Evaluation of the USGS national elevation dataset and the Kansas Biological Survey's FLDPLN ('floodplain') model for inundation extent estimation.

摘要

While riverine flooding is a natural and ecologically beneficial phenomenon, severe flood events continue to result in loss of life and property. Recent major flood events in the Midwest, including record flooding in Kansas in 2007 and Iowa in 2008, have shown that currently available inundation extent information is insufficient from both a planning and a response perspective. Current initiatives at the Kansas Biological Survey are aiming to bridge the gap between limited information that is presently available and what is needed to adequately prepare for and respond to a range of inevitable, and in some cases unprecedented, flood events. The focus of this effort is the development of a library of modeled flood inundation extents, using the FLDPLN model, for major streams across the state that can be accessed in near real-time to provide valuable information to disaster responders. This research (1) examines the USGS National Elevation Dataset (NED) and evaluates the affects of errors in the elevation data on flood inundation extent estimation; and (2) evaluates the capabilities and limitations of the FLDPLN model for inundation extent estimation. Results showed that, although the accuracy of pre-LiDAR NED for the Kansas study area is better than published figures, modeled flood extents vary significantly when using LiDAR-derived vs. pre-LiDAR NED elevation data inputs. Comparison of modeled flood extents for HEC-RAS, HAZUS, and FLDPLN models for both hypothetical and empirical floods events showed greater correspondence at high flood stages. Improved elevation data and empirical low flood data would offer improved flood extent estimates and more robust model evaluation.