Contact:  Randee Exler              FOR IMMEDIATE RELEASE
                                    5/23/96

NATIONAL WEATHER SERVICE ISSUES DISASTER SURVEY SUMMARY

The National Weather Service today released a summary of the disaster survey team report reviewing weather service performance for the April 21 storm system that spawned a tornado which struck Fort Smith and Van Buren, Ark. The survey team results were released at a Fort Smith news conference today along with the announcement of steps that will be taken in response to the teamūs findings.

The Tulsa, Okla., National Weather Service Forecast Office, over a two-hour period, tracked four tornadoes during the storm: two in Oklahoma and two in Arkansas. Using WSR-88D radar data the warning team in Tulsa issued tornado warnings for all four tornadoes: two with 30-minute lead times, one with a 25-minute lead time and one with a four-minute lead time.

The survey teamūs results show that as the storm neared Fort Smith, forecasters who were relying on one particular WSR-88D radar product believed the storm, while still dangerous, was unlikely to spawn tornadoes. Based on that analysis, the Tulsa forecasters issued a severe thunderstorm warning rather than a tornado warning 16 minutes before the onset of the tornado. The Tulsa Forecast office upgraded that warning to a tornado warning four minutes before the tornado touched down in Fort Smith at 11:12 p.m.

Communication line failures due to power disruptions resulting from the storm prevented this warning from reaching local emergency service officials. As a result, the Civil Defense sirens were not activated. Subsequent analysis of the data available during the storm indicates the signature of a tornado was present.

The survey team concluded that three factors influenced the decisions made by the forecasters:

 

     1) Information from one particular radar product, which was
     interpreted as showing a decrease in the intensity of the
     storm; 
     2) inconsistencies between spotter reports of funnel cloud
     locations and radar data; and 
     3) a lack of damage reports from areas where the storm had
     already passed.

     Based on these findings, the survey team recommends: 

     1) forecasters should issue warnings based on a variety of
     reports and products, not just a particular radar product;
     2) forecasters should judiciously question and clarify
     spotter reports that are not consistent with other available
     data; and 
     3) forecasters should not rely as heavily on damage reports
     when other information is available.

The NWS will immediately begin working with local officials in the Fort Smith area to mitigate the impact of future severe weather occurrences. As a result of the findings, the NWS will continue its training program for forecasters nationwide to ensure that a high level of understanding of radar products and usage of them in severe weather situations is achieved.

In addition, the NWS will work to improve coordination efforts with local officials, and will work with the community to identify alternative methods to improve the communication of critical weather information.

"The Tulsa forecast office issued numerous watches, forecasts and warnings throughout the afternoon and evening of the event, which alerted citizens throughout eastern Oklahoma and western Arkansas to the potential of severe weather including the threat of tornados," said Jim Belville, disaster survey team leader and director of the weather service's NEXRAD Operational Support Facility in Norman, Okla. "This coupled with the responsiveness of the media and a generally aware public, without doubt, saved lives."

During this event, personnel in the Fort Smith Weather Service Office interacted with emergency management and law enforcement officials relaying critical warning information generated by the Tulsa Forecast Office. Nevertheless, Fort Smith employees expressed some confusion over their role in the warning process during severe weather events. The team recommends that the NWS work to clarify these roles, particularly in instances where county warning responsibilities have been transferred.

The same storm system that spawned the Fort Smith/Van Buren tornado produced three other tornados between 9:27 p.m. and 12:15 a.m. as the storm tracked across Oklahoma and Arkansas. The Tulsa Weather Forecast Office provided 25 - 30 minute warning lead times before the onset of these three other tornados. The national average lead time for tornado warnings is eight minutes.

From 1986 to 1990 (before installation in Tulsa of the WSR- 88D radar), the average lead time for a tornado warning was 1.2 minutes in the seven-county area including Fort Smith. Since July 1994, when Tulsa took over the warning responsibility for the same seven-county area, the average lead time for tornado warnings has been 18.6 minutes.

The variety of radar products that the survey team reviewed demonstrated that the Tulsa WSR-88D radar detected the supercell thunderstorm throughout its life cycle. The path of the storm that spawned this event was within the range of radar coverage provided by the Tulsa NWS office. An area of degraded radar coverage to the south of Fort Smith, separate from the area where this tornado struck, was identified in an October 1995 report to Congress by the Secretary of Commerce, and will result in an additional radar for that area.

                               ###