On January 26, 2016, ICLR conducted a Friday Forum seminar with representative of LexisNexis Risk Solutions and JBA Risk Management, entitled 'IBC's new flood maps: Leveraging data to effectively assess and manage flood risk.
With the increase in the frequency and severity of flooding and extreme weather events there is a greater need to provide Canadian homeowners with adequate protection against flooding and effective flood insurance. LexisNexis has been working with the Insurance Bureau of Canada (IBC) to develop and manage its flood risk program for Canada. A key component of this initiative is the creation of all new pluvial and fluvial flood maps for Canada. These new models, produced by LexisNexis partner, JBA Risk Management, fully leverage local river, rainfall, snowmelt and higher resolution Digital Terrain Model (DTM) datasets, resulting in completely updated river flow and rainfall estimates based upon a much more detailed hydrological study where snowmelt is now explicitly modelled. The resulting maps represent the most comprehensive and detailed set of Canadian flood maps, reflecting the uniqueness of the local environment and greatly improving risk assessment outcomes. Using LexisNexis technology, IBC will be able to quantify Canada’s flood risk and clearly identify the number of properties at risk of flooding and the associated economic losses for virtually any geography in Canada. LexisNexis is also working with Canadian insurers as well as UK, Irish and European insurance customers to help them gain better insight into their flood exposure and increase visibility into their overall property underwriting.
In this talk, speakers reviewed the newly developed flood models and presented key findings from the data in terms of exposure hotspots across Canada. They also discussed best practices for using the data and applying the flood models, as well as other key risk data, in pricing and underwriting decisions, based on their experience working with Canadian, UK and European insurers.
2. Slide 2
CONFIDENTIAL
Introduction/context
Not yet any national mapping standards
Provincial/municipal maps:
Have variable coverage
Are often outdated
Are provided in inconsistent formats
Ambitious project – mapping for every single river, plus pluvial
flooding
Considering data availability, what could we achieve?
Downtown Calgary floods, June 21st, 2013
(Flickr user Wilson Hui - Attribution License 2.0)
3. Slide 3
CONFIDENTIAL
Mapping approach
Traditional engineering approaches to flood mapping, but with a
hi-tech twist!
Extensive validation and correction at all stages
Generate river
network and
catchment
boundaries
GIS analysis of
elevation data
Correction of
any errors
Hydrological
modelling
Generate river
flow estimates
and rainfall
totals
Different
methods to
reflect different
river regimes
Hydraulic
modelling
Largest
dedicated flood
modelling grid
in the world
Different
methods to suit
different
elevation data
4. Slide 4
CONFIDENTIAL
Elevation data
Digital Surface Model (DSM)
represents the earth’s surface as detected by radar
includes features such as tree canopies and tall buildings
Digital Terrain Model (DTM)
representation of the true ‘bare-earth’ ground level
34% of national population mapped on bare-earth terrain data
DTM DSM
5. Slide 5
CONFIDENTIAL
Hydrology introduction
Purpose: methods to calculate river flows and rainfall totals for any
ungauged location in the country
No published Canada-wide methods for estimating these
However, high quality data plentiful
6. Slide 6
CONFIDENTIAL
Hydrology overview
Input data
Water Survey of Canada - HYDAT database of 1,664 river gauge records
Environment Canada – IDF curves and snow depth records at 565
gauges
Ecological Framework of Canada – ecozones to calculate losses e.g. from
urban drainage
Represents local characteristics
Significant influence of lakes
Snow-melt
Frozen impermeable ground
Environment Canada rain gauges
20yr rainfall surface – 1 hr duration
Rainfall surface - 20yr - 1hr duration
High : 55 mm
Low : 5 mm
55mm
5mm
7. Slide 7
CONFIDENTIAL
Design rainfall estimation
Rainfall hyetographs for every catchment across Canada
Intensity-duration-frequency (IDF) statistics at 565 gauges used to
calculate rain depth per return period
Three storm durations for each return period, to capture critical storm
duration in different topographies
Adjusted to account for snowmelt & frozen ground
Interpolated to create
continuous rainfall surfaces
Hyetograph generation
using US Soil Conservation
Service method
Percentage runoff varies
with land use
8. Slide 8
CONFIDENTIAL
Snowmelt & frozen ground
Rain on snow can lead to significant flooding
Losses reduced by:
Priming of depressions (snow-filled)
Frozen ground
Lack of vegetation
Volume of water from snowmelt on successive days calculated from snow
depth gauges
Snowmelt amounts combined with rainfall totals to generate winter runoff
at each gauge
Considered alongside summer rainfall totals – worst case taken forward
Stirling Flood, 25th Jan 2010 (Flickr user
Robert Taylor. Attribution License 2.0)
9. Slide 9
CONFIDENTIAL
Peak flow estimation for main rivers
Adapted from Faulkner, D., Warren, S. and Burn, D (2015) – Design floods for all of Canada. Accepted by Canadian
Journal of Water Resources
Input data (HYDAT flow data, lakes, catchment boundaries)
Data checking & development of flood peak dataset – daily mean and
instantaneous AMAX flowsCalculation of an index of lake attenuation
Estimation of the index flood (QMED) at suitable
gauging stations
Development of regression model to estimate QMED from effective
catchment area and FARL
Spatial interpolation of residuals from regression model using kriging, to
derive local adjustment factors for QMED
Regionalization: spatial interpolation of L-moment ratios
using kriging
Calculation of QMED for ungauged catchments
Calculation of return period design flows at ungauged locations using growth
ratios and QMED
Flood Frequency Analysis - fitting of growth curves (GEV
distribution) at gauged locations
10. Slide 10
CONFIDENTIAL
Hydrology summary
We developed methods that enabled us to:
Calculate rainfall amounts for all catchments in Canada
Calculate river flows for 24,000 locations along the national river
network
Account for the impacts of snowmelt
Represent different land uses and ground conditions
These form some of the inputs into our hydraulic models…
Whitemud Park, Edmonton, 23rd June 2013. Flickr
user Kurt Bauschardt. Attribution License 2.0
11. Slide 11
CONFIDENTIAL
Hydraulic modelling
RFlow®
Extremely fast
Credible outputs
Used in largely
unpopulated areas
JFlow®
Fast
Specifically designed
for large-scale hazard
mapping
Developed since 2002
Benchmarked by UK
government
River flow
estimates
Rainfall
estimates
Elevation
data
Maps of flood
depth and extent
12. Slide 12
CONFIDENTIAL
What’s the result?
National flood hazard maps for 7 return periods at 30m resolution
> 5 million km of river mapped
Pluvial flooding for every location
Full 2D hydrodynamic simulations for three storm durations, for each
return period, on a 30m grid!
Calgary: 1/100-year fluvial map Calgary: 1/100-year fluvial & pluvial map
14. Slide 14
CONFIDENTIAL
Application of flood hazard maps
Insurance metrics
Underwriting/risk scores
Pricing tools
Eg Annual Damage Ratios
Probabilistic models LOCATION B
Total insured value of building =
C$300,000
MeanADR Buildings = 0.000008
Buildings flood premium = $2.40 LOCATION C
Total insured value of building =
C$300,000
MeanADR Buildings = 0.00266
Buildings flood premium = $798
LOCATION A
Total insured value of building =
C$300,000
MeanADR Buildings = 0.00012
Buildings flood premium = $36
Flood warning &
forecasting systems
15. Slide 15
CONFIDENTIAL
What’s next?
Understanding of flood risk across Canada is better than ever – but
more to do!
Local hydrology & hydraulic studies
To do more, we need more:
Flood defence data
High quality terrain data – LIDAR
Investment
Accuracy
Cost
We are here!
16. Slide 16
CONFIDENTIAL
Thank you!
All photos: https://creativecommons.org/licenses/by/2.0/uk/legalcode
Flickr user Ryan McGilchrist. Taken April 21st 2013. Highway 11. Attribution License 2.0