Theoretical models help determine flood risk by studying which conditions would alter water levels and to what degree, said Sumner.
"In other words, the intensity of a storm, the duration of a storm -- all those things that contribute to extraordinary water flow -- would be placed into a model, and that model would determine which areas would be flooded under that particular condition," Sumner said.
This is what FEMA does, he added. It builds scenarios using topographic data, and determines possible outcomes. For this reason, FEMA will be involved in floodplain assessment across the country, even though the topographic base maps are coming from local sources.
Less Room for Error
Turning to local governments enables FEMA to utilize larger-scale data with a lower error rate, which improves accuracy when determining flood zones, said Tim Scharff, Cobb County's GIS manager.
GIS map scales relate distance on the map to distance on the ground. The concept is similar to taking a photo: the farther away a photographer is from the subject, the less detail the picture contains, even though more area is captured. A small scale of 1:100,000 -- where one unit on the map represents 100,000 units on the Earth's surface -- would have a larger acceptable error rate than a large scale of 1:1,200, because detail is sacrificed to cover more ground.
"On a scale of 1:100,000, data is going to be tens or hundreds of feet off," said Scharff. "So you can see how quickly that could be the size of a lot [of land] or more." He also said FEMA's maps suffer from small-scale error rates.
"You take a look at the scale developed for FEMA and you begin to question if that should be used for a particular house as opposed to a broader area," said Scharff.
Localized maps are more conducive to large-scale accuracy, something that Scharff said can further FEMA's attempt to produce accurate floodplain maps.
"Let's say this has the potential to allow FEMA to map at a different scale for this county because we've got some nice elevation data layer developed at a scale that's larger than what they typically develop data in," Scharff said. "As long as whatever else they combine it with isn't at a much smaller scale -- because the strength of any map is only as good as its weakest link -- then they can get larger-scale data that provides more detail."