Amelia Martin, a CEIE rising senior (second from left in photo); Daniel Scolese, a May 2013 BS graduate (second from right in photo); and Michael Wright, a CEIE PhD student, (not pictured) presented their research in poster sessions at the American Geophysical Union Science Policy Conference in Washington DC on June 26. The conference features work on the interface of science and policy. Professors Burak Tanyu and Celso Ferreira (the casual dressing professors) look on proudly as the advisors of these students; Professor Mark Houck, also an advisor, is not pictured.
Amelia’s project was entitled Potential Impacts of Hurricane Flooding in the National Capital Region: What if Hurricane Sandy Made Landfall in the Chesapeake Bay? Her undergraduate research project evaluated hurricane surge flood risks in the National Capital Region combining a state-of-the-art hydrodynamic and wave model (SWAN+ADCIRC) to simulate hurricane flooding; an asymmetric wind model based on southerly shifted tracks of hurricane sandy (National Hurricane Center Best Track Data); and the HAZUS-MH Flood Model (developed by the Federal Emergency Management Agency [FEMA]) to gain a better understanding of possible damages incurred from hurricane storm surges in the Washington DC National Capital Region.
Daniel’s research focused on the comparison of the actual and predicted flood-damage using a computer model to evaluate the effectiveness of models to predict or assess damage before or after storm events. His poster, entitled Evaluation of the HAZUS-MH Model to Predict Flood Damage Along the New Jersey Coast After Hurricane Sandy, presented the results of a HAZUS sensitivity analyses concerning the selection of depth-damage curves, first floor elevation and building inventory distribution; the results of the HAZUS level 1 and level 2 analyses and a comparison of the calculated expected damage to the FEMA post-storm damage survey and the authors field campaigns. These results help quantify uncertainty in hurricane flood damage estimation and support policy and decision-making in the future.
Michael’s research, Risk Quantification for Extreme Precipitation Events: Developing a Parallel Algorithm, is part of his PhD dissertation. In this research, an automated algorithm for the identification of low-error regions based on sampling from enumerated regions using a distance- and a frequency-based statistic was tested on subsets of Minnesota’s high-density gauge network. The ‘embarrassing parallelism’ of Hosking and Wallis’ Monte Carlo calculations is exploited using parallel computing to generate diagnostic statistics for the enumerated list of all possible regions in each subset, and to estimate the root mean square error of quantile estimates for different combinations of regions. Results across a range of parameterizations were presented for the algorithm, indicating its ability to locate low-error estimations automatically.
Further research descriptions of this work are available at http://spc.agu.org/2013/