The AIR Inland Flood Model for the United States In Spring 2011, heavy rainfall and snowmelt produced massive flooding along the Mississippi River, inundating huge swaths of land across seven states. As the flood waves traveled downstream, river levels in many areas surpassed those of the Great Flood of 1993. The 2011 floods damaged or destroyed more than 21,000 homes and businesses, some as a result of opening levees in an effort to reduce the overall impact.
THE AIR INLAND FLOOD MODEL FOR THE UNITED STATES At any given time, a storm with the potential to produce flooding is occurring somewhere over the contiguous United States. While thousands of towns and cities in the U.S. are located on floodplains, many areas far from these low-lying river valleys are also susceptible to flood damage and loss. And as seen during the Mississippi River floods of 2011, even sophisticated flood defense systems cannot always protect homes and businesses in harm s way. The enormous watershed of the Mississippi River covers an area of 1.2 million square miles, and major floods can overwhelm defenses and cause a huge amount of destruction while disrupting businesses and infrastructure. Up until now, risk managers have been unable to account for the many complexities that affect flood risk across the country, which include the continual changes in the characteristics of insured properties as well as understanding the severity, frequency, and location of potential flood events on and off the floodplain. A probabilistic approach is the only method that can account for all the complex variables associated with flood risk. A Comprehensive View of Flood Risk across the United States The AIR Inland Flood Model for the United States provides companies with the most detailed probabilistic model available for assessing and managing inland flood risk at high resolution, for locations on and off the many and varied floodplains across the U.S. Number of Flood Dis as ter Declarations 1974-2013 1 2 3 4-13 2 Number of flood disaster declarations by county, 1974 2013. Long and damaging floods are common in many regions, including an area of the Upper Midwest between the Missouri and Mississippi rivers.
THE AIR INLAND FLOOD MODEL FOR THE UNITED STATES An Innovative Solution for Simulating Realistic Precipitation Patterns AIR s innovative approach to simulating rainfall couples a global Community Atmospheric Model (CAM) with a regional Weather, Research, and Forecasting (WRF) model; the results are realistic and statistically robust storm patterns, over space and time. A sophisticated downscaling technique enables the model to capture both large- and small-scale precipitation patterns, including the bursts of extreme rainfall that contribute to highly localized flooding. A AIR s model captures inland flood risk for 18 hydrological regions across the contiguous U.S., an area of more than 3 million square miles, with a river network 1.4 million miles long. It includes 335,000 distinct drainage catchments, and all streams with a minimum drainage area of 3.9 square miles are modeled explicitly. B C The model simulates both the weather systems and the flooding they cause, and accounts for a broad range of climates, local soil conditions, land use, and topography. Water is routed through 1.4 million miles of the country s extensive river networks, including every stream with a drainage area of at least 3.9 square miles. Simulated flood propagation accounts for the effects of more than 20,000 lakes and reservoirs. Damage is determined by calculating the inundation depth at each affected location, taking into account the country s extensive system of levees (and their probability of failure), as well as regional differences in building codes and building practices. The model s high level of detail captures events of all levels of severity. The catalog includes catastrophic floods that produce insured losses in excess of USD 25 million the ISO Property Claim Services (PCS ) threshold for issuing a catastrophe serial number as well as non-catastrophic floods, with losses as low as USD 10,000. The global Community Atmospheric Model is coupled with the regional Weather, Research, and Forecasting model (A). The output (B) is downscaled to yield realistic precipitation patterns (C). The potential for flooding depends greatly on antecedent conditions the amount of prior rainfall or snowmelt which determine the saturation levels of the soil. The high resolution output from AIR s precipitation model is used to compute, on a continuous basis at a local level, the soil-water balance at the beginning of a storm. When soils cannot absorb any more water, the additional rainfall streams downhill as surface runoff. 3
THE AIR INLAND FLOOD MODEL FOR THE UNITED STATES CUSTOM AND SECONDARY RISK MODIFIERS The AIR Inland Flood Model for the United States supports several individual risk characteristics (secondary modifiers) that can be used to modify the vulnerability of an exposure. For example, for basement foundations, a finished or unfinished basement can be specified. Flood damage is highly dependent on elevation, and the AIR model provides options for specifying ground elevation, height of the first floor above ground level, and the floor of interest. Additional modifiers include the height of a custom-built levee or flood wall protecting a building. The level of contents vulnerability can also be provided if the insured has flood-proofed their contents. 1.0 Probability of Failure Fragility curve typically assumed Severity of Loading Standard of protection provided by defense True Fragility Curve AIR s fragility curves provide accurate estimates of flood defense failure, fulfilling a crucial part of U.S. inland flood modeling. calculated for any point of interest on the floodplain. These cross sections include details of the terrain, which were extracted from a high-resolution digital terrain model (DTM). Water flow is also affected by the surface roughness, which is based on land use and land cover data from the USGS. 0 1 2 0.5 Miles 0 1 2 0.5 Miles AIR s hydraulic model realistically simulates flood extents. Water elevation is determined at the centerline of the river at each cross section (left panel). This data is then used, in part, to create continuous flood extents at the same resolution as the underlying 30-meter DTM (right panel). Sophisticated Methodology for Determining Flood Inundation Depth The AIR model uses a physically based hydraulic model that calculates the extent of the flooding and the depth of inundation at each location of interest on the floodplain. Hydraulic calculations are performed at more than 4 million river cross sections at approximately 500-meter intervals. This enables flood depth and extent to be Flood Defense Modeling Recognizing that levees can fail is a critical part of flood risk management in the U.S. A sizable portion of the nation s extensive levee network was built more than 50 years ago, and maintenance has become increasingly complex and generally underfunded. Many U.S. levees have weakened over time, increasing the probability of failure. The AIR model accounts for the current state of the country s levees by analyzing detailed data from more than 14,000 miles of the levee network. The model uses fragility curves to determine if and when a levee fails. Although each levee is designed for a specific standard of protection, it may fail at flood levels below that standard, or even withstand floods above the standard. AIR s fragility curves incorporate the uncertainty. The levee system along the Mississippi River is one of the world s largest, particularly along the lower stretch of the river from Cairo, Illinois, to New Orleans where about 1.5 million homes rely on more than 1,00 miles of the mainstem flood protection system. Here, the model relies heavily on levee data and fragility curves to simulate floods along incremental stretches of the river. 4
THE AIR INLAND FLOOD MODEL FOR THE UNITED STATES Modeled levee breach producing a 100-year flood extent along the lower Mississippi River north of Memphis, Tennessee. (Source: AIR) Building Damage Is Determined at the Component Level The majority of the damage caused by most floods is to contents and nonstructural components. After the floodwaters recede, mold is a problem with a severity that depends on flood depth and the construction material used for internal walls, as well as the thoroughness of drying that takes place. To capture the relative vulnerability of building fabric, fixtures and fittings, and service, the model uses a componentbased approach to estimate damage for all lines of business. Basement contents damage from the 2013 floods in Illinois (left panel); Mold in a home that was submerged during the 2011 Mississippi River flood (right panel). (Source: FEMA) OFF-FLOODPLAIN RISK A SIGNIFICANT DRIVER OF LOSS The AIR Inland Flood Model for the United States supports several individual risk characteristics (secondary modifiers) that can be used to modify the vulnerability of an exposure. For example, for basement foundations, a finished or unfinished basement can be specified. Flood damage is highly dependent on elevation, and the AIR model provides options for specifying ground elevation, height of the first floor above ground level, and the floor of interest. Additional modifiers include the height of a custom-built levee or flood wall protecting a building. The level of contents vulnerability can also be provided if the insured has flood-proofed their contents. Among the factors influencing the extent of damage from floods are construction type and occupancy, which often relate to a building s level of protection or to the presence of a basement. Height is another variable that can affect a building s response to flood; typically, greater engineering attention is given to high-rise buildings attention that often includes flood protection. In addition to these primary building characteristics, the model supports a variety of secondary characteristics, including foundation type, number of basement levels, first floor height, contents vulnerability, and floor of interest. 5
THE AIR INLAND FLOOD MODEL FOR THE UNITED STATES To estimate time element losses, specific damage functions were developed that vary by occupancy and account for building damage level, building size and complexity, and business characteristics, such as resiliency and ability to relocate. DEFINING FLOOD EVENTS The AIR Inland Flood Model for the United States defines flood events based on a physical understanding of what causes flooding and how floods propagate through the country s extensive river networks. Extreme runoff and river flows are clustered into separate flood occurrences based on their spatial and temporal proximity. The methodology allows the model to accommodate the current practice in the industry of defining events using the 18-hour clause. Extensive Validation Confirms the Model s Robustness All components of the inland flood model were rigorously validated, from the spatial distribution of precipitation and its accumulation, to modeled flood extents for different return periods and modeled historical events. (A) 14 Flow (1,000 m 3/s) 12 10 Observed 8 Modeled (B) 4 2 ul 17 ul 12 7Ju l 2Ju l un 27 un 22 un 17 12 AIR modeled flood extent (A) and aerial imagery of maximum flood extent from DigitalGlobe (B) near Ashland City, Tennessee, during the 2010 Cumberland River flood. un 0 Peak discharges for a catchment along the Susquehanna River during the 200 Northeast flood and a validation of modeled peak discharges for the gauging station (red triangle) at Marietta, Pennsylvania. AIR also obtained gauge station data, from more than 9,000 river gauging stations across the country, to validate modeled river discharges. The modeled river flows for historical events were compared with observed peak discharges at the stream links corresponding to each gauge station.
THE AIR INLAND FLOOD MODEL FOR THE UNITED STATES CONTENTS The modeled flood extents generated for historical events compare well with observed footprints derived from satellite imagery taken during the actual event. AIR s modeled loss estimates validate well against reported insured losses from floods across the country, for both onand off-floodplain locations. Understanding, Pricing, and Transferring Inland Flood Risk A probabilistic approach is the only method that can account for all the complex variables associated with flood risk. AIR s approach provides underwriters with the most detailed probabilistic U.S. inland flood model available, enabling you to assess and manage your inland flood risk for locations on and off the floodplain. The AIR model also helps establish a currency for understanding and quantifying risk in insurance and reinsurance transactions. Insured Loss Insured Loss Single Family Homes Single Family Homes Modeled Observed (trended to 2012 values) Apartments Commercial Total BUILDINGS Modeled Observed (trended to 2012 values) Apartments Commercial Total Modeled insured losses compare well with observed insured losses for residential and commercial buildings (top) and contents (bottom), California flood of 1995. MODEL HIGHLIGHTS Incorporates an innovative and sophisticated approach to simulating large-scale precipitation by coupling a Global Climate Model and a Numerical Weather Prediction (NWP) model to produce realistic and statistically robust rainfall patterns in space and time, across seasons State-of-the-art approach to downscaling simulates precipitation fields at high resolution Fully accounts for flood defenses in hydraulic modeling Explicitly models off-floodplain losses to capture a major source of insured flood losses Features a component-based approach one that divides a building into building fabric, fixtures and fittings, and services to estimate the vulnerability of buildings and their contents for all lines of business Primary risk characteristics include occupancy and construction type, height, and year built; 10 secondary modifiers are supported to cover foundation type, basement finish, first floor height, contents vulnerability, floor of interest, and other characteristics Identifies separate flood occurrences spatially and temporally, reflecting the physical properties of a flood while accommodating a 18-hour clause Validated through extensive analyses of detailed loss experience data obtained from FEMA for historical flood events Enables improved risk selection and accurate assessment of portfolio risks 7
Model at a Glance Modeled Perils Supported Lines of Business Supported Construction and Occupancy Classes On-floodplain riverine flooding for a river network of 1.4 million miles, including all streams with a minimum drainage area of 3.9 miles, with 335,000 drainage catchments. Off-floodplain flooding for areas away from floodplains. Residential, commercial, small industrial, large industrial facilities, agricultural buildings, and automobiles Number of Supported Construction Classes: 43 Number of Supported Occupancy Classes: 110, including 2 for large industrial facilities Unknown Damage Functions: When detailed exposure data (for example, construction type or height) is unavailable, the model applies an unknown damage function that takes into account regional construction characteristics. ABOUT AIR WORLDWIDE AIR Worldwide (AIR) provides risk modeling solutions that make individuals, businesses, and society more resilient to extreme events. In 1987, AIR Worldwide founded the catastrophe modeling industry and today models the risk from natural catastrophes, terrorism, pandemics, casualty catastrophes, and cyber attacks, globally. Insurance, reinsurance, financial, corporate, and government clients rely on AIR s advanced science, software, and consulting services for catastrophe risk management, insurance-linked securities, site-specific engineering analyses, and agricultural risk management. AIR Worldwide, a Verisk (Nasdaq:VRSK) business, is headquartered in Boston with additional offices in North America, Europe, and Asia. For more information, please visit www.air-worldwide.com. 2018 AIR Worldwide A Verisk Business