Meteo 101

Meteo 241

Meteo 361

Meteo 410


Page 1



by Philip Lutzak – April 2006



 In what has already turned out to be a very bad severe weather season for the country, another outbreak of severe thunderstorms and tornadoes sprang up in the U.S. on April 2nd, this time in the Tennessee and Missouri Valleys, flattening entire communities and killing 27 people. This is the deadliest weather event so far in 2006, and follows the long term trend: April holds the distinction of having more killer tornadoes than any other month in the United States.

  On the right (figure 1) is a picture of multiple tornadoes in a supercell thunderstorm over Dunklin County, Missouri, just before sunset on Sunday, April 2, 2006. One person was killed there, and these particular tornadoes did tremendous damage in Dunklin and Pemiscot counties in southeastern Missouri, destroying 80% of the homes in some of the towns there.


 A number of other supercell thunderstorms were occurring at this time in a large area of the mid-west and south as the outbreak neared its peak. In Figure 2 below is the storm reports map from April 2, from the Storm Prediction Center (SPC), showing the huge extent of the severe weather.




Figure 1. Multiple vortices in Dunklin County, Missouri, April 2, 2006. (click for wide shot) Photo by Duncan Phenix. Courtesy NWS Forecast Office, Memphis, TN.

  The 872 reports comprised the second largest single outbreak of severe weather on record with the SPC. The entire episode covered 16 states and produced 27 fatalities. Here is the complete storm report for April 2nd, 2006.


                     Figure 2. Storm Reports Map for April 2, 2006. Courtesy SPC, Norman, OK.



  Since this outbreak covered such a large area, and it's not possible to cover it all in this report, I intend to focus primarily on the tornadic supercell thunderstorms that occurred in southeastern Missouri, northeastern Arkansas and western Tennessee and the particular atmospheric conditions that occurred there. Below in figures 3, 4 and 5 are some photos of the damage done by the tornadoes that roared into Dyer and Gibson counties of Tennessee shortly after sunset on April 2nd. These tornadoes ranged from F1 to F3 on the Fujita scale, with two of the three F3 tornadoes causing all of the 24 fatalities in Tennessee.


Figure 3. Courtesy National Weather Service Office, Memphis, Tennessee. Figure 4. Courtesy National Weather Service Office, Memphis, Tennessee.


Figure 5. Courtesy National Weather Service Office, Memphis, Tennessee.  


  The map below, from the NWS at Memphis, Tennessee, shows the paths of the tornadoes, along with their Fujita scale ratings, that devastated parts of Missouri, Arkansas and Tennessee. The F3 tornado that occurred in Arkansas and Missouri carved a path 75 miles long and killed one person in southeastern Missouri. The F3 tornado in westernmost Tennessee had an 18 mile long path, from just east of the Mississippi River to Newbern, and killed 16 people. The easternmost F3 storm had a 20 mile path from Dyer to Gibson, and caused 8 fatalities.


Figure 6. Courtesy National Weather Service Office, Memphis, Tennessee. Click here for large version.



  The large-scale, or synoptic weather conditions that produced this outbreak came, as is often the case at this time of year, from the west and south. This surface map from 7:00 AM EST on April 2nd shows low pressure over southern Nebraska, which by mid-afternoon was over Iowa, with its trailing cold front moving through Missouri and Arkansas towards Illinois and Kentucky. As this system was pulling very warm and moist air at the surface into the southern and central states from the Gulf of Mexico, a strong 500mb trough with a high vorticity max was approaching from the west, adding a good deal of divergence and thus lift, aloft. In addition, a strong jet stream with embedded powerful jet streaks began moving in overhead, adding additional large amounts of lift to the unstable air below. 

 As noted by the SPC in their 2:55 PM CDT Convective Outlook, the 1800 Springfield, Missouri sounding showed that CAPE (Convective Available Potential Energy, or amount of potential instability in the atmosphere) was reaching very high values of 1000 - 2000 Joules/kg over the Missouri Valley, indicating potentially extremely unstable air moving into the region (see also this 1800 map of CAPE .) Just as importantly, wind shear values were becoming very high over the area (see SPC 0-6km shear from 1800). Because of this high potential for instability, the strong vertical wind shear and the previously mentioned strong upper-air support, the SPC shortly thereafter issued tornado watches 132 and 133 for the area. These visible and water vapor satellite images from 4PM CDT show the strong low over Iowa and the powerful squall line developing south of it in Illinois, eastern Missouri and Arkansas. By 4:30 PM central time, dozens of tornadoes and severe thunderstorms were descending on Iowa, Illinois, Arkansas and Missouri, producing widespread damage and injuries. At 4:40 PM, the SPC issued tornado watch 134 for Arkansas and north Mississippi through western Tennessee, since it was apparent that the surface and upper-air conditions that had produced the severe weather were in no way abating. In fact, the SPC was so concerned that they issued another MCD at 7:06 PM CDT, specifically noting the danger developing for west Tennessee. The SPC radar composite shown below in figure 7 from the same time shows the multiple supercells over the area.


            Figure 7. NWS Radar Reflectivity Composite for 04/02/2006 at 7:06 PM CDT. Courtesy SPC, Norman, OK.

Note in particular the two supercells over the boot-heel of Missouri. These were the eastward-moving storms that produced the tornadoes and damage pictured earlier in this report. The SPC was concerned that these storms would continue unabated into Tennessee, and in fact could produce more tornadoes even as other tornadic supercells developed around them. Besides the high surface dewpoints, the MCD noted the strong 40 knot low-level jet over western Tennessee and Kentucky, and the "nose" of the mid-level jet (notice the area of orange, 58 knots or higher) moving in overhead. The eastward advance of these jets into the storm environment added a significant amount of vertical wind shear to the developing thunderstorms: speed shear through a strong increase in wind speed from the surface up into the mid and upper-levels, and directional shear by moving very strong westerly winds in over the low-level southeasterly and southerly winds below them, providing the classic veering in wind direction with height that we see in many tornadic supercells.



  As we saw from the SPC radar composite and storm reports, a number of supercells in the area were producing multiple tornadoes of significant intensity by late afternoon. Since Doppler radar is so essential to forecasting and monitoring events such as this, I have devoted a separate RADAR PAGE for the illustration and explanation of some radar images from the time, including radar image loops. 

  In addition, it is worth looking at the radar summary chart from NOAA IWIN services below in figure 8. This analysis gives us an approximation of the height of cloud tops by displaying the echo returns of the highest detectable hydrometeors. We can see that the heights over west Tennessee were from 52,000 to 64,000 feet. Keep in mind that these thunderstorm heights are approximate values and that they can be used here with confidence because we can see the high precipitation echoes correlate with the areas where we know thunderstorms were present at the time. These thunderstorms are known as "high-topped".


                        Figure 8. Radar summary showing tops of echo returns from April 2, 2006 at 10:45 PM CDT. Courtesy NOAA Aviation weather.




  A discussion of storm relative winds is important in studying this outbreak, for several reasons. First of all, they are key to any discussion of storm relative helicity, which we will examine below in the Mesoscale Analysis section. Second, they are useful in themselves in forecasting supercells and tornadoes. And third, because they can help us identify the type of supercells we are examining. This last characteristic turns out to be important in the discussion of storm motion in the next section. Please see the STORM RELATIVE WINDS PAGE.  




  One of the tools that meteorologists use to predict the direction and speed of storm motion is the "Rasmussen technique", developed by Erik Rasmussen and David Blanchard at the Storm Prediction Center. This technique uses the idea that the difference between the wind speed and direction at the surface up to roughly 4,000 meters or 13,000 feet can be used to predict the average motion of supercell thunderstorms. I've illustrated the method in detail in this storm motion graphics page. After looking at this method, we can see how we arrived at the storm motion vector value of 29042, which means a mean storm motion from 290 degrees, or from the west-northwest, at 42 knots, or 48 mph.



  To determine how well the Rasmussen technique worked, we will calculate the actual storm motion. For the direction of motion, I used the map of actual tornado paths provided by the NWS in Memphis. For the forward speed, I used the speeds provided by the National Weather Service in the tornado warnings they issued. Both are good proxies of the supercell motion, since tornadoes generally travel at the same speed and direction as their parent supercell, with few aberrations.

  I have included, in figure 9 on the right, a close-up of the map of tornado paths from figure 6, and added the average originating direction in degrees.  The four tornado tracks shown all occurred from the same two supercells between 7:00 PM and 9:30 PM CDT. I have not included the tornado from further south at Brownville, TN because it occurred later in the evening at 10:30PM and from a different storm. As you can see, the origin of motion ranged from the west to slightly south of west to northwest. When we average their "from" directions we get 274 degrees.


      Figure 9. Courtesy National Weather Service Office, Memphis, Tennessee. Large version.


  Notice that the actual motion of 274 degrees deviated from our predicted motion of 290 degrees from the Rasmussen technique. In this case, there is a reasonable explanation for the actual storms' deviation in direction from the forecasted direction. First, it must be remembered here that these storms were in the class of supercells known as high precipitation (HP) supercells, covered in the previous section on STORM RELATIVE WINDS. In a study on Supercell Morphology by Rasmussen and Straka 1998, they state that "While HP storms propagated much more erratically, several of these storms tended to propagate more rapidly and in a direction more along the BL–4-km shear vector, so that the SR upper-tropospheric flow was reduced further." So there is evidence that with HP storms the direction of motion may stay closer to the original surface-4km shear vector direction that we start with in the Rasmussen technique, i.e. there is less inclination to the right than that method employs. If that is so, then our actual motion would be less "right-moving" than the predicted motion, which is what happened here. Rasmussen also states that more research needs to be done on this, although right now there is speculation that the strong outflow in HP storms may affect the way these storms propagate forward. Professor Lee Grenci of Penn State University has stated that "Rasmussen once remarked that his storm-motion predictor worked best for classic (CL) and LP storms, and not as well for HP supercells. It's probably because HP storms have so much rain-cooled outflow that their gust-front dynamics affect the propagation."



  The following are excerpts from the actual tornado warnings issued by the NWS for the four tornadoes whose tracks are shown above at the time of their occurrence. For brevity's sake, I have included only the portion of the warnings that included the storm location and motion. The storm speeds are highlighted in red. The complete listings from which these were taken are available in this selection of tornado warnings from April 2nd.








 The average of these forward speeds is 48.7 mph or 42 knots. That's exactly what the Rasmussen technique predicted.

  In conclusion, the Rasmussen technique worked very well in predicting the speed component of our storm motion here. It was off by 16 degrees in predicted direction, but we have evidence from the literature that the method doesn't work as well for the HP type supercells that we observed here. In addition, predicting our storm direction is most accurate when we can use soundings as close as possible to the area we need to examine. In this case we had a good representative sounding from Nashville, where the environment was very similar to the area of western Tennessee where our supercells occurred. Unfortunately, there are many areas of the country where soundings are more sparse and then our only available readings may not adequately represent the environment that we are examining. So we should make sure that our sounding data is in fairly close proximity to the storms, and in the same air mass, before relying on this method.






Figure 10. The SPC's Mesoscale Discussion map from 4/2/2006, 7:06PM CDT. Courtesy SPC.



  As alluded to in the synoptic analysis, the above map in figure 10, taken from the SPC's Mesoscale Discussion at 7:06PM CDT on April 2nd, shows just how concerned they were about this particular outbreak by the evening of April 2nd, 2006. For the National Weather Service to be this urgent and direct in a warning, there must be very specific conditions occurring in an area. To find out why, we need to examine the mesoscale conditions that existed at the time, and some of the most state of the art tools describing mesoscale atmospheric conditions available to us at this time are provided by the SPC. I chose readings at or near 8PM CDT, since these well represent the conditions of the atmosphere at the time the worst of the tornadoes were occurring.





  The most important indicators for thunderstorm development that can produce supercells and tornadoes are all related to the concept of very unstable air undergoing directional shear, or turning, in the lower atmosphere as it is drawn upwards at increasing speeds, and then undergoing additional impetus to spin in the higher levels. While theory on the development of tornadoes still has a long way to go, we now have better tools available to predict them. The three that have consistently proved most valuable at this point in our history of understanding are CAPE (previously defined), wind shear and storm relative helicity, all provided to us at hourly intervals by the SPC. Since some of the meteorological definitions I've linked to in the AMS glossary are quite technical, I will try to use somewhat less complex explanations below. At the end of the section I have included Lifting Condensation Level (LCL) values and finally, the SPC composite index known as the significant tornado parameter.





  The first, CAPE, mentioned earlier, deals with the potential instability of the air in a particular environment. The more it exhibits a tendency to keep rising as it moves upward, the more unstable it is. Although there are several measures of CAPE, taken from different starting points in the atmosphere, it's often preferable to use the mean layer cape, or 100mb mlcape, as an indicator of instability when the boundary layer is well-mixed, i.e. the bottom layer of the atmosphere has been well mixed, or "turned over" by strong winds and/or convection. This 18Z skew-T from Springfield, Missouri shows a reasonably well-mixed boundary layer, so mlcape is a useful indicator here. The chart below in figure 11a, of mean layer CAPE from 8PM taken from the SPC (marked in red lines and numbers), shows values in western Tennessee and southeastern Missouri that were extremely high, at 2,500 to 3,000 J/kg. In addition, this chart displays CIN, or convective inhibition (marked with blue shading and numbers), which is the amount of energy needed to lift an air parcel upward to the point called the LFC or level of free convection, where it will be able to rise freely on its own. It is basically a measurement of how much inhibition to rising is present in the atmosphere, or the resistance to CAPE. The very low values in western Tennessee, eastern Arkansas and southeastern Missouri indicate that there was almost no resistance to the potentially explosive upward movement of the air.   


     Figure 11a. 100mb mean layer CAPE from 4/2/2006, 8:00PM CDT. Courtesy SPC, Norman, OK.


  In addition, I have included another commonly used value of CAPE, since it also clearly shows how unstable the environment was at the time. Figure 11b shows surface based CAPE (sbCAPE) and CIN, with very high CAPE values of 1000-2000 J/kg and more. Although CIN here is a little higher than the mlCIN above, there was so much forcing from the approaching front and upper levels (see synoptic discussion above) that it was not an impediment to supercell development.

        Figure 11b. Surface based CAPE from 4/2/2006, 8:00PM CDT. Courtesy SPC, Norman, OK.



  Vertical wind shear is a measure of the change in wind direction and speed as we go upward in the atmosphere. The more veering, or clockwise turning (directional shear) we have in the winds from the surface to 6km, and the faster these winds increase with height (velocity or speed shear), the more likely that supercells can acquire a rotating updraft, and in turn, tornadoes. One of the best tools to use for measuring vertical wind shear is 0-6km shear, which gives us a good measurement of the vertical shear up to 6,000 meters or roughly 20,000 feet.  In most cases, 35-40 knots of shear or more is ideal in seeing supercells develop. The readings in figure 12a below show strong shear values of 55 to as much as 75 knots in the outbreak area of concern. And although very high shear values can sometimes be an impediment to supercells, it would not be a factor with high-topped thunderstorms such as these.

      Figure 12a. 0-6km shear vector from 04/02/2006 at 8:00PM CDT. Courtesy SPC, Norman, OK.


   Another good measurement of vertical wind shear available to us is profiler data, which measures the wind speed and direction in a vertical profile each hour, for a set number of hours. The following chart in figure 12b, taken from Wolcott in southern Indiana near the Kentucky border, is well representative of the powerful wind shear occurring in the outbreak area on the afternoon and evening of April 2nd. We can clearly see the veering of the wind with height from southeast to southwest, and increasing from 10-20 knots at the surface to over 50 knots at mid-levels by 1700. By 7PM the winds were 50 knots at the 6km level, increasing even further to 70 knots between 8 and 12 kilometers.


                  Figure 12b. Wind profiler data from 6AM to 7PM CDT on April 2nd, 2006. Courtesy RAP/UCAR.





  Storm relative helicity is, in the simplest sense, a measure of the amount of spin and rate of inflow into a thunderstorm in the lowest layers of the atmosphere, which is key to the ability of air parcels to spin further as they are drawn upwards into a supercell circulation and ultimately cause tornadic circulations.

  To understand this concept, it helps to picture a tube of air, in the lowest kilometer or so of the atmosphere, feeding into the thunderstorm. As air parcels move along this tube, they start to spin in a clockwise manner around the tube, acquiring horizontal vorticity. This motion is called "streamwise vorticity". But these parcels spinning around the tube are also moving along it, in towards the thunderstorm. The resultant motion of that streamwise vorticity around the tube and at the same time in towards the thunderstorm (the storm relative winds), causes a helical shaped wind flow. Also, if you picture looking at these clockwise or anticyclonically spinning tubes from behind as they spin in towards the thunderstorm updraft, you can visualize how the spin will become counter-clockwise or cyclonic as it changes from a horizontal tube along the ground to a vertical tube moving upwards. It makes sense that the higher the speed at which the streamwise vorticity is moving, and the faster the storm-relative winds are blowing in towards the thunderstorm, the higher the possibility that the updraft will rotate, possibly allowing tornadogenesis. We call the rate at which an updraft ingests this streamwise vorticity the Storm Relative Helicity, or SRH. The SPC uses SRH as an indicator of an increased threat of tornadoes with supercells, and uses thresholds of >100 m2/s2 for 0-1km SRH, and values of >250 m2/s2 for 0-3km. Another SRH commonly used is effective SRH, which can help us discriminate between tornadic and non-tornadic supercells. Values over 300m2/s2 indicate a probability of strong tornadoes of F2 or higher potential.

  It must be emphasized that there are no known realistic thresholds that we can use to tell if a supercell will be tornadic or not, so like all of our other measurements, these must be used in concert with other tools and used judiciously.     

   As for the storm relative helicity in this event, below is the 0-1km chart of SRH from 8 PM on April 2nd (figure 13a). Note that values well above the threshold of 100 m2/s2 cover the entire area where the outbreak occurred, and values go very high, to over 300 and 400 m2/s2 in the areas where our worst tornadoes occurred in Missouri, Arkansas and Tennessee.  In figure 13b, the 0-3km SRH values are also exceptionally high, with values over 300 m2/s2 or more in the entire outbreak area, and in fact they are 350-450 m2/s2 in the area where the worst tornadoes occurred. Finally, the effective SRH in figure 13c shows values of 200-400 m2/s2 in the worst-hit areas, also indicative of tornadic supercell development.

Figure 13a. 0-1km storm-relative helicity from 04/02/2006 at 8:00PM CDT. Courtesy SPC.



    Figure 13b. 0-3km storm-relative helicity from 04/02/2006 at 8:00PM CDT. Courtesy SPC.


         Figure 13c. Effective storm-relative helicity from 04/02/2006 at 8:00PM CDT. Courtesy SPC.




  Since it can also be an important indicator of strong tornadoes, and we need it to calculate the significant tornado composite examined below, we should look at the height of the Lifting Condensation Level (LCL) in this outbreak. It has been shown by Rasmussen and Blanchard 1998 that very low LCL heights have a high correlation to the potential for F2 or higher tornadoes. During this outbreak, we had levels of 800-1200 meters (see figure 14 below), readings which qualify as quite low, and taken with the other factors we've examined here, indicated a higher probability of strong tornadoes with the supercells that developed. 


        Figure 14. 100mb mean parcel LCL height from 04/02/2006 at 8:00PM CDT. Courtesy SPC.




  The final SPC tool I would like to examine is one of their composite indices, which the SPC has developed in an attempt to combine some strong indicators of severe weather into one indicator which may help in future forecasts. The one I chose is the Significant Tornado Parameter, or STP, which gives an indication of the chances of tornadoes capable of F2 damage or higher occurring in a supercell thunderstorm in a particular area. The STP combines mlCAPE/CIN, LCL, SRH and SHEAR. Here in figure 15 is the SPC's definition of the STP, taken from the SPC's Mesoscale Analysis Page:


Significant Tornado Parameter

A multi-parameter index that includes effective bulk shear, effective SRH, 100-mb mean parcel CAPE, 100-mb mean parcel CIN, and 100-mb mean parcel LCL height.

The index is formulated as follows:

STP = (mlCAPE/1500 J kg-1) * ((2000-mlLCL)/1500 m) * (ESRH/150 m2 s-2) * (ESHEAR/20 m s-1) * ((200+mlCIN)/150 J kg-1)

When the mlLCL is less than 1000 m AGL, the mlLCL term is set to one, and when the mlCIN is greater than -50 J kg-1, the mlCIN term is set to one. Lastly, the ESHEAR term is capped at a value of 1.5, and set to zero when ESHEAR is less than 12.5 m s-1. A majority of significant tornadoes (F2 or greater damage) have been associated with STP values greater than 1, while most non-tornadic supercells have been associated with values less than 1 in a large sample of RUC analysis proximity soundings.

Figure 15. Significant Tornado Parameter from 04/02/2006 at 8:00PM CDT. Courtesy SPC, Norman, OK.


   The STP was developed in 2003 by meteorologists at the SPC. It was based on looking at RUC upper-air sounding data collected from 916 different sites in the United States. These sites were all in close proximity in time and location to areas experiencing supercell thunderstorms and/or tornadoes. By examining the range of values associated with these readings, the meteorologists were able to find threshold values for each parameter, above or below which significant tornadoes had occurred. Each parameter has been adjusted in the equation to make its value come out to 1 if the threshold value is reached. For example, since they found that average values of mlCAPE above 1500 J/kg were found in the F2 tornadoes studied but below 1500 J/kg in weaker tornadoes or non-tornadic supercells, the mlCAPE parameter is divided by 1500 so that its value will be greater than 1 if the mlCAPE is higher but less than 1 when the mlCAPE is lower than 1500. By doing this with each parameter, it's easy to see how an index above 1 will indicate a higher probability of significant tornadoes. When the parameters are multiplied together, the lower than 1 values will decrease the total value, as the higher values will increase it, resulting in a balancing of all of the parameters against each other. Also, since CIN is an inhibitor of convection, these values are adjusted so that they will be less that 1 when they are high, so that higher CIN values will reduce the overall index. Obviously, this index can be misleading and should be used with caution, since 1 or more parameters may have very high values while the others do not, resulting in a high total value that is meaningless. 

  The most important caveat for us to remember here is that composite indices are a helpful tool in gauging whether certain types of severe weather may occur, but they are NOT intended to replace the other parameters which we need to rely on. This index, for example, should raise eyebrows when it starts exceeding a value of 1, but when that happens, it means that we should continue looking at the basic three discussed in this paper - CAPE, SHEAR and SRH, as well as the other atmospheric indicators that we have learned to use, and then try to make an intelligent forecast based on ALL of the data. 


   Here is the latest paper on STP from the SPC with an explanation of the STP parameters, and how they've been upgraded since the index was first created.

   And finally, below in figure 16, is the STP chart taken from 8PM CDT on April 2nd, when multiple tornadoes of F2 and F3 intensity were occurring in western Tennessee.

          Figure 16. Significant Tornado Parameter from 04/02/2006 at 8:00PM CDT. Courtesy SPC.

   These very high values of 4 to as high as 7 are rarely seen, and correlated well with the area where very powerful tornadoes occurred. Since we have already examined the individual parameters that make up this index, we can see why it was so high in this outbreak, especially in our area of study - eastern Arkansas, southeastern Missouri, and western Tennessee.  



  In conclusion, the parameters that we currently most rely on to predict tornadic supercells were all indicative of a significant outbreak of tornadoes. Very high values of CAPE, strong vertical wind shear, and high storm-relative helicity had developed over the region, so we knew that any decent lifting that occurred over the area would probably induce supercell development, and that some of them could be tornadic. Unfortunately we also wound up with an even stronger contribution to vertical wind shear from the low and mid level jets, and an approaching 500mb vorticity max and jet streaks at 300mb provided plenty of lift. Thus the likelihood of powerful supercells with tornadoes became a reality, and we saw many of them. The strong tornadoes that resulted occurred in high-topped, fast moving HP supercells, many if not all of them likely rain-wrapped, thus making it a very dangerous situation for the people in their paths, especially at night.

  Finally, it is still the unfortunate reality that no matter how well forecasted, we can't guarantee that everybody will hear or heed the warnings, especially when these storms occur after dark, as they did here. So although the event was well forecasted, there were still quite a few fatalities in the rural areas of western Tennessee. 






Storm reports:

NWS Forecast Office, Memphis, TN April 2, 2006 - Storm Damage Surveys and Survey Photos

SPC Complete Storm Report for April 2nd, 2006



Technical papers:

Variations in Supercell Morphology. Part I: Observations of the Role of Upper-Level Storm-Relative Flow

Predicting Supercell Motion Using a New Hodograph Technique

An update to the supercell composite and significant tornado parameters