Keqi Zhang, Ph.D.

Professor of Department of Earth and Environment andDirector, Laboratory for Coastal ResearchInternational Hurricane Research Center

Florida International University

October 30, 2014

The Florida Public Hurricane Loss ModelStorm Surge and Freshwater Flood

Enhancement Project

Storm surge and Freshwater Flooding Enhancement Project

• In 2013 work started to enhance the FPHLM by adding both a storm surge and fresh water flooding component.

• The SSFF enhancement project will take 3 to 4 years and is expected to be operational in 2017 contingent on timely acquisition of all data.

• There are over 140 proposed deliverables (tasks) that will have to be completed (about 40 completed so far).

• Participating institutions are:– Florida International University (lead institution)– Florida Institute of Technology– Florida State University– University of Florida– University of Miami– Notre Dame University– AMI Risk Consultants

Project Team• The modeling team include

– Storm surge numerical modelers– Hydrologist (freshwater flood expert)– Meteorology experts– Structural engineers– Coastal engineer– Actuary– Financial, insurance expert– Computer scientists– Database manager– Statistician– GIS expert

Dr. Shahid Hamid Dept. of Finance and IHRC, Florida International University (FIU)

Dr. Shu-Ching Chen School of Computing and Information Sciences, FIUDr. Keqi Zhang Dept. of Earth and Environment and International

Hurricane Research Center, FIUDr. Jean Paul Pinelli Dept. of Civil Engineering, Florida Institute of TechnologyDr. Steven Cocke Dept. of Meteorology, Florida State UniversityDr. Kurtis Gurley Dept. of Civil and Coastal Engineering, Univ. of FloridaDr. Mark Powell Hurricane Research Division, NOAA (now at H-Wind)Dr. Omar Abdul-Aziz Dept. of Civil Engineering, Florida International UniversityDr. Andrew Kennedy Dept. of Civil Engineering, University of Notre DameMr. Bachir Annane CIMAS, University of MiamiDr. Mei-Ling Shyu Dept. of Electrical and Comp Engineering, Univ. of

MiamiMs. Gail Flannery FCAS Actuary, Vice- President AMI Risk ConsultantsDr. Yuepeng Li International Hurricane Research Center, FIUDr. Yi-Cheng Teng International Hurricane Research Center, FIUDr. Sneh Gulati Dept. of Statistics, FIUDr. B. G. Kibria Department of Statistics, FIU

Plus many graduate students

Key Team Members

• Assess storm surge and hurricane related rain flood risk and estimate both the insured and uninsured losses they may create.

• Provide an innovative, transparent, combined model that can separate wind losses from flood losses and try to resolve the issue of who should pay for damages.

• Conduct simulations and scenario analysis that can help state and local government with disaster planning and land use planning.

• Help assess the cost-benefit of disaster mitigation strategies• Help conduct stress test to assess solvency in case of surge

and flood catastrophe.

What will the storm surge enhancement do?

Components of the Storm Surge and Flood Model

• Wind spatial and temporal interpolation model• Storm surge model• Ocean wave model• Nearshore wave transformation model • Freshwater hydrological flood model• Engineering vulnerability simulation models for storm surge

and wave• Engineering vulnerability simulation models for inland flood • Engineering damage models for surge and wave• Engineering damage models for inland flood• Probabilistic ground up and actuarial loss model for surge and

flood• Scenario based loss model for surge and flood• Mitigation model for surge and flood• GIS analysis model for surge and flood• These components will be tested and validated • Software will be developed and tested for these components

FPHLM Architecture

• Storm Track and Intensity Model• Inland Storm Decay Model• Wind Field Model• Terrain Roughness Model• Gust Factor Model• Precipitation Model

Wind and Precipitation Models

Storm Surge Model• 2D Curvilinear fine difference model-CEST• Overland flooding algorithm considering the land cover effect• 5 overlapped Grids to cover the Florida Coast and Lake Okeechobee• NOAA ETOP1, Coastal Relief, and Integrated Coastal Topo/Bathy data,

USGS DEM, FLDEM LiDAR data• Hydrological feature data from water management districts,

counties, and NOAA• 2011 National LCLU data

• Validate the model using NOAA, FEMA, USGS, USACE storm surge observations from historical storms

• Conduct sensitivity analysis of grid extent and resolution

• Examine the numerical stability of the model

Model Validation

Wave Model

• Modified USACE STWAVE model

• Topo/Bathy and land cover data used by the surge model

• Wave transformation in shallow water, wave breaking, and dissipation over the land

Sensitivity Analysis

• Overland dissipation• Spatial resolution• Using output wind speeds and

water levels from surge model• Selecting wave grids where

maximum surge level exceeds threshold

• Run times of Ivan for a single 40m grid using one CPU: 10s. Total over 22 active grids: 268s ~4 minutes

Freshwater Model: Objectives

• Develop large basin-scale freshwater flood models by using the modified the EPA-SWMM 5.0 (open-source) model to predict the freshwater flooding during extreme rainfall events .

• Calibrate and validate the SWMM models using the historically available data.

• The storm surge depth (sea water) will be aggregated with SWMM model output to estimate the combined freshwater-seawater mass and flooding depths.

Freshwater Model: Delineation of Basins

Mexico Gulf coast

Central Florida

East coast

• The level-6 USGS Hydrologic Unit Codes (HUC-6) for Floridawere combined into 7 large hydrologic basins.

• Developing 7 regional stormwater models using EPA SWMM5.0

Currently building the two SWMM models:

Florida SoutheastCoasts Basin

Area ~9,000 km2

Central Florida Basin (Kissimmee River)

Area ~10,000 km2

Freshwater Model: Pilot Study with a Mid-size Basin

Myakka River Basin; Area: 1557km2

NSE = 0.63

Freshwater Model: Pilot Study ResultsMyakka basin SWMM calibration

Engineering Vulnerability Model

The enhanced FPHLM will combine:

• Existing wind and rain vulnerability models Personal residential model (including manufactured homes) Low-rise commercial residential model (1 to 3 story high) Mid/high-rise commercial residential model

• Storm surge vulnerability model (in progress) Personal residential model (including manufactured homes) Low-rise commercial residential model (1 to 3 story high) Mid/high-rise commercial residential model

• Fresh water inundation vulnerability model (in progress) Personal residential model (including manufactured homes) Low-rise commercial residential model (1 to 3 story high) Mid/high-rise commercial residential model

Storm Surge and Fresh Water Inundation Vulnerability Models

This is a work in progress. Expected completion date: 2016-17

The tasks include:

• Identify key building components for vulnerability to evaluate/classify the overall system performance Foundation (Slabs, piers, piles, basement) Structure (roof framing, structural envelope, and exterior walls) Interiors (floor covering, drywalls, partition walls) Mechanical Systems (Ducts, heating and ventilation) Electrical Systems (lighting components and wiring) Plumbing Systems

• Define a series of damage states from no damage to total loss

• Develop fragility curves for each state that will relate the probability of damage to different hydrological states (combination of flooding height, wave, current, etc.).

Combined Vulnerability ModelThis is a work in progress. Expected completion date: 2017

The objective is to identify a method to:• calculate the combined surge/inundation damage and

wind/rain damage• calculate the contribution of each hazard to the total

combined damage• Avoid double counting damage

Several methods are currently being investigated including:• Simple averaging• Weighted averaging• Linear regression• Step-wise regression• Expert rules

Calibration and Validation of Vulnerability Model

This is a work in progress. Expected completion date: 2017

Several sources of data will be used for calibration and validation of the data

• A library of post storm damage assessment databases has been compiled: http://research.fit.edu/whirl/storm_damage.php

• A library of case studies is being built: http://research.fit.edu/whirl/projects/fpublic_studies.php

• Claim data from NFIP and private insurers is being pursued.

• Exposure study from different sources like tax appraiser databases

Actuarial Model• Probabilistic actuarial loss model for surge and flood.• Scenario based loss model for surge and flood.• Develop method to assign wind vs. flood related losses based

on the assignment of separate wind and water vulnerability functions and the relative timing of wind and water provided by meteorology and surge modeling teams.

• Actuarial modifications for existing policy characteristics.• For model development, calibration and validation need

current flood insurance exposure data and exposure and claim data for past hurricane.

• The main source of insurance data for flood exposure and claim is the FEMA National Flood Insurance Program data. It is proprietary data and we are in the process of acquiring it.

• Other potential sources are Department of Emergency Management in Florida and state agencies in Texas, possibly Louisiana and New York.

Computational Architecture and Facility

High-level overview of the modified system architecture of the FPHLM showing the enhancement components in red.

New hardware to accommodate additional computational requirements:• 7 compute nodes and 1 storage expansion chassis

• 448 CPU cores @ 2.5 GHz• 3.5 TB RAM• 56 TB of storage• All nodes interconnected using 10 Gbps links