U.S. Department of Commerce
William M. Daley, Secretary

National Oceanic and Atmospheric Administration
D. James Baker, Administrator

National Weather Service
John J. Kelly, Jr., Assistant Administrator

National Weather Service, Southern Region Headquarters
X. William Proenza, Acting Director

May 1998

A tornado struck parts of northeast Georgia early Friday morning, March 20, 1998. The National Weather Service's Southern Region Headquarters formed a Regional Service Assessment Team to review the operations and effectiveness of products and services of NWSFO Peachtree City (Atlanta) associated with the event. The Assessment Team consisted of (Team Leader) Steven Cooper, Acting Chief of Meteorological Services, Southern Region Headquarters, Fort Worth, Texas; Steve Letro, Meteorologist-in-Charge, NWSO Jacksonville, Florida; Jim Ward, Science and Operations Officer, NWSFO Austin/San Antonio, Texas; and Jim Butch, Warning and Coordination Meteorologist, NWSFO Jackson, Mississippi.

We express our sincere appreciation to the Team for their quick response in documenting and preparing this Assessment report. We also wish to thank the staff at NWSFO Peachtree City who assisted with gathering data and providing other information via personal interviews. Appreciation is also extended to media interviewees, Emergency Management Officials and to those individuals interviewed who live or work in the Hall and White Counties area.

X. William Proenza
Acting Director, Southern Region Headquarters

May 1998

Event Summary

On the morning of March 20, 1998, a tornado developed rapidly in western portions of Hall County, Georgia and moved quickly through northern Hall into southern White County. The tornado traveled approximately 11.5 miles through the hilly terrain of North Georgia. Damage on the Fujita scale ranged from F2 (113-157 mph) to F3 (158-206 mph). The initial touchdown was about six and one-half miles north-northwest of Gainesville, just west of State Highway 60. The tornado's path width was up to one-quarter of a mile as it raced northeastward around 55 mph. The last damage was noted about 16 miles north of Gainesville or about three miles northeast of the US Highway 129/White-Hall County line. In its wake 12 fatalities were directly related to the tornado. Eleven of these were in Hall County with the remaining one in White County. Eleven of the twelve deaths occurred in mobile homes. The twelfth occurred in a truck that was tossed off the road. One additional indirect fatality occurred when a sheriff's deputy stopped to assist traffic control and was struck by a passing vehicle. Hall County officials reported 96 injuries, while White County officials reported 75. In Hall County, thirty homes were destroyed with 105 damaged. Thirty mobile homes were destroyed with another fourteen damaged. Officials from Hall County estimated approximately $13.5 million in damage. White County officials estimated $1.7 million damage with twenty-two homes destroyed and nine damaged. Thirteen mobile homes were destroyed with nine damaged.

In response to the tragic event, the Southern Region Headquarters assembled a Service Assessment Team to review the actions, products, and services delivered by the local NWS office located in Peachtree City, Georgia. The Team assembled at the Peachtree City office Monday afternoon, March 23, through midday Tuesday, March 24, and reviewed forecast and observed data, WSR-88D data, and met with four of the individuals working the March 20 event. The Team traveled to Gainesville Tuesday afternoon and spent Wednesday visiting with local Emergency Management officials in Hall and White Counties. Officials provided a very detailed tour of the damaged areas allowing Team members to stop and talk with individuals struck by the storm. Most of the victims reported seeing lightning and hearing thunder, a sudden burst of very heavy rain, and a quick onset of the tornado. The typical "freight train" sound was mentioned with one individual describing the sound as a "bomb explosion" that didn't end until the tornado passed. There were many stories of families hearing the storm and running to their bathrooms. Some survivors were swept away by the winds while others described holding onto their toilet or bathtub, only to discover it was the only portion of the home remaining above the flooring of their home. As described earlier, except for the indirect traffic fatality and the individual tossed into the school while driving along Highway 60, all victims were in mobile homes. No one in their frame/foundation homes lost their lives - even in those homes that suffered total destruction!

One of the main concerns about this tragedy was the lack of a warning from the Peachtree City office prior to the onset of the tornado. The Team reviewed the forecast and observation data. A compact vigorous upper level low pressure system was moving eastward from northern Alabama into northern Georgia. The wind profiles and instabilities (both observed and forecast) indicated the potential for severe thunderstorms with wind and hail as the greatest threat. Because of this potential, the Storm Prediction Center (SPC) in Norman, Oklahoma, had issued a Severe Thunderstorm Watch that included northeast Georgia with a valid time from 12:45 a.m. until 6:00 a.m. (all times Eastern Standard Time, unless otherwise indicated). The SPC Mesoscale Convective Discussion issued at 4:48 a.m. agreed with the trends indicated by satellite and radar that the severe weather activity which had moved into Georgia from Alabama was decreasing significantly in intensity, and the Watch covering Hall and White Counties would be allowed to expire at 6:00 a.m. The forecast staff from Peachtree City agreed with this assessment. As a result, several required updates for the removal of the watch information occurred between six and seven o'clock including the Zone Forecast Product, the State Forecast Product, and the Thunderstorm Outlook for the area. These updates were in addition to the regularly scheduled products such as the Terminal Forecasts (TAFs).

An additional forecaster arrived at 2:15 a.m. to assume WSR-88D duties. He relieved a forecaster held over from the evening shift. During his shift, Severe Thunderstorm Warnings were issued for three counties in west and northwest Georgia. These were based on radar data that had been identified as a mesoscale cyclone (meso). None of these warnings verified, as only pea or one-half inch sized hail was observed. Several hours passed until the next algorithm derived meso alarmed around 6:26 a.m. (from the 6:21 a.m. WSR-88D volume scans). The radar operator investigated the storm and wanted additional continuity before issuing a Severe Thunderstorm Warning. The radar algorithm did not generate a meso alert for the 6:27 a.m. or 6:33 a.m. volume scans. The system did generate a meso alert (around 6:44 a.m.) for the volume scan that began at 6:39 a.m. This was about the same time the first reports of damage were being received in the Peachtree City office. A Severe Thunderstorm Warning was issued at 6:46 a.m. for Hall County. Due to the damage reports being received at the NWS office, a Tornado Warning was issued at 6:51 a.m. for White County.

Initial reports that the WSR-88D did not detect the storm were misleading. In fact, the storm was easily identified in both the reflectivity and storm relative velocity displays. The problems with providing a warning for this storm were the extremely rapid spin-up of the mesocyclone and the accelerated (55 mph) movement toward the northeast. Additionally, range folding near the area of the mesocyclone on the 6:15 a.m. volume scan may have inhibited the algorithm from defining a meso and alerting the radar operator around 6:21 a.m. The Team believes the radar performed as designed. As meteorologists and radar operators gain an increasing knowledge of these fast spin-up, low-level type storms, and as the WSR-88D algorithms and computing capabilities improve with time, it is possible that better lead times can be accomplished for these difficult storms.

OBSERVATIONS

Finding: The nearby ASOS's appeared to be operating well throughout the night and during the event. The ASOS site, Gilmer Memorial Airport, in Gainesville, is located very near the tornado path. There was a pressure falling rapidly/rising rapidly couplet at the time of the tornado, with very little if any wind response.

The tornado event likely was a low-level spin-up related to upward vertical motion near a surface boundary. The surface boundary may have resulted from previous rains that night. During the nighttime, without visible satellite imagery, small-scale boundaries are often difficult to detect. The surface observation network array is typically coarse, with nearby observations in Atlanta, Gainesville, and Athens. As more of these type tornadoes are documented, the need for a mesoscale observing array becomes more evident. SPC forecasters stated that some areas across the United States seem to favor tornado development, some likely orographically induced.

Recommendation: The Peachtree City Office should continue their efforts to work with State and other officials to incorporate current observing systems and look for resources to develop a statewide mesoscale observing network (such as the University of Georgia Environmental Network). Priority areas should be those that historically have had a recurrence of tornadoes, such as the Hall County area.

Finding: Frequent quality satellite imagery was available to the NWSFO forecasters. The NWSFO has RAMSDIS capabilities, and it was utilized that morning. The imagery did show increasing convection across the area of concern, but specific signatures of severe weather were not highly evident to assist in any needed tornado warning. Perhaps more frequent satellite imagery displayed with surface, radar, and other data on an AWIPS workstation would have provided a better assessment of the situation.

Recommendation: No action necessary.

Fact: From all evidence and according to the radar operator that morning, the WSR-88D radar at Peachtree City, Georgia, was operating well throughout the night and early morning of the event.

Fact: The Peachtree City, Georgia, WSR-88D was displaying a line of strong convection, associated with the cold front, moving from northeastern Alabama into northwestern Georgia between 0800 (3:00 a.m.) and 0900 (4:00 a.m.) UTC. Maximum reflectivities generally were in the 60-65 dBz range. The line of strong convection extended southward from a large cyclonic circulation pattern across northeast Alabama. Between 0800 (3:00 a.m.) and 1100 (6:00 a.m.) UTC the convective pattern moved eastward across much of the 125-nm radar scope. Severe Thunderstorm Warnings were issued for three counties as the VIL (vertically integrated liquid water content) readings reached 40 kg/m². These storms earlier in the period did exhibit comma shapes, but they did not persist. Rotational values remained weak or minimal in these earlier storms (in general 20 to 30 knots at 15 to 35 nm from the radar). No severe weather was reported from these storms of concern.

Between 1000 (5:00 a.m.) and 1100 (6:00 a.m.) UTC there was a general decrease in thunderstorm intensity and coverage. This general decrease was the main factor in SPC's allowing the Severe Thunderstorm Watch to expire without the issuance of another watch. Although there was a general decrease in thunderstorm coverage and intensity, particularly along the southern region of the line, a cluster of thunderstorms to the northwest of Atlanta persisted. Between 1000 (5:00 a.m.) and 1100 (6:00 a.m.) UTC this cluster began to take on a comma shape (cyclonic head and associated bow-type echo). The pattern was similar to the northern part of a mature squall line, with a developed cyclonic head. The cyclonic pattern was much smaller than most mature squall line patterns of this nature.

Between 1100 (6:00 a.m.) and 1115 (6:15 a.m.) UTC, the comma-shaped thunderstorm complex moved rapidly northeastward toward Hall County. However, only weak wind shears were evident in the circulation of the comma head. As the thunderstorm system moved into Hall County, stronger rotations became evident on the WSR-88D Storm Relative Velocity Maps, and VIL values increased slightly. Maximum reflectivity values varied around 60 dBz. The WSR-88D derived rotations continued across Hall and White Counties between 1115 (6:15 a.m.) and 1200 (7:00 a.m.) UTC.

Of particular note was the existence of inflow notches in the reflectivity. One inflow notch was from west to east into the comma head. A second inflow notch was from southeast to northwest into the comma head. The tornado occurred in the vicinity of the confluence of these two inflow notches.

Finding: The comma-shaped echo pattern evident on the Peachtree City WSR-88D is the type that has been documented as possible straight-line wind damaging or tornado-producing systems. The inflow notches also have been documented to be associated with damaging winds or tornadoes. Such structures have not been a part of major training efforts by the Agency in the past, compared to descending mesocyclones in supercells.

Recommendation: Although other indicators showed a diminishing threat of severe weather, the existence of a comma-shaped echo thunderstorm complex and the existence of reflectivity inflow notches should alert WSR-88D operators to thoroughly investigate any developing winds in the bow or rotations in or near the comma signature. Continuing education of WSR-88D operators regarding convective circulation and line structures should be high priority for all NWS offices.

Finding: The 11:09 UTC volume scan indicated a minimal rotation between 20 and 25 knots at 60 to 65 nm from the radar. The maximum unambiguous range (Rmax) was located around 80 nm and the velocity pattern with the storm was unobscured. The first significant rotations became evident on the WSR-88D on the 1115 (6:15 a.m.) UTC volume scan on the 1.5 degree level; however, the Auto PRF caused the Rmax to jump into the 63 nm ring area on the 1115 UTC, obscuring portions of the storm.

The Auto PRF is designed to minimize the range folding of velocity data, but sometimes at the expense of obscuring certain storms. This range-folded data over the storm of concern may have prevented the radar-defined mesocyclone to be calculated at the 1115 UTC volume scan. Manual calculations utilizing the V/R Shear feature of the PUP indicated rotations of around 35 knots around 65 nm from the radar.

Recommendation: Radar operators must be aware of range folding that may occur when Auto PRF is on. The Auto PRF feature may move the Rmax ring and obscure important storms. In such instances Auto PRF should be turned off, and an Rmax manually assigned which will keep range folded data away from important storms.

Finding: A radar-detected mesocyclone occurred on the 1121 (6:21 a.m.) UTC volume scan. The mesocyclone was detected in the apparent cyclonic feature of the reflectivity pattern. Such a pattern increases the likelihood of either strong bowing winds, or in more rare events, a tornado. In this case, the mesocyclone was occurring in the lower two volume scans, 0.5 and 1.5 degrees.

Recommendation: When mesocyclones are detected in comma-head features in the lowest elevations of the radar scan, statements or warnings should be issued to emphasize the threat of damaging winds or possible tornadoes with the system.

Finding: On the subsequent volume scan at 1127 (6:27 a.m.) UTC, the radar failed to continue detecting the mesocyclone. There was no range folding or significant data dropout during this scan. However, V/R Shear calculations indicated a continuing moderate to strong mesocyclone at 0.5 and 1.5 degrees.

Recommendation: Although the radar does not detect a mesocyclone on the scans subsequent to a detected circulation, the radar operator should investigate the velocity shears and apply operator-defined methods, especially if the rotations are in the lowest layers.

Finding: The mesocyclone and associated tornado spun up in the lowest layers quite rapidly near the rain/no-rain line as indicated on the Storm Total Precipitation product. Recent field investigations have pointed out the need for forecasters to monitor the existence of low-level boundaries to assist in locating areas where thunderstorms may interact with those boundaries and generate fast spin-up tornadoes at the surface. Locating rain-cooled areas can assist in monitoring these surface boundaries.

Recommendation: Radar operators and mesoscale forecasters should utilize the radar precipitation estimate patterns to help them identify surface boundaries, especially in areas where there is no mesoscale surface observation network, and at night when visible satellite images are not available.

GUIDANCE

Fact: A vigorous upper trough in the Oklahoma/Arkansas area at 0000 (7:00 p.m. 19 March) UTC, 20 March 1998, moved rapidly eastward into the Mississippi/Alabama area by 1200 (7:00 a.m.) UTC. An associated cold front located in Mississippi at 0000 (7:00 p.m.) UTC moved into western and northern Georgia by 1200 (7:00 a.m.) UTC. The atmosphere in northern Georgia at 0000 (7:00 p.m.) UTC was slightly unstable. The atmosphere in Alabama near the surface front was moderately unstable. During the nighttime, the moderately unstable air moved eastward with the front into northern Georgia. Wind patterns exhibited strong upper-level diffluence across Georgia between the small jet associated with the upper trough and the strong subtropical jet across the Gulf of Mexico. Northern Georgia was located in the left exit region of the small jet streak associated with the upper trough.

Fact: The ETA model initialized at 0000 (7:00 p.m.) UTC on 20 March appeared to have performed quite well in depicting the movement of the upper trough and associated cold front. The 12-hr ETA forecast, valid near the time of the tornado, indicated deep-layer lifting and a dry tongue of air at 700 mb from the southwest moving into northern Georgia. CAPE values were low to moderate, between 500 and 1000 J/kg. Twelve hour forecast storm-relative helicity values ranged between 200 and 400 m²/sec².

Fact: The SPC issued numerous outlooks and watches across the southern and eastern States on 19 and 20 March as the upper trough crossed those regions and tapped an increasing supply of moisture from the Gulf of Mexico.

Finding: The Day 1 Convective Outlook issued at 0607 (1:07 a.m.) UTC, 19 Mar 1998, included all of Georgia in a Slight Risk Area. Severe thunderstorms were expected to develop in the Arkansas/Louisiana area and move eastward evolving from a supercell environment to a unidirectional shear environment across Georgia (lines with embedded bow echoes). Damaging winds would be the greatest severe weather threat by the time the system moved into Georgia. The outlook was updated at 1446 (9:46 a.m.) UTC to decrease the moderate risk area over the Louisiana/Arkansas area, and "Thunderstorm activity should move across Georgia during the overnight hours ... with damaging wind gusts being the primary severe weather threat."

The Day 1 Outlook issued subsequently at 1915 (2:15 p.m.) UTC shifted emphasis and areas. The new outlook dropped a moderate risk area that had been over northern Alabama and Mississippi and decreased the slight risk area considerably. The new slight risk area included only extreme northwest Georgia. A second slight risk area extended across central Florida. The reasoning for this major adjustment included a concern about widespread low clouds that were limiting daytime heating and veering wind ahead of the cold front that was limiting convergence along the front. Also, greatest lifting over northern Alabama and Mississippi near the warm front would not coincide with the larger CAPE values farther north. The outlook stated: "It appears large hail and damaging gusts will be the primary threats...though marginally favorable storm-relative low/mid level winds suggest an isolated tornadic supercell will still be possible."

The update of the Convective Outlook at 0212 (9:12 p.m.) UTC decreased the slight threat area, but it still included extreme northwest Georgia. CAPE values around 1000 J/kg were expected across northern Alabama into northwest Georgia. The message stated: "Although winds are not particularly strong in the lower levels...veering vertical wind profile...enhanced low level convergence and instability should be sufficient to support a few severe storms through the night."

The last Convective Outlook before the tornado event was issued at 0607 (1:07 a.m.) UTC and was for the period 1200 (7:00 a.m.) UTC, 20 March, through 1200 (7:00 a.m.) UTC, 21 March. The slight risk area included all but northwest and southwest Georgia. The slight risk area included the Hall and White County area. A line of convection was expected to be present at the beginning of the period, around 1200 (7:00 a.m.) UTC, 20 March, from eastern Kentucky all the way down through the central Florida Panhandle. CAPE values near 1000 J/kg and increasingly long curved hodographs would continue to support the eastward-moving squall line through the morning hours.

Recommendation: Most of the outlooks were generally consistent in reasoning and timing. The issuance at 1915 (2:15 p.m.) UTC provided excellent reasoning for the adjustment of the moderate risk to a slight risk (persistent low clouds, lack of low-level convergence, and the disconnect of CAPE and vertical motion), but no reasoning was provided for the elimination of the slight risk from southeast South Carolina, across Georgia, and the Florida Panhandle. Such reasoning should be provided in the text, so that downstream locations can assess the situation more fully.

Fact: SPC Watches issued through the night leading up to the tornado event were consistent with the reasoning provided in the Outlook and with the movement of the weather system. Severe Thunderstorm Watch 112 was issued at 0212 (9:12 p.m.) UTC, valid from 0230-0700 (9:30 p.m. - 2:00 a.m.) UTC, generally for the northern half of Alabama. Emphasis was on the negative-tilted upper trough rotating east northeastward, sufficient instability, dry air aloft, and a veering wind profile.

Severe Thunderstorm Watch 113 was issued at 0509 (12:09 a.m.) UTC, valid from 0545-1100 (12:45 a.m. - 6:00 a.m.) UTC, for eastern Tennessee/Western North Carolina southwestward across eastern Alabama and the northwestern third of Georgia. A strong squall line with embedded bow echoes was expected to continue to move from the earlier watch area into the new watch area. Strong unidirectional wind fields and adequate moisture/instability were sited.

A Watch Status Report issued at 0752 (2:52 a.m.) UTC indicated that bowing segments were evident along the squall line in northeastern Alabama. The report expressed some concern that model surface-based inversions might limit the potential for damaging winds across east-central Alabama and western Georgia; however, the presence of a strong low-level jet, dry mid-level air still could produce strong winds, and high CAPE still could produce large hail.

Another Status Report was issued at 0948 (4:48 a.m.) UTC which continued the watch over the eastern counties of the watch. However, the report indicated that radar and satellite imagery showed significant decrease in convective activity. The decreasing trend was expected to continue for the next few hours. Thus, the remaining valid portion of the watch was allowed to expire as scheduled at 1100 (6:00 a.m.) UTC. The report further stated that SPC would "continue to monitor areas in Georgia and the Carolinas...ahead of the cold front and along the warm front...for new convective development later in the morning."

A special Mesoscale Discussion from SPC at 1134 (6:34 a.m.) UTC summarized their concern over the potential for developing severe thunderstorms. They indicated that they will continue to monitor the situation, but present indications are that a watch will be issued soon.

Tornado Watch 114 was issued at 1154 (6:54 a.m.) UTC, valid from 1230-1800 (7:30 a.m. - 1:00 p.m.) UTC, for much of western North Carolina, northwestern South Carolina, and extreme northeastern Georgia. The watch sited rapid development of thunderstorms ahead of a strong short wave trough over eastern Georgia. Increasing moisture and favorable wind shears should produce supercells and isolated tornadoes. Damaging bow echoes also were mentioned.

Finding: The SPC forecasters were diligent in monitoring the weather system and in providing reasoning in timely status reports, watches, and discussions. The killer tornado occurred during a brief period between watches. The rapid development of convection and the tornado, where convection had been decreasing during typically the most stable time of the day, was an isolated event and extremely difficult to forecast.

Recommendations: No action necessary.

PREDICTIONS

Finding: The Short Term Forecast is an appropriate vehicle for keeping users informed on non-routine weather factors expected to affect their area in a non-severe situation. In this case, Short Term Forecasts were timely with appropriate specifics as thunderstorms approached the metro Atlanta area. Once the storms exited that area, the Short Term Forecasts were vague and contained little in the way of time or place specifics. The last Short Term Forecast before the event for the affected area was issued at 5:10 a.m. and mentioned that "Scattered showers and thunderstorms will continue across the area through 7:00 a.m. Some of the stronger storms may contain frequent lightning...heavy downpours and small hail." The next Short Term Forecast was issued at 7:19 a.m., after the event had passed. It appears Short Term Forecasts, in this case, were underutilized.

Recommendation: Short Term Forecasts should follow guidance provided in a 10/9/97 memo to Forecasters/Interns from the SOO that emphasized the importance of time and place specifics in all Short Term Forecasts and avoid using generic statements such as "scattered showers and thunderstorms will move across the area."

Fact: Zone forecasts were updated in a timely manner, responding to watch issuances and cancellations. The zone forecast in effect at the time of the event had been issued at 6:02 a.m. in response to the cancellation of Severe Thunderstorm Watch #113 and forecasted "Mostly cloudy. A 30 percent chance of showers and possibly a thunderstorm this morning. High in the upper 50s. Southwest to west winds 15 to 20 mph and gusty. Caution is advised on area lakes." Relatively low POP's were based on the apparent decrease in aerial coverage of the thunderstorm area over northern Georgia over the preceding few hours.

Fact: The thunderstorm outlook issued at 6:45 a.m. mentioned a slight risk of severe thunderstorms for the area east of the event area. It also included the statement "scattered thunderstorms are expected...but should not be severe." This outlook was apparently based on the SPC Day 1 outlook rather than analysis of currently available radar data, and was before first damage reports became available.

Finding: An F3 tornado touched down in Hall County, Georgia, at about 6:25 a.m. EST on 20 March 1998. It lasted approximately 12 minutes. The first warning on this event was a Severe Thunderstorm Warning issued for Hall County at 6:46 a.m., valid until 7:00 a.m, apparently based on an initial report from WSB-TV that damage and a possible tornado had occurred. This warning was followed up with a Tornado Warning for White County at 6:51 a.m., valid until 7:15 a.m, upon confirmation of the tornado and damage in Hall County via a follow-up call to Hall County 911. Neither warning contained any time nor place specifics regarding the storm's projected path, nor any mention that damage reports had been received. The White County warning contained an incorrect and potentially misleading reference to southern Hall County that was not detected or corrected.

Recommendation: The bullet-formatted style for warnings was implemented nationwide on 15 April 1998. This format includes time and place specifics regarding the storm path and affected communities. The staff should review all products closely for accuracy and issue corrections as necessary, even if detected after transmission.

Finding: Staffing at the time of the event consisted of two forecasters, one radar operator, one Meteorological Intern (performing HMT duties), and an HMT working on the morning Upper Air observation. Station staffing plans are thorough and appropriate for recognized and anticipated events. In this situation, however, while staffing was adequate for the perceived threat at the time of the event, it proved insufficient to handle the sudden workload generated once the event was recognized. Due to the responsibilities of other staff members, the radar operator was suddenly in the position of having to handle radar monitoring, warning preparation, coordination and documentation duties by himself.

Recommendation: Reemphasize operational priorities to ensure that severe weather duties are given top priority when severe events occur with non-team staffing. Follow station instructions that call for delaying routine products if necessary instead of severe weather products and Short Term Forecast.

Finding: No contact appears to have been initiated with counties affected by the parent thunderstorm until approximately 6:40 a.m. to 6:45 a.m. despite radar indications of a potentially severe storm 15 to 20 minutes earlier. This was apparently due to staff members involved with other duties and lack of any severe history up to that point.

Recommendation: Consider revising station instructions to encourage real-time information exchange with areas affected by potentially severe storms. Such calls can serve as a "heads-up" to affected areas, allowing for quicker response by emergency management and other officials if a warning is required and for real-time feedback on the nature/progress of storms throughout the event.

Finding: Initial notification of the event seems to have been a phone call from WSB-TV at approximately 6:40 a.m. to 6:43 a.m. based on recollections of the radar operator and the Forecaster in Charge (FIC). The FIC indicated subsequent calls were made to the 911 center between 6:40 a.m. and 6:45 a.m. The FIC reported no answer on the first call. A subsequent call did get through and indicated that "several tornadoes were on the ground." No written documentation of any of this information exchange was kept, and times are therefore estimated. A Severe Weather Spotters Log is available for this purpose and station instructions provide for it. The first entry on the log of this event was at 7:03 a.m.

Recommendation: Ensure that station instructions give higher priority to logging of coordination calls and spotter reports and are followed.

EXTERNAL COORDINATION AND DISSEMINATION

Finding: Hall and White Counties in Georgia are covered by NOAA Weather Radio (NWR) broadcast from the Peachtree City NWS and transmitted on a frequency of 162.400 MHz from a tower near Athens, Georgia. All warnings for these counties are coded on the Specific Area Message Encoder (SAME) for input into the Emergency Alert System (EAS) and are given a 1050 MHz tone alert. Reception of the NWR signal in these counties is good.

Finding: No logs or records are kept as to the dissemination times of watch or warning products on the Athens NWR. A printer attached to the Atlanta NWR transmitter only logs warnings/watches/statements issued for that transmitter.

Recommendation: Continue all broadcasts of warnings into these counties using SAME and the tone alert. The Peachtree City office should develop a system to log the times of all warnings/watches/statements broadcast on each NWR transmitter in their area of responsibility.

Finding: The Emergency Manager in Hall county was not thoroughly familiar with the SAME feature of NWR.

Recommendation: The Peachtree City NWS should continue aggressively promoting NWR. The office should consider using NWR and other sources to educate people about NWR and the features of the SAME radios.

Finding: The office at Peachtree City maintains a Severe Weather Spotters Log to record calls and severe weather reports made from the office or received by spotters or other contacts. During this event, only six entries were made on this log. Office personnel did not feel there was enough time to record entries on this log. Recalling specific reports from this event was extremely difficult for the staff.

Recommendation: There were several other key calls made out of or into the Peachtree City office that were not recorded on the log during this event. The station should make every effort to place a higher priority on recording all contacts on this log. This will give a permanent record of the actions taken by the office and storm reports received by spotters or other individuals.

Finding: The Peachtree City office normally will read watch and warning information over the Georgia National Warning System (NAWAS). Hall County has a NAWAS drop in their county that is staffed 24 hours a day. White county does not have a NAWAS drop in the county. The Peachtree City office staff attempted to contact Hall County via the 911 operator. However, the 911 operators were deluged with calls within a few minutes after the tornado and were unable to respond to NWS personnel. Neither county received any communication from the Peachtree City office via NAWAS.

Recommendation: Contacting Hall County emergency officials via the 911 system proved to be ineffective because of volume of calls from Hall County citizens. Using NAWAS to contact Hall County (and other officials with NAWAS) is an alternative that many offices should explore.

Finding: There are four network affiliated television stations that cover the affected area. All these TV stations have automatic watch and warning receiving equipment that ingests its input from the GTE/Contel Weather Wire Service (NWWS). This equipment will scroll the information on the bottom of the TV screen and then put the affected counties on a map until they reach their expiration time. All four networks were satisfied with the services generally provided by the Peachtree City NWS. One station commented about not having a current phone number to reach the NWS.

Recommendation: The WCM should call upon these outlets to ensure they have telephone numbers for access into the NWS.

RESPONSE

Fact: The Peachtree City NWS conducted two spotter training programs in Hall county and one in White County during the last two years. The Peachtree City office has an outreach program called Adopt-a-County, in which a staff member adopts a particular county and is the main liaison with local officials. This individual also makes periodic visits to the area. Both Hall and White counties had one visit each during the last two years.

Finding: Twelve people died as a direct result of this tornado and over one hundred fifty people were injured. Eleven of the fatalities occurred in mobile homes.

Recommendation: Continue to emphasize the dangers of mobile home living in all spotter, safety, and preparedness activities. People in mobile homes need to evacuate them and get to a more sturdy structure before severe weather threatens. It is extremely important for everyone to own and use a NOAA Weather Radio. This is especially important for mobile home dwellers due to the high occurrence of tornado fatalities in mobile homes.