April 17, 2015 by Jason Contant, Online Editor
Guy Carpenter, a global risk and reinsurance specialist and a wholly owned subsidiary of Marsh & McLennan Companies, has launched its new G-CAT Canada Flood Model, what it calls the “first-ever probabilistic flood model for Canada” that will allow insurers to assess their exposure to the peril.
“By launching the G-CAT Canada Flood Model, Guy Carpenter addresses the concern of extreme losses from natural hazards in Canada, such as flooding,” said Donald Callahan, president and CEO of Guy Carpenter Canada. “To produce this state-of-the-art flood model for the Canadian insurance market, Guy Carpenter collaborated with experts in hydrological and hydraulic modeling to produce simulations for hurricane, rainfall and river flow flooding catastrophes.”
The product was launched on Thursday at the Design Exchange in downtown Toronto. The flood model evaluates fluvial (or riverine) overland flood risk, together with the off-plain component of such events, but does not currently include pluvial, coastal flooding, flash flooding, mudslide (slope failure), groundwater, sewer back-up or tsunami risk. By focusing on riverine risk, Guy Carpenter felt that it had “captured the majority of the hazard,” Joseph Becker (pictured right), research hydrologist at Guy Carpenter, told attendees.
The flood model also features:
• Pre-built specific historical Canadian events;
• User-defined scenarios developed upon request;
• Collaboration with JBA Risk Management on hazard maps & stochastic event set;
• Variable resolution grids, with the finest resolution of the flood hazard 10 metres in urban areas;
• A stochastic event set of nearly 80,000 synthetic events across Canada; and
• The incorporation of existing river defenses, such as dams and levees and probability of defence failure.
The model enables clients to evaluate both exposure to floods and the financial cost of a range of possible flood events. It includes historic flood events, such as the June 2013 Alberta floods, considered among the worst in Canadian history, with insured losses estimated by Guy Carpenter at Cdn$1.8 billion and economic loss estimated at Cdn$6.2 billion.
Flood maps are created using two hydraulic modelling techniques that take into account Canada’s large geographic area and the diversity of its landscape, as well as the population density of the area being modelled and a comprehensive view of river flood defenses throughout the country.
“Flood maps are the basis of flood management,” Paul Kovacs (pictured right), executive director of the Institute for Catastrophic Loss Reduction, said during his presentation. “Very few provinces have come up with a strategy to deal with urban flooding,” he added.
Becker told attendees that the G-CAT Canada Flood Model implements four major modules: hazard (including defence implementation and historic flood scenarios, such as the recent Alberta floods, and those in Winnipeg/Red River in 1997, Saguenay, Que. in 1996 and Winnipeg/Red River in 1950); exposure (including land use and census data); damage (including damage distributions fitted and calibrated on the basis of Canadian claims data from recent flood events); and financial (including the calculation of insurance terms and conditions, such as deductibles and limits).
Other key features of the model include the following:
• Hydrological modelling based on rainfall data; modelled river flows calibrated using historical discharge data from catchments representing the main hydrological regimes across Canada;
• Taking account of information on standards of protection from river defence structures available from Canadian river authorities or in the public domain. “Where concrete defence information is not available, assumptions as to defence standards are made using land use and/or population data,” Guy Carpenter said.
• Resulting event set contains information on extreme rainfall, river flood and a combination of two (“coupled” rainfall-river events); event set also identifies flooding events that are caused by hurricane activity;
• Rainfall-runoff models used to model river flows from precipitation data are designed specifically to incorporate the influence of the spring snowmelt phenomenon;
• A module that “features an accurate disaggregation scheme that uses recent census data counts from Statistics Canada (2011 vintage) and CEC 2010 land use data”; and
• Vulnerability development informed by recent Canadian flood claims information (representing all lines of business) collected by Guy Carpenter from a number of Canadian insurers.
— Steve Wilson (@InsuranceMedia) April 16, 2015