radar (SAR) systems, have been successfully applied for flood mapping and monitoring applications. However, the quality and accuracy of the flood masks and derived flood parameters always depends on the scale and the geometric precision of the original data as well as on the classification accuracy of the derived data products. The incorporation of auxiliary information such as elevation data can help to improve the plausibility and reliability of the derived flood masks as well as higher level products. This paper presents methods to improve the matching of flood masks with very high resolution digital elevation models as derived from LiDAR measurements for example. In the following, a cross section approach is presented that allows the dynamic fitting of the position of flood mask profiles according to the underlying terrain information from the DEM. This approach is tested in two study areas, using different input data sets. The first test area is part of the Elbe River (Germany) where flood masks derived from Radarsat-1 and IKONOS during the 2002 flood are used in combination with a LiDAR DEM of 1 m spatial resolution. The other test data set is located on the River Severn (UK) and flood masks derived from the TerraSAR-X satellite and aerial photos acquired during the 2007 flood are used in combination with a LiDAR DEM of 2 m pixel spacing. By means of these two examples the performance of the matching technique and the scaling effects are analysed and discussed. Furthermore, the systematic flood mapping capability of the different imaging systems are examined. It could be shown that the combination of high resolution SAR data and LiDAR DEM allows the derivation of higher level flood parameters such as flood depth estimates, as presented for the Severn area. Finally, the potential and the constraints of the approach are evaluated and discussed.
Highlights
Mapping of large scale flood events is of major concern for disaster response teams and flood management officials and poses a key task to hydrologists and the industry in order to generate reference data and calibration information for dynamic flood models, damage estimates, flood plain mapping tasks and further applications
The pixel size of 12.5 m has a negative influence on the accuracy of the flood mask when compared to a high resolution DEM of 1 m pixel spacing
Precise high resolution synthetic aperture radar (SAR)-data is expected to be more suitable for such detailed studies
Summary
Mapping of large scale flood events is of major concern for disaster response teams and flood management officials and poses a key task to hydrologists and the industry in order to generate reference data and calibration information for dynamic flood models, damage estimates, flood plain mapping tasks and further applications. Voigt: Improved estimation of flood parameters when processing the respective flood masks in GIS operations or when generating flood maps This is the case when flood masks are combined with digital elevation data to derive spatially distributed estimates of inundation depths or to exactly locate the land-water boundary in a digital terrain model (Sanders, 2007; Mason et al, 2007; Ling et al, 2008). In such cases even small geometric inaccuracies during the geocoding process or slight classification errors (local or general in character) can significantly reduce the quality of higher level products such as maps of inundation depths. The methods presented in this paper seek to reduce such residual errors through local matching operations
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
