Russell Pfost, NWSFO Miami, FL


The storm system that produced the Natchez downburst around 1010 AM CST on February 26, 1998, was part of a very classic looking, dynamic storm system that affected a large part of Louisiana and Mississippi besides Natchez. The event began around 730 AM CST with wind gusts greater than 100 mph reported at an offshore oil rig south of Cameron Parish, Louisiana. Many wind damage reports followed across southwest and central Louisiana through the early morning hours, and an F1 tornado touched down at Chicot State Park in Evangeline Parish at 908 AM CST where a large number of trees were downed. Before the storm system left Louisiana 12 persons were injured and hundreds of thousands of dollars of damage had been done.

At 1010 AM CST, winds in excess of 70 mph struck the city of Natchez, Mississippi, which is situated on a high bluff by the Mississippi River. Natchez is a very historical city with a large number of antebellum mansions and buildings that make it a popular tourist attraction. The storm system caused over 30 million dollars of damage to Natchez, including several of the most historical homes and attractions. As the storm system swept northeast, a second F1 tornado was spawned around 1115 AM CST near the Natchez Trace scenic highway in Claiborne County, Mississippi, near the Rocky Springs camp ground. Four persons were injured in Mississippi, for a total of 16 in the two states. The storm system resulted in many more reports of damage across Mississippi as it broadened and moved northeast at more than 50 mph.

Such a long lived system must be highly organized to produce so much damage across a wide area. In fact, this storm system satisfies Johns' (1987) criteria for a derecho:


  • a concentrated area of reports consisting of convectively induced wind damage and/or convective gusts > 26 m/s (50 kt). This area must have a major axis length of at least 400 km (250 nmi).
  • The reports within this area must also exhibit a non-random pattern of occurrence (chronological progression).
  • Within the area there must be at least three reports, separated by 64 km (40 nmi) or more, of either F1 damage and/or convective gusts of 33 m/s (64 kt) or greater.
  • No more than 3h can elapse between successive wind damage events. (Johns, 1987)

Using storm data, a rough plot of the damage reports from this day was constructed. The following figure shows the progression of the storm system very well.

This event was classic in other ways, too! The synoptic pattern strongly suggested potential for severe weather, with a major negatively tilted trough rotating through the lower Mississippi valley, a strong jet max rounding the base of the trough, and very strong vertical motion and enhanced divergence across the Jackson CWA.

300 mb chart

500 mb chart

700 mb chart

850 mb chart

surface chart


The Jackson sounding for 12Z was fairly stable with a significant cap between 800 and 700 mb that would have to be overcome by either mechanical lifting or cooling or moistening of the layer. The 32km eta model sounding for Natchez for 15Z was nearly pseudoadiabatic and moist through a deep layer...much more so than the observed Jackson 12Z sounding.


The Jackson hodograph for 12Z showed substantial shear (almost 30 knots) through the lowest 6 km and helicity about 220 m2/s2 through the lowest 3 km. These values are sufficient for severe weather in the south but are not unusually strong.

Level II archive data from the KLCH Lake Charles WSR-88D was retrieved and analyzed through WATADS. The cause of the extensive damage in Jefferson Davis Parish east of Lake Charles is readily seen through these two radar products at 823 AM CST as the bow echo was moving through Fenton, LA. Small areas of velocities greater than 80 knots are visible at Fenton and west of Welsh, LA, as well as in the extreme southwest corner of Jefferson Davis Parish.

1423Z 0.5 base velocity product

In the reflectivity product, these areas of damaging wind correlate well with weak reflectivity echo notches evident from DeQuincy to Fenton, from Lake Charles to Welsh, and from Hackberry to the extreme southwest corner of Jefferson Davis Parish.

1423 0.5 base reflectivity product

Level II archive data from the KJAN WSR-88D was also retrieved and analyzed through the WATADS software package. Several interesting views of the bow echo follow. The first one is a 0.5 degree elevation base reflectivity product as the bow echo was causing the damage at Natchez about 10:10 AM. Note the weak echo channel behind the bow echo which is a signature of a very strong rear inflow jet (RIJ) and the hint of a developing comma head over Adams County.

1609Z 0.5 base reflectivity product


Here is a cross section of reflectivity at the same time (10:10 AM). Even though we can't see the lowest parts of the storm due to earth curvature effects, this is obviously a very organized storm with 50 dBz extending as high as 22,000 feet (remember this is winter).

1609Z reflectivity cross section through the Natchez storm


This cross section of storm relative velocity at 10:53 AM does show the classical expectations of a bow echo with the RIJ clearly visible as is the storm relative inflow (red) feeding into the convection and then rising up and over the RIJ.

1653Z reflectivity cross section through the storm


The controversy surrounding this event will never be completely satisfied since we have no low level radar data to analyze and conclusively prove that no tornado occurred in Natchez on this day. However, the radar data we do have strongly suggests that the damage in Natchez on February 26, 1998 around 10 AM was caused by a bow echo and associated straight line winds in excess of 70 mph!


Johns, Robert H., and William D. Hirt, 1987: Derechoes: Widespread Convectively Induced Windstorms, Wea. and Fcstg., 2, 32-49.




Local Climate Water & Weather Topics:
Current Hazards, Current Conditions, Radar, Satellite, Climate, Weather Safety, Contact Us is the U.S. government's official web portal to all federal, state and local government web resources and services.