Florida Commission on Hurricane Loss Projection Methodology Flood Standards Development Committee October 30, 2014 KatRisk LLC 752 Gilman St. Berkeley, CA 94710 510-984-0056 www.katrisk.com
About KatRisk q Founded in 2012 q Founding members have extensive science and engineering backgrounds including 40+ years of CAT modeling experience q Emphasis on global flood modeling, including tropical cyclone rainfall and storm surge q On-going collaborations with contributing consultants o Tropical cyclone track modeling o Storm surge modeling q Have developed flood hazard layers on an 10 meter resolution for the entire US. Development of an event based probabilistic model starting later this year and into next year. 2
Some Key Considerations in Flood Modeling q Completeness o Minimum size of catchment modeled o 1-d vs 2-d everywhere o Consideration of fluvial and pluvial flooding q Drivers of flooding that need to be considered o Tropical cyclone and non-tropical cyclone rainfall o Storm surge q Resolution - quality and resolution of DTM 3
Size of Catchment Area Considered Red outlines FEMA 100 year flood zones q FEMA FIRMs cover much but not all of the US q In many areas they cover the main rivers but not smaller streams and surface water flooding q Need to model the the water getting to the rivers as well as out of the rivers Blue high resolution model including pluvial (surface) and fluvial (riverine) flooding 4
Pensacola Flooding April 2014 5
Pensacola Flooding Detail Flooded Downtown Area Outside of FEMA Hazard Zones 1 9 4 7 Blue Shading KatRisk Flood Model Red Hatched FEMA Zones A and V 6
Modeling of Tropical Cyclone Rainfall q For each year, count whether the maximum precipitation is from a Tropical Cyclone 7
Flood Model Data Sources q Near Surface Meteorology: NOAA/NCEP/EMC and ECMWF q q q q o KatRisk co-authored papers: A 51-year reanalysis of the US land-surface hydrology Precipitation: NOAA/NCEP/CPC and EMC River Flow: USGS, USACE DTM o o Varying resolution data sets available from USGS. We have used finest scale DTM country wide available. Extensive quality checks and corrections are needed (e.g. bridges), the coarser the resolution of the DTM, the more corrections that will be needed Vulnerability: o o USACE, FHRC (mdx.ac.uk) Claims data less available as compared to hurricane 8
Expertise q Meteorology / Climatology o Global atmospheric data sets o Large scale probabilistic event creation q Hydrology o Rainfall runoff and routing models o Hydraulic modeling o Calibration / validation q Engineering o Flood defenses o Vulnerability and loss modeling q Statistics and Actuarial Science q Software Engineering and IT o Web framework, data bases, Big Data, server management o Various different programming languages and programming frameworks (R, CUDA, C, Fortran, shell scripting, Postgres, MPI, OpenMP, etc.) 9
Validation and Uncertainty q Almost too much to discuss in a short time to be comprehensive. This has been a subject of many scientific papers and projects in the last 20 years q KatRisk co-founders have participated and co-authored many of these studies (PILPS, DMIP, NLDAS, GLDAS, etc.), e.g. o Real-time and retrospective forcing in the North American Land Data Assimilation System (NLDAS) project o The Global Land Data Assimilation System o Overall distributed model intercomparison project results q Co-authored papers on prediction uncertainty o Predicting the discharge of global rivers o The effect of large-scale atmospheric uncertainty on streamflow predictability o Ensemble streamflow forecasting with the coupled GFS-NOAH modeling system q Climate change adds to the uncertainty (co-authored Nature paper in 2011 on flood risk attribution) o Anthropogenic greenhouse gas contribution to flood risk in England and Wales in autumn 2000) 10
KatRisk Flood Model Development q Released 100 and 500 year inland flood depth hazard data on a 10 meter resolution grid. 10, 20, 50, and 200 year return period data completed by the end of this year. q Includes both riverine and pluvial (surface water) flooding 11
Examples in Florida Orlando Tallahassee KatRisk in Blue FEMA zones red hatch overlay 12
TITAN Supercomputer This research used resources of the Oak Ridge Leadership Computing Facility at the Oak Ridge National Laboratory, which is supported by the Office of Science of the U.S. Department of Energy under Contract No. DE-AC05-00OR22725. 13
US Hurricane Modeling q A hurricane track set has been developed for the Atlantic Basin. 25 sets of 10,000 years of tracks (250,000 total years) have been developed conditioned on SST and ENSO. Sample Tracks Loss Costs 14
Running Current Wind Models q To give the models a trial spin, visit our Katalyser site and run one location or as many as 1,000 locations (desktop or on a mobile device). q Scaled back version of the model in terms of the number of simulated years and model resolution with no insurance policies included. q Investigating different platform options for all of our probabilistic models 15
Next Step in the Model Development Process q Modeling of tropical cyclone rainfall and storm surge with our track sets q Development of a correlated event model considering inland flood as well as tropical cyclone rainfall/flooding and storm surge. Initial model planned for later in 2015. 16