Using information from maximum historical flooding events in floodplain management

Portions of Southeast Texas experienced severe flooding from Tropical Storm Allison in 2001. Copyright Harris County Flood Control District. Reproduced with permission.

Extreme rainfall events occur in the United States every year, bringing high water that inundates large parcels of land—damaging crops, destroying livestock, disrupting transportation and commerce, and flooding homes and other structures. In the worst cases, human lives are lost in rain-swollen streams.

Engineers study these major storms to gain information that may be used to protect the public from flooding. Such studies are used to develop relationships between the amount of rain that falls, the duration of the rainfall, and the probability of flooding associated with it. These depth-duration-probability relationships are used to design storm sewers, drainage channels, and other critical facilities. They are also used in studies that establish flood elevations, the extent of a floodplain, and minimum finished floor elevations for insurable structures.

In developing depth-duration-probability relationships from historical data, individual storms are identified, ranked, and passed through a statistical analysis. Larger, less frequent storms are associated with lower probabilities, and vice versa. Extreme storms, the "big ones," are mixed with other storms in completing these statistical analyses. In some cases, the storm identified as the 1-percent design event is less severe than one or more historical storms because those extreme storms are judged to have a probability of occurrence of less than 1 percent in any given year. Thus, historical events, and the lessons that they contain about extreme weather conditions, are often not directly used in floodplain management.

This raises a number of questions. Are these extreme storms so rare that we can comfortably assume that similar storms will never again occur? Should these events be recognized and used? Should the public be educated regarding these storms and the flooding they caused? What about prospective buyers of homes and property in the areas where they have occurred? Is it the duty of floodplain managers to disclose information about these storms?

We must face the fact that floods more severe than the 1-percent design event can and will occur, and very severe storms can and will occur more than once in a particular area. These storms can be important in defining the potential magnitude of flooding in a given location, especially when designing critical facilities such as levees. Civil engineers should adopt an attitude of "never say never" in connection with floodplain management and public education.

Current flood evaluation procedures
Floodplain management data is largely based on U.S. Geological Survey (USGS) streamflow gage data and published Flood Insurance Studies. The USGS maintains streamflow gaging stations at many locations around the United States. Data collected at those gaging stations includes flow depth versus time. Those flow depths are used with rating curves (the relationship between stage, or flooding depth, and flow rate) to produce storm hydrographs and peak flow rates for individual storms, as well as average data over long periods of time.

Flood Insurance Studies provide information on 10-year, 50-year, 100-year, and 500-year storms, which have annual probabilities of 10 percent, 2 percent, 1 percent, and 0.2 percent, respectively. A Flood Insurance Study includes a written report and a set of Flood Insurance Rate Maps that illustrate flood plain boundaries for the 1-percent and 0.2-percent storms, base flood elevations for the 1-percent event, and floodway boundaries based on the 1-percent storm.

Modeling studies are typically completed in connection with Flood Insurance Studies and updates to those studies. In some cases, statistical analyses relating flow rate to frequency of occurrence are used to produce flow rates for use in hydraulic modeling studies. However, since individual historical storms are not typically analyzed in connection with Flood Insurance Studies, the studies and maps do not give any real idea of the severity of record historical storms. In fact, they do not indicate in any way that such events ever occurred.

This leaves a prospective property owner or homeowner with little or no information on the magnitude of "maximum historical events" for local streams. He or she will need another source of information to evaluate a property under a maximum historical event scenario.

Figure 1: A portion of a flood tracking chart compiled by the U.S. Geological Survey for the upper San Jacinto River watershed north of Houston

USGS flood tracking data
The USGS provides some historical storm data for use in evaluating the likelihood and severity of flooding in a particular location. Figure 1 represents a portion of a flood tracking chart compiled by the USGS for the upper San Jacinto River watershed north of Houston. The original chart lists maximum flood stages for a number of historical events and provides information that may be used to convert the maximum stage values to elevations. Table 1 summarizes the data provided on the tracking chart for Spring Creek, a tributary to the West Fork of the San Jacinto River.

The USGS indicates on the flood tracking chart that the information is intended to allow a property to be evaluated in terms of potential inundation by flood waters. This is a valid and useful application of historical storm data. While the standard Flood Insurance Rate Map provides flooding data based on a storm-probability approach, information from this type of flood tracking chart allows an evaluation based on historical events. However, the maximum historical flood evaluation is based on limited data (at a limited number of discrete points along a channel or within a watershed) and is therefore difficult to apply to locations that are not in close proximity to a gage location.

A more meaningful evaluation
Obviously, the use of such limited information as that provided on a USGS Flood Tracking Chart will yield meaningful results for a relatively small number of locations. A better approach involves a combination of historical storm data and the data illustrated on Flood Insurance Rate Maps. To fully evaluate the extent and magnitude of a historical flood event, a prospective property owner needs both a Flood Insurance Rate Map and historical data such as that provided on a USGS Flood Tracking Chart—information that is rarely available.

As an example, consider the storm that occurred over the Spring Creek watershed in Montgomery and Harris Counties, Texas, in October 1994. That storm created flood levels approximately 5 feet higher than 100-year levels shown on the official Flood Insurance Rate Maps. Those flood levels are the highest ever recorded along Spring Creek. To evaluate a given location (say one in the vicinity of the I-45 crossing of Spring Creek) against the maximum historical event, locate the point or area of interest, determine the 100-year flood level from the Flood Insurance Rate Map, and add 5 feet. This approximates the maximum elevation reached by flood waters during the October 1994 event.

Figure 2: Flood Insurance Rate Map for a portion of the Spring Creek watershed in Montgomery and Harris Counties, Texas

Figure 2 illustrates floodplain data for a portion of the Spring Creek watershed near the existing gage location at I-45. For a sample application of the suggested procedure, consider the area outlined on Figure 2 with a red rectangle. The Flood Insurance Rate Map indicates a 100-year flood elevation of 100.8 feet in the central portion of that area. Adding 5 feet to this value yields a maximum historical flood level of approximately 105.8 feet. The use of this elevation along with an approximate ground elevation allows establishment of a historical flooding depth at the selected site.

Figure 3 illustrates ground elevation contours developed for the area of interest from LIDAR data, which is collected using airborne laser ranging equipment, as well as the 100-year floodplain (shown in blue) and the approximate extent of flooding associated with the 1994 storm (shown in green).

Figure 3: Ground elevation contours developed from LIDAR data, the 100-year floodplain (shown in blue), and approximate extent of flooding associated with the 1994 storm (shown in green)

Using the LIDAR ground contours, a ground elevation of 102.0 feet is estimated as the average natural ground elevation in the area of interest. This puts the flooding depth for the maximum historical storm event at 105.8 feet—102.0 feet = 3.8 feet. Knowledge of this depth allows a prospective property owner to determine the height to which a home or other structure would have to be elevated to have remained dry during the maximum historical flood event in the area.

It is imperative that floodplain managers, engineers, and others tasked with protecting the public interest use historical data to evaluate hazards posed by extreme floods and make that information available to the public, especially when flooding may be accompanied by loss of life or damage to critical facilities. In light of recent events such as the Mississippi River flooding in Illinois and Missouri during the summer of 2008 and the tragedy brought to the city of New Orleans by Hurricane Katrina in 2005, such improvements in public safety should be pursued diligently and through all possible means.

Duane Barrett, P.E., heads the Storm Water Management Department at R.G. Miller Engineers, Inc., in Houston. He can be contacted at dbarrett@rgmiller.com.