SR SSD 99-20

Technical Attachment


David Andra
Dan Smith (ed.)
NWS Southern Region Headquarters

In the wake of the May 3, 1999, outbreak of violent tornadoes in Oklahoma there has been considerable interest concerning the performance and application of both the AWIPS and WDSS systems during the event. The majority of the AWIPS system design and software applications were developed by the NOAA/ERL Forecast Systems Laboratory, with the contractor, PRC, responsible for integration and fielding of the final system. The purpose of WDSS is obvious from its name (Warning Decision Support System), and it was developed by NSSL, in conjunction with the WSR-88D Operational Support Facility in Norman, to provide a test-bed for new radar algorithms, many of which will find their way into future AWIPS software loads. The philosophy at NWSO Norman has been to use the best attributes of both systems and to find ways to integrate and customize both systems to optimally support operations.

Attributes of AWIPS include access to all three WSR-88D radars associated with the NWSFO, integrated warning generation software (Warngen), integration of other meteorological data sets, six full-function workstations - each with dual graphics and an X-terminal for text generation, easily developed procedure macros, very good time-lapse functionality, and automatic updates of all data sets.

WDSS, on the other hand, offers access to the full wide-band suite of radar reflectivity and velocity data, improved algorithms to facilitate warning decision guidance, and dynamic tables which automatically rank storms according to their algorithm-derived severe weather threats.

Experience at NWSFO Norman leads us to believe that now is the time to ensure AWIPS planning takes full advantage of the arrival of the NEXRAD Open RPG, and that the strengths of the WDSS system make their way into AWIPS at all sites. Some basic WDSS functionality is already under development for AWIPS as part of SCAN (the System for Convection Analysis and Nowcasting), we believe it is already clear this is only a first step toward incorporating full-fledged WDSS functionality. Following are specific comments and observations based on experience gained during the May 3 outbreak.


1. AWIPS was critical to the success of the May 3rd warning operations. Already, it is hard to imagine working an event of this magnitude using only two PUPS and a couple of PCs connected to AFOS...which was normal operations prior to AWIPS.

2. The new notification server process performed flawlessly and greatly aided our ability to keep up with three WSR-88Ds and a rapidly evolving storm situation.

3. Only one ICP reset was required, and it was performed by the NCF (they were previously advised of impending critical weather and were poised to act when the call was made).

4. It is unlikely we could have provided the rapid flow of information and warnings without the use of Warngen. At one point, severe weather statements and tornado warnings were being composed simultaneously at four of the six AWIPS workstations. While Warngen was very successful in this instance, that was in large part due to substantial customization to templates, modification to the cities info file, and the creation of detailed municipal boundary maps; all accomplished by the local staff after AWIPS delivery. Without this customization, considerable hand edits and delays would have incurred.

5. The warning forecasters relied on locally developed procedure macros to load data and set up displays for stormscale analysis and Warngen use. Based on this event, and others in the recent past, we strongly advocate forecasters be trained to develop and use procedures for warning operations. Displays which require two to three minutes to establish using menu options can be loaded in a few seconds with procedures!

6. Between about 6 pm and 10 pm telephone communications were unreliable and at times no outgoing calls could be made. This points to a potentially serious shortcoming in the present NCF support system. Had we required emergency assistance with AWIPS, we could not have reached the NCF for help. An alternate means for reaching the NCF in emergency situations is needed.


1. The dynamic cell/meso/tornado tables allowed forecasters to maintain an overview of the outbreak and helped enhance confidence that forecasters who were basing their warning decisions on AWIPS capabilities were in fact seeing all they needed to see, and that no severe storms were being overlooked.

2. WDSS allowed forecasters to see the true intensity of the signatures (not limited by 16-level quantization or pixel replication distortions), and at the full spatial resolution of the wideband data. In a couple of instances we were able to confidently track TVS with delta velocity values greater than 210 kt across adjacent gates! This increased confidence in issuing strongly worded statements and warnings for storms which occurred in rural areas after dark.

3. The WDSS at NWSFO Norman runs on an HP-715 dating from 1995, and it is too slow to serve as a stand-alone platform for severe weather operations, especially when we do an 'xhost' second session from another workstation.

4. Although the wideband-3 connection remained solid for this event, this is often not the case and we remain concerned about the reliability of this connection.


1. During this event the NWSFO employed sectored warning operations. This method of operation allows us to use multiple warning forecasters, each assigned a subset of the CWA and an AWIPS workstation, based upon the meteorology at hand and that expected over the next one or two hours. Sectors are arbitrarily defined using available map backgrounds and reassigned as the threat moves or changes character. On May 3 each warning forecaster was responsible for understanding the mesoscale environment, monitoring stormscale evolution, making decisions, and issuing warning products for their subsection of the CWA. Our experience from previous years using DARE workstations at Norman, and now with AWIPS, suggests this method of operation is more efficient and intuitive than our previous method of assigning warning forecasters by radar/PUP, etc.