Mission count before landfall Before Hurricane Katrina made landfall, mission estimators already knew, with a few percent, the damages that would occur thanks to estimate models honed over years of experience. NOAA photo. Back to Features

Katrina mission estimators use models for a fast response
By Dave Harris, public affairs specialist, Louisiana Recovery Field Office  9/22/07

NEW ORLEANS, LA … As Hurricane Katrina grows in fury and intensity in the Gulf of Mexico, one after another, an Army Corps of Engineers disaster response team member turns away from the television and exchanges that knowing look with a loved one sharing the family room. 

They know that taskers will flow soon from emergency ops centers, armed with amazingly savvy predictions of needed first-response resources. 

“Where did I stash those steel-toed boots from the last deployment?” 

Even before landfall, Allen Morse, debris subject matter expert at Corps Headquarters in Washington D. C., and modeling expert Terry Siemsen of the Louisville District Emergency Management GIS –Geographic Information System - group, are studying a model to predict th level of damage and Corps response.  That estimate will eventually result in a 1700-person volunteer team in Mississippi and 1300 in Louisiana.  Mistakes cost millions if not billions. 

“The models are based on GIS.  We create polygons of expected wind bands and rainfall and often also storm surge and expected power outages,” Siemsen said. Polygons are used in computer graphics to compose images that are three-dimensional in appearance. “The software then calculates debris, possible ice and water commodities, temporary roofs and temporary housing based on databases previously built at a census tract level of detail.”

Census tracts were chosen as counties were considered too "coarse" of a unit for estimating the storm effects. Census tracts are much smaller, typically an area that includes 4,000 to 10,000 persons. 

In earlier storms, debris estimates were only within 30 percent accuracy, Morse said, but in some cases the model has proved to be within 15 percent.  The goal is to continue tweaking and modifying the model to narrowed the gap to 10 to 12 percent. 

Not bad for an up-front estimate. The experts studied their models and predicted 27 million cubic yards of debris in Louisiana back in August 2005. Nearly two years later, the mission tapped out a 28.5 million, raised by the addition of unforeseen minor non-traditional missions, like 50 million pounds of rotting meat in New Orleans warehouses. 

“As the storm comes in, we use the planning model and look at wind fields and population density typical for the Gulf Coast,” Morse said. “The model estimates cubic yards of curbside debris per household for each category hurricane. As the category goes up, the cubic yardage per household goes up, e.g. for instance a category 1 is 3 cubic yards per household and a category 5 jumps dramatically 100.  A whole house with contents is about 225 cubic yards. 

Morse explained that the quantities of debris shown by the model consider the effect of wind damage, and do not include flooding. Commercial density and vegetative cover also become part of the equation. 

“We had to superimpose flooding and later demolition of houses; afterward we added the impact of thousands of saltwater-killed trees,” he said. “All these factors gave us good footing for planning and what we’d need for the initial response. We plug in more details, provided by from boots on the ground, as time goes by.” 

With each storm event, information on the number of persons and the number of households that may be within the envelope of hurricane force winds will also be provided by the model, Siemsen said.  

Commercial density database is influenced by population and household data - more people require or demand more services and associated facilities. This database is used in the modeling effort to account for debris that is likely to be generated from non-residential sources.  

Additional databases that are used in the event modeling process include critical facilities such as hospitals, police stations, fire stations, schools, potable water treatment plants and wastewater treatment plants.   

“The debris equations consider five primary factors - number of households, vegetation density, commercial density, storm wind intensity, and rainfall intensity,” Siemsen said. “The initial work with these equations was done at county-level and was considered a basis for determining possible amounts of debris to be handled in a clean-up effort and to estimate the number of debris reduction sites that may be needed following a hurricane landfall.” 

The volume estimated is a total amount of debris from a storm and not necessarily the amount of debris that will require clean-up with federal funds, he added. Nor does the model account for debris that might result from flooding caused by storm-related rainfall.  

Besides debris estimates, Siemsen said, needs for quantities of three liters of water and eight pounds of ice per person per day are built into the model program code. Also part of the model code are the typical truckload quantities of each commodity, 18,000 liters of water and 40,000 pounds of ice. 

Newer models can even predict the need for temporary housing and blue roofs. The models are intended to provide the mission response teams with a sense of the “scale and scope” of the potential action that may be required, he said.   

As hurricanes have revealed the patterns of their behaviors, the ever-clearer historical realities help refine the modeling process for each storm event enabling planners, with eyebrow-raising accuracy, to fulfill what some folks may regard as near-apocalyptic prophecies of impending doom. 

 The finely tuned estimated quantities become forecasts relied upon by capable Corps-monitored suppliers, trucks and boots on the ground worn by debris-hauling souls and managers who marvel at the pinpoint precision of the matching numbers.