Risk Frontiers Flood Hazard Data Flood Hazard Data Flood Forum Victoria Nov 2014
About Risk Frontiers Risk Frontiers is an independent Research & Development company based at Macquarie University Our mission is to improve the pricing of impacts of natural perils, and decision making in respect to the management of natural hazard risks through scientific research and analysis We have been working closely with the (re)insurance industry and emergency managers since our inception in 1994
Insured cost of weather-related perils: present present In the dollars of the day - - -
Normalised insurance costs of weather- related perils: 2011 societal conditions Updated from Crompton and McAneney (2008)
Australia: Coastal Development Gold Coast Main Beach circa 1970 Local Studies Library, Gold Coast City Council Gold Coast Main Beach 2003
NAT CAT Loss Models for Australian key perils EventRankingYearNormalised cost Sydney Hailstorm 11999$4.3 Billion Tropical Cyclone Tracy 21974$4.1 Billion Newcastle Earthquake 31989$3.2 Billion QLD Floods52011$2.5 Billion Ash Wednesday Bushfires 71983$1.8 Billion Normalised insured losses as if all events were to impact upon 2011 societal conditions (Source: ICA/Risk Frontiers) HailAUS 6.1 CyclAUS 3.0 QuakeAUS 4.0 FloodAUS 3.0 FireAUS 2.0
Depending upon the application, areal coverage, data availability, output detail required and budget, flood mapping can be examined at various levels: 1: Broad-scale mapping to delineate floodplains and those areas not subject to flooding. Risk Frontiers’ Flood Exclusion Zones, FEZ, is an example of this sort of modelling. 2: Medium-level mapping based on observed historical water levels and flows. 3: Detailed mapping with full hydrodynamic analysis. Only practical if study area is relatively small. Flood Modelling
8 Flood inundation modelling: water depth and 2011 flood extent (Brisbane)
National Flood Information Database Data integration from two dozen sources
Maribyrnong Flood surface DTM Water Depth
Release history of NFID VersionDateAddresses with Flood Risk Data 117-Dec-08672, Dec-081,185, Mar-091,436, May-091,436, Jul-091,556, Dec-091,631, Mar-102,570, Jun-104,635, Dec-104,965, May-115,151, Sep-115,630, May-125,714, Oct-125,742, May-135,870, Sep-135,928, May-146,038,672
Comparison between modelled flood extent in NFID (white areas) and the 2011 Brisbane Flood inundation extent (red polygon) Validation – NFID vs 2011 Floods
Comparison: NFID versus 1974
Central Kempsey showing Address points over closed road regions and FloodAUS 2001 Design flood ARI=20 yr depth map
Maitland
Gosford
FEZ™ Other companies pursuing (Statistical modelling) FEZ: Alternative to statistical modelling White: modelled floodplain in FEZ Red: PMF extent from trad. flood modelling
Evidence-based spatial integration / filtering process 1 - Distance to river network (at Drainage Division group level) 2 – Water depth at PMF level (at Drainage Division group level) 3 – Same two attributes (at catchment level) 4 – Average slopes in known flood extents 5 – River stream orders (determined from average slopes of waterways) 6 - Distance to shoreline Lots of attributes:
Distance to river networks: comparison Group AGroup B
Red outline – observed flood extent from flooding Blue outline – flood-prone areas delineated by QRA flood overlay project White area – flood-prone areas in FEZ classification Risk Selection: e.g. Bundaberg In absence of NFID data or flood studies for Bundaberg
Bundaberg flooding (imagery 29/01/2013, 50cm-resolution, Qld govt) Yellow – observed flood extent Red – modelled flood plain in Risk Frontiers’ Flood Exclusion Zones (FEZ) FEZ boundary, as a conservative estimate, captures current flood extent well.
FloodAUS Loss Model -- Main Features Uses National Flood Information Database (NFID) – Flood surfaces created by hydraulic engineers for state and local governments – High resolution address data and digital terrain model (DTM) to assess water depth Claims data on residential flood damage used for deriving vulnerabilities (Brisbane, 1974; Katherine, 1998; Central Coast, 2007; validated against 2011 Queensland events) Inter-catchment correlations included from stream gauge data
Flood – equations are well known but computationally difficult and data constrained -- need catchment scale modelling Insurers have a suite of tools for risk selection – FEZ, NFID and, in some cases, in-house modelling Some local councils still hostile to release of data In case of Tweed, release of newer modelling reduced number of at risk homes from ~16,000 to ~8,000 Victoria – Melbourne Water and CMA have been very good at releasing data Victoria – mostly the 1-in-100 extents from State Governments Conclusions