Mastering Disaster -- Earthquake Modeling Piloted

Want to know what would be waiting for you after an earthquake rolled through you hometown during a rainstorm at 4:24 a.m.?

by / September 30, 1997
If you shake a stubborn bottle of catsup hard enough, its contents will loosen up and pour. During an earthquake, the same thing can happen to otherwise solid ground. The process is called "liquefaction," and when it occurs during an earthquake, buildings can shift or sink, and underground storage tanks -- like those at the local gas station -- can float to the surface. Unfortunately, the pipelines attached to those tanks float at a different rate, causing line breaks that may lead to fires or flooding.

In many cases, collateral damage may be more destructive than the quake itself: fires, flooding, loss of vital services or destruction of key structures -- such as hospitals, highways or fire stations -- can dramatically increase the total losses associated with an earthquake.

The tremendous variety of possible damage and the influence of local circumstances have traditionally made it difficult for community leaders to predict expected damage levels and plan emergency responses to earthquakes. Perhaps more important, the lack of effective risk analysis has made it difficult to bring public attention to bear on legislative, zoning and building code changes that could help mitigate the effects of earthquakes. And, of course, all these local and state impediments to reliable estimates roll up to the federal level, making it all but impossible for federal authorities to realistically estimate the national earthquake risk.

In Search of a Standard

Congress began to address some of these issues in October 1977 with the passage of the Earthquake Hazards Reduction Act. Its purpose is to reduce "the risks to life and property from future earthquakes in the United States through the establishment and maintenance of an effective earthquake hazards reduction program." The act established the National Earthquake Hazards Reduction Program (NEHRP), which coordinates activities between various federal agencies, including the Federal Emergency Management Agency (FEMA). In the early 1990s, under the auspices of NEHRP, FEMA partnered with the National Institutes of Building Science (NIBS) to develop a standardized methodology for estimating earthquake damage.

"We started in about 1992 to develop an earthquake loss methodology," said Claire Drury, FEMA project officer for what is known as the HAZUS Loss Estimation Methodology Project. "We found that people weren't aware of potential losses and you can't promote seismic building codes unless people realize there is a benefit to doing this. So what we have done is build the model that will be used by regional and local governments to run loss estimates to help them recognize the potential hazards."

The project, directed by NIBS, established an eight-member Project Work Group (PWG) -- consisting of earthquake experts -- and an 18-member project oversight committee (POC), which represented user interests in the earthquake community. Additional assistance was elicited from over 80 corresponding members of the POC, whose views represented user and technical interests.

In 1993, PWG and POC defined the components of the loss estimation methodology, prepared an extensive set of objectives for developing the methodology, and generated a standardized list of earthquake-caused economic and social losses as methodology outputs.

Risk Management Solutions (RMS) of Menlo Park, Calif. -- a company that specializes in providing software and information services to the insurance and financial industry -- was selected to develop the methodology and was also hired to do the software implementation.

"The HAZUS methodology itself is a nationally applicable methodology," said Scott Lawson, an associate vice president at RMS with a Ph.D. in structural engineering who headed RMS's part of HAZUS. "The information that was gathered was on a national scale -- data like building inventory information, census data and the EPA-supplied toxic-site information. We have information on highways, bridges and dams.

"FEMA had collected a whole bunch of information during the civil defense era," Lawson continued, "and we took all that information and put it in there -- all the data you need for dealing with a nuclear disaster, you also need for dealing with an earthquake disaster. There is a wealth of other data, such as FEMA flood maps and U.S. Geological Survey land use and land cover information. We also have probabilistic ground-shaking mappings, information on elevation levels and hurricane data."

What RMS was studying were the mechanics behind how to theorize an earthquake in a particular location, and understand the impacts on the local community, region or state.

In Lawson's opinion, HAZUS' biggest contributions to the art of loss prediction are the way it analyzes and estimates building damage and its ability to estimate indirect economic effects. For example, it can provide economic "snapshots" of a region one to 15 years after an earthquake.

As described in a February 1997 paper by Robert V. Whitman, professor emeritus at MIT, and Henry J. Lagorio, professor emeritus at the University of California at Berkeley, the HAZUS methodology consists of three elements:

* Classification systems for assembling information on the building stock, the components of highway and utility lifelines, and demographic and economic data.

* Methods for evaluating damage and calculating various losses.

* Databases containing information usable for calculations and as default data.


These three components are implemented in the software package developed by RMS, which is based on MapInfo GIS (an ArcView version is in the offing). Part of the development process included two pilot studies -- one in Portland, Ore., the other in Boston. Mei Mei Wang, director of Earthquake Programs for the Oregon Department of Geology and Mineral Industries was involved in supplying information to the Portland pilot and was impressed with what she saw.

"I give it five stars, but you have to use your judgment," said Wang. "The further out you are [the broader the scope of the estimate], the better the results. When I was at the training session, I made the state of Oregon a study area. The program built all the census tracts in the state and figured out the building worth at $160 billion [not including bridges and highways], which is the right order of magnitude. But in a smaller area, my uncertainty gets bigger and bigger."

This uncertainty is particularly acute when only default information is used. While default-based estimates have a place, the full methodology calls for customizing the data to get more accurate estimates. In the Portland pilot, localization of much of the data was done to raise the level of confidence in HAZUS' results. And even though Portland is not generally considered a likely victim of earthquakes, it is subject to an interesting and complex set of geological and environmental circumstances that needed to be customized -- beginning with the buildings.

"In Portland, what was really important was categorizing the building stock, because the national database did not accurately portray the actual stock at all, so it would have underestimated the hazard," said Wang. "The buildings are worse than you would expect with many URM [unreinforced masonry] buildings."

URM buildings are particularly susceptible to earthquakes and are a key factor affecting how well or poorly a metropolitan area will fare during an earthquake. To improve the Portland estimate, a complete building inventory was done using a technique known as the FEMA 154 method. This is an easily learned technique that involves a 10-minute evaluation of each building in a locale. Portland State University Civil Engineering Department students were enlisted to help and spent several months combing the city, evaluating each building. That information was entered into the HAZUS databases and was used to improve the evaluation.

"Another important part of the Portland pilot was soil type," said Wang. "We have the Portland Hills, which tend to have landslide problems, and we have the Williamette and Columbia Rivers. The Williamette runs through downtown Portland, and the areas around the rivers would be subject to liquefaction and lateral spreading during an earthquake."

There are seven bridges in the greater Portland area and although HAZUS does not include bridge model capability, these would be strongly affected by an earthquake. It is common for bridge supports on each shore to move toward the center of a river during an earthquake due to liquefaction of the supporting soil. This would almost certainly be a problem in Portland and, according to Wang, the Portland bridges would be "a wreck."

"When you look at some of these old bridges you say 'that is how you would not design a bridge to withstand earthquakes,'" said Wang. "For example, one of the Portland bridges has big concrete counterweights, and when you lower the counterweights, it raises the bridge. In an earthquake, you don't want structures weighted at the top."

Portland at Fault

Geologically, Portland provides some other challenges. "There is a local fault -- the Portland Hills fault, which runs at the base of a hill very close to downtown," noted Wang. "It's a big fault structure and it's a young fault; and young faults are a concern because they might be triggered. With an earthquake there, you would have landslides in the Portland Hills."

However, that isn't the only fault potentially affecting the city. The Cascadia Subduction Zone exists off the Oregon shore. The Portland Hills fault occurs where two geologic plates are moving sideways past each other. In the Cascadia Subduction Zone, the Juan de Fuca plate in the Pacific is sliding underneath the North American plate. Although subduction zones such as this generate earthquakes less frequently, they have the potential of generating even bigger quakes than large lateral faults, such as California's San Andreas.

Because the Cascadia Subduction Zone is offshore and would tend to produce a vertical motion, tsunamis (large ocean waves) would be a likely effect. This wouldn't affect Portland itself, but could be devastating to the Oregon coast. However, subduction zone faults tend to produce many low frequency waves that travel further than high frequency waves. Such low frequency waves impact buildings 10 stories and higher because the buildings themselves resonate with the low frequency waves. Therefore, despite the distance from the Cascadia Subduction Zone to Portland, the taller Portland buildings are likely to be strongly affected by any major Cascadia earthquake.

Evaluating the Risk

With two major faults, the Portland pilot modeled the effects of both a local earthquake in the Portland Hills -- which would tend to generate higher frequency waves and cause landslides -- and a more remote earthquake in the Cascadia Subduction zone -- which would tend to affect the taller buildings. With the data entered, HAZUS evaluated two scenarios -- one, a magnitude 6.5 earthquake in the Portland Hills, the other, a magnitude 8.5 earthquake in the Cascadia Subduction Zone. The HAZUS estimate determined that the Portland Hills quake would produce the greater damage, although both would have a serious impact.

Because HAZUS' accuracy depends on the quality of the data, as well as the sophistication of the algorithms, it is generally believed that the estimates will improve over time as both data and algorithms improve. Nonetheless, the results to date have been quite positive.

"FEMA is very pleased with it," said Drury. "HAZUS gives us a standardized methodology that is applicable nationwide. Localities have done studies in the past, but they were done only if a community knew it had a risk."

Communities that didn't perceive a risk usually haven't invested the time or money in doing studies. What's more, because there was no standardized methodology, one community's studies weren't generally comparable to others. These factors contributed to the difficulty of making useful national estimates.

HAZUS was designed to overcome these problems and so, once the methodology was out of the pilot stage, FEMA began to distribute the software to state and local communities. The software was sent on CD-ROM, one for the East Coast and one for the West Coast. Each CD contains the default level building inventory data, as well as various national databases, and is available for free to state and local officials through NIBS. FEMA has also provided training seminars to help communities get their estimates started. In all, 45 states have had representatives trained on HAZUS.

Various communities have started their own estimate projects, but it may be some time before their results become generally available. In the meanwhile, FEMA is doing an annualized loss estimate for the United States by running estimates on individual counties and using probabilistic capabilities within the model to make generalized conclusions for the whole country.

Earthquake loss estimation is only the beginning for HAZUS. The intention all along was to expand the program to encompass other types of disasters, and work is already under way on these other components.

"We plan to expand the methodology to include wind and flood," said Drury, "and we have panels of experts similar to the earthquake panel at work on it. It will take a couple of years to complete -- we're still very much at the seminal periods for those projects."

As RMS' Lawson said, "If you build a robust enough data environment, then you can overlay disasters and assess the damages. You can throw an earthquake at it, or a flood, and see how it reacts to the disaster itself."

Such a tool gives planners the opportunity to catch a glimpse in advance of what they may face during a full-blown disaster. By understanding the possible effects, communities and their leaders have the opportunity to mitigate the effects of disasters, and that helps everyone.

October Table of Contents
David Aden
David Aden is a writer from Washington, D.C.