The Assessment, Improvement, and Application of the GeoFlood Flood Inundation Mapping Framework PDF Download

Author: Alec Carruthers
Publisher:
ISBN:
Category :
Languages : en
Pages : 198

Get Book Here

Book Description
GeoFlood is a flood inundation mapping package that utilizes a geodesic, cost minimization algorithm to extract channel networks from high resolution terrain data along with National Water Model forecasts and a Height Above Nearest Drainage approach to create near real time flood inundation maps. Earlier work has applied the GeoFlood framework at the scale of a single Hydrologic Unit Code 12 watershed, but this work extended the application to seven counties across the state of Texas. One meter resolution lidar digital elevation models were generated for each county and segmented by watershed, resulting in approximately 400 gigabytes of input data. Two cost functions were found to improve the channel network extraction capabilities. The unique aspect of the first cost function was the inclusion of a cost threshold, which served to increase the penalty for pixels that had a cost above a given limit, thereby restricting the extracted channel to only the minimum cost path and preventing shortcuts. A second cost function, featuring a binary representation of the National Hydrography Data high resolution flowlines, was used to improve the performance of the threshold based cost function in the presence of artificial features or within low relief topography. Four variations of a channel bed slope calculation were tested, two that were end point based and two that utilized a linear regression. The end point based methods were shown to have synthetic rating curves with a smaller percent error and variance across the first three meters of stage height, as well as less area separation from the corresponding United States Geological Survey synthetic rating curves than the regression based methods. Identification of the reaches in which the slope calculation method was most significant was conducted by analyzing the variance and standard deviation across the four computed slopes. Artificial and canal reaches at a low level (1 - 2) and high stream order (5 - 7) were found to have the most variance across their computed slopes. A reach being hydro-flattened was determined to most likely not be the sole limiting factor when it comes to the accuracy of GeoFlood derived products. An approximately equal number of low and high error synthetic rating curves were produced from hydro-flattened reaches. While improvement to the model can always be made, the application of GeoFlood across seven counties in Texas, using high resolution terrain data, was a step forward in regard to showing that GeoFlood can be applied to larger study areas than just a single watershed, including the potential for statewide and or national implementation