Transformative Household-level flood and hurricane risk assessment tools for the City of Houston Leonardo Dueñas-Osorio (CEVE) Robert Stein (POLI) Devika Subramanian (COMP) Candase Arnold, Brian Bue, Rachel Carlson William McGuinness, Alexander Tribble, Jeffrey Weed Office of Emergency Management Houston, Texas January 22, 2008
56
Embed
Transformative Household-level flood and hurricane risk ...devika/evac/Rice-OEM... · Transformative Household-level flood and hurricane risk assessment tools for the City of Houston
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Transformative Household-level flood and hurricane risk assessment tools
for the City of Houston
Leonardo Dueñas-Osorio (CEVE)
Robert Stein (POLI)
Devika Subramanian (COMP)
Candase Arnold, Brian Bue, Rachel CarlsonWilliam McGuinness, Alexander Tribble, Jeffrey Weed
Office of Emergency ManagementHouston, Texas
January 22, 2008
Presentation Outline
I. Research Motivation
II. Risks from wind and flood hazards
III. Structural risk estimation
IV. Risk computation and visualization
V. Perceived risk characterization
VI. Summary and future work
I. Research motivation
Collective Dynamics
• Hurricane Rita and Houston’s Infrastructure Usage
Evacuation Zoning
• Current disaster management strategies
ZIP code level evacuation policy
- Requires large scale evacuation
- Relies on intercity sheltering
- Aggregates data at the ZIP code level
Structural Safety Tagging
• Transformative disaster management strategies
- Avoids unnecessary evacuation
- Encourages intra-city sheltering
- Offers flexibility of data aggregation
- Reveals geospatial variability of risk
Household-level risk assessment strategy
SAFESAFESAFE
SAFESAFEUNSAFEUNSAFE
UNSAFE
UNSAFEUNSAFEUNSAFE
RestrictedRestrictedRestricted
Agent-Based Modeling
• Multi-disciplinary Approach
Estimation of floodand wind risk at
individualhouseholds
Behavioralcharacterization
of households
Agent-basedmodeling of
evacuation andsheltering
Data-Rich Environment
Demographic characterization
Post-disaster utility availability
Best intra-city sheltering and escape routes
Evacuation policies
What is the actual risk?
What is the perceived risk?
What are the resulting collective dynamics?
II. Risks from wind and flood hazards
Geocoding
Visualization
Database of residences
Computational Risk Assessment
Risk Estimation Methodology
Data Source 1
• Harris County Appraisal District (HCAD) data
– Aimed towards keeping detailed Harris County property tax records
– Exterior wall type, property dimensions, assessed structure quality (A-F), foundation type, year built
– Contains approx. 1 million residential properties
Data Source 2
• Hurricane Rita survey data (Stein, 2005)
– Performed shortly after the events of Rita (09/05)
– Input from 648 homeowners
– 216/407 total in Harris County
– 175/216 located in HCAD db (thus far)
– Wind and flood risks manually assessed by civil engineers in our group (214/216)
Geocoding
• Geocoding: street address → (latitude, longitude), elevation
• Addresses are automatically coded via Google maps web interface
• Performance: ~ 10k addresses / hour
Geocoding is essential for assessing geospatial proximity – relevant for individual household flood and wind risk assessment
Automated Risk Assessment
• Problem with source data 1: far too many properties in HCAD to characterize manually
• Approach: computationally assess risks by analyzing residence characteristics in HCAD data for given hazard scenarios. Use geospatial proximity and similarity of structural properties to scale up household level evaluation sample size.
III. Structural risk estimation
HAZUS-MH
• HAZards United States – Multi-Hazards
• GIS-Based Program freely distributed by FEMA
• Developed for institutions
• Evaluates hurricanes, flooding, and earthquakes
Leveraging the HAZUS Framework
• HAZUS
– Evaluates vulnerability at the census-tract level
– Informs about evacuation for entire areas, regardless of building vulnerability
– Contains specific structural performance data that is useful when detailed inventory is available
– Available only to GIS users
– Complex data outputs
• Rice/OEM
– Evaluates vulnerability at the building level
– Allows for household-level evacuation decisions
– Uses public HCAD inventory that is useful when structural behavior data for individuals households is available
– Available to anyone with Internet access / mobile phone
– Easy to use output scale
Hazard Scenarios for Pilot Test
• Choose scenario for:– Wind (“Rita-like”
hurricane)– Flooding (100-yr flow
+ surge)
• Determine wind risk and flood risk for area households
• Once one scenario is run, ANY scenario can be run
III. a. Wind scenario
HAZUS Fragility Curves
• A fragility curve provides the probability of specific damage states as a function of hazard intensity.
• HAZUS contains specific fragility curves for multiple building types.
• 80/112 applicable residential housing fragility types for Houston.
Wind fragility curve
Parametric HAZUS Fragility Curves
• HAZUS curves are prototypical of common structural types. Parameters include:– Design (stories, garage, roof shape)
– Construction (roofing nail size, roof/wall connection)
– Location (surrounding terrain)
• HCAD data contains indirect information to determine HAZUS parameters:– Year Built � Existence of roof straps
– Examining perimeter of floor v. total sq. footage �Number of Floors
– Exterior wall type � Structure type
Coupling HCAD and HAZUS
• Use HAZUS fragility curves to develop HCAD fragility curves
• Report risk assessment at the household level
• Validate and compare Rice/OEM approach with HAZUS and building inspectors
- Average nailing patterns per structural type
- Average roof configurations
- Average of houses with and without shutters
HCAD Parameters in HAZUS
HAZUS Fragility Curves
Implies highly accurate knowledge of individual structures
Creating HCAD Fragility Curves
• Group similar HAZUS types– Averaged the fragility of similar homes
• Obtained log-normal probability distribution parameters– Creates smooth graph
– Searchable for ANY wind speed
– Useful to report risk estimates at the household level
An HCAD fragility curve
Wind Risk Assessment
• HCAD fragility curves determine damage likelihood for a given wind speed
• Plug scenario-determined wind and estimate vulnerability from curve
• Determine risk from a Rita-like storm to all area households
III. b. Flooding scenario
Floodplain Map Challenges
• Floodplain Maps
– Hazard based
– Flooded or not flooded
– Few rainfall scenarios
– Only one surge scenario
• Rice/OEM Output
– Risk-based
– Allow for water depth to be calculated
– ANY rainfall scenario
– ANY surge scenario
Flooding Profiles
Using TSARP Models in Corps of Engineers’ HEC-RAS – Harris Gully to create Floodplain Map
Longitudinal Centerline Profile
Stream Cross-section
Flooding Area
Overlaying LIDAR Data in ArcGIS –Harris Gully 100-yr storm
Flooding Vulnerability
Comparing Data and Creating Coordinate-Searchable Map – Harris Gully 100-yr storm with 7’ surge
Flooding Vulnerability
Expanding scenario to area within Beltway 8
Flooding Vulnerability
Creation of “Category 5” scenario – 500-yr storm, 25’ surge
IV. Risk computation and visualization
Output Visualization
Web Interface
• Web access:– Risk assessment
information tailored to each address
• Access via Google Earth:– All data can be
exported to a Google Earth-compatible format (for offline access)
Household-Level Flood Risk Analysis
Regional Flood Hazard Analysis
Image credit: TSARP
Perceived vs. Estimated Flood Risk
Perceived risk Estimated risk
For the Houston Beltway Area:
V. Perceived risk characterization
The Policy Problem
• Increasing numbers of persons (households) are making choices in the face of severe weather experiences that put themselves and others at risk (e.g., congestion of roadways and shelters).
Multidisciplinary Research Questions
• How are perceptions about risk from severe weather experiences related to estimated risk?
• How are perceptions about risk from severe weather experiences and estimated risk related to evacuation behavior?
• What are the most efficacious strategies for informing/persuading (as opposed to coercing) the public about the estimated risks from severe weather experiences and how best to respond?
Research Design
• Rice/Houston Chronicle Survey– Sept. 29 – Oct. 3, 2005– 650 Households from 8 county Houston area– 85% response rate– Analysis limited to Harris County residents
N=216 with some missing data.
• Residential property’s vulnerability to flooding in Harris County
Evacuation behavior Harris County
• Did you evacuate or attempt to evacuate your home to go someplace safer before Hurricane Rita hit?
• Yes: 51%• No: 49%
N = 216
Estimated vulnerability to Flooding
• Likelihood residential structure would flood in severe storm from Tropical Storm Allison Recovery Project:
– 5: houses are well within the 100-yr floodplain
– 4: houses on the fringe of the 100-yr floodplain
– 3: houses outside the 100-yr floodplain
– 2: houses are the edge of the 500-yr floodplain
– 1: houses far from any floodplains
Perceived vulnerability
• Now, aside from the risk of an initial storm surge, consider the risks of flooding from rain and rising water from hurricane Rita. In your opinion was your residence/home located in a high risk, a medium risk or a low-risk area for flooding from rain and rising water from Hurricane Rita.