Friday, May 27, 2011

The Joplin tornado environment - poor visibility with large shear & instability

The death toll from last Sunday's horrendous and tragic EF5 tornado in Joplin, Missouri (see my prior blog discussion) is now up to 132 people, according to reports on Friday 5/27/11. As the deadliest single U.S. tornado in over 60 years, it is worth taking a brief look at the storm environment that helped generate it.

The RUC analysis sounding for Joplin at 2200 UTC (5:00 pm CDT, see 2nd image above), an estimate of the environment roughly 40-45 minutes before the tornado struck, showed very large instability (MLCAPE > 4000 J/kg) and quite sizable low-level wind shear, with 0-1 km storm-relative helicity (SRH) near 300 m2/s2. This matched values shown on SPC mesoanalysis maps at 2200 UTC (not shown). On the same SRH-CAPE diagram where I plotted the Tuscaloosa AL tornado environment in a blog post a few weeks back (see 3rd image above), this combination of SRH and CAPE places the Joplin tornado environment in the same general area and magnitude on the diagram as the Moore OK tornado in 1999 and the Greensburg KS tornado in 2007, both rated EF-5 as well. Although the NAM 12-hr model forecast under-represented the energy-helicity index values (EHI, combinations of SRH & CAPE; see 4th image above), it did forecast a good estimate of the pattern, with an EHI maximum indicated over northwest AR and southwest MO. So it is not a total surprise that the tornado was so strong and deadly.

What is a surprise is the huge death toll, which was partly a result of the tornado being rain-wrapped and very difficult to see (see my photo at top above), and the size of the tornado (around 3/4 mi wide) going down through a very populated area. Going back to the RUC sounding above, relatively weak winds of 40 kts or less at storm anvil level (around 300 mb or 30,000 ft and above) probably contributed to the high-precipitation (HP) rain-wrapped nature of the storm, allowing significant precipitation to fall in and around the updraft/mesocyclone area where the tornado was located, rather than being blown downwind away from the updraft. This seems a little unusual for an EF5 tornado, as most such environments I have examined (e.g., Greensburg KS and Moore OK) have had much stronger winds at that level, instead of the hodograph doubling back on itself with weaker winds in the 9-12 km AGL elevation range. This exacerbated the visibility issue on a day that was already hazy and murky with moisture and humidity, and may have fed into the number of deaths as residents could not get a visual sense of urgency until the tornado was almost right on them.

Another possible ingredient that may have helped contribute to the tornado was a subtle WSW to ENE boundary that appeared visible in the low-level cumulus field on satellite at early afternoon across the Joplin area (see 4th image above). As the Joplin storm complex moved across or "phased" with this subtle feature, it might have provided some focus for the low-level wind shear and SRH to help spin up the tornadic circulation. But this is only speculation.

The last point I'll make is the complex evolution of the Joplin storm, as can be seen in the radar images in the last graphic above. The original supercell and mesocyclone formed near Parsons KS, but began to fall apart shortly before 5:00 pm CDT over northern Cherokee County KS. At this time, three new cells were rapidly developing on the original storm's right/southern flank; two of these cells ("A" and "B" in the images above) quickly merged, while the southern-most of these newer cells ("C") generated a new mesocyclone around 5:15 pm CDT that prompted an NWS tornado warning for Joplin proper at 5:17 pm CDT. This new mesocyclone produced the Joplin tornado about 20-25 minutes later as it merged into the complex that had been cells A and B. The relevance, if any, of these complicated mergers and rapid evolutions to the intensity of the Joplin tornado is not immediately clear. It can be said that, with each new right flank cell raining into the one to its northeast and merging into the complex, this probably further created visibility problems that helped make the monster Joplin tornado even more deadly.

This intensely tragic event has turned literally thousands of lives upside down. Please consider making a donation to one of several organizations assisting in the Joplin area. See

- Jon Davies 5/27/11

Monday, May 23, 2011

Unfathomable - the 22 May 2011 Joplin tornado

Update, Thursday 5/26/11: The Joplin tornado death toll has risen to 125, making it the most deadly single U.S. tornado since the Woodward OK tornado of 1947. Yet another outbreak of dangerous tornadoes occurred in the Oklahoma/Kansas/Arkansas area on Tuesday 5/24/11, killing several people.

It is now one of the largest single city tornado death tolls in the past 70 years... 89 deaths and climbing. It is the 2nd deadliest tornado event in Missouri history (just behind the 1896 St. Louis tornado in number of deaths). Yesterday's tornado at Joplin, Missouri is yet another massive tragic event in this year of killer tornadoes and huge death tolls.

Shawna and I drove through downtown Joplin minutes before the tornado hit. Sirens were going, but we could not see the tornado, and many people were out and about seemingly unaware or unconcerned. We had been following this storm complex for a couple hours without observing any tornadoes, and though we were a little nervous, we weren't anticipating anything of the magnitude that buried Joplin around 5:45 pm CDT.

We blasted southwest on Interstate 44 to circumvent the approaching mesocyclone on radar that suddenly appeared very strong. As we looked to the north through the trees we caught a brief glimpse of the hard-to-see tornado (see images above, the second enhanced to make the tornado more visible). It was so large and the visibility so poor, that this may have had something to do with why so many people were killed.

Shawna wrote a paper for one of her college classes back in April before the Dixie Super Outbreak on the 27th. In that paper she suggested that, while death tolls from tornadoes have dropped dramatically in the past 50 years due to better warnings and public awareness, that trend might soon begin to reverse again. She noted that "up-close-and-personal" tornado videos may be jading people's attitudes towards the dangers of tornadoes. And, with so many diverse information and entertainment sources distracting people's attention these days, the public may have increasing difficulty "hearing" warnings and grasping the importance of that information through the wall of "noise" produced by our information technology.

Given what has happened in the past month, I wonder if she may well be right. Bonar Menniger's excellent book about the 1966 F5 Topeka tornado, "And Hell Followed With It", details how many people followed and heeded tornado warnings from TV and radio that day (there were only local stations focusing on local information back then). The death toll of only 17 from that evening in Topeka might in some ways suggest a more focused attention and response from the public regarding truly hazardous events back then contrasting with today.

It is clear that the Joplin event was a difficult one to prepare for, with a rain-wrapped hard-to-see tornado forming just west of town when there hadn't been tornadoes from the same storm complex prior to that. And the tornado was huge and violent, around a mile wide going right down through a populated area. But with around 20 minutes of advance warning from the NWS, the death toll is still staggering, and leaves me pondering what could have been done to reduce it. I am so saddened this morning by what has occurred.

I will mention that there were a number of storm chasers who stopped to help with search and rescue in Joplin, including Tyler Costantini and Jay Cazel, Mira Lee, and the Cloud 9 tours group. That's awesome... all those who stopped and helped in the aftermath deserve big credit for doing so.

Shawna's and my thoughts and prayers truly go out to all who have suffered losses in the tornadoes of recent days.

- Jon Davies 5/23/11

Sunday, May 22, 2011

The Reading KS tornado - Kansas weather turns deadly on 21 May 201

Severe weather turned deadly suddenly on Saturday evening 5/21/11 when a tornado struck Reading KS (northeast of Emporia) around dark, killing 1 person, injuring others, and destroying a number of homes. Jim Saueressig's photo above shows the impressive storm structure of the Reading storm after dark.

The environment was one of the most rapidly changing that I've seen on Saturday. At late afternoon, the SPC Sig Tor Parameter (STP) was weak across eastern Kansas with marginal values due to both deep-layer shear and low-level shear being rather weak or marginal for tornadoes (0-6 km shear around 30 kts or less, 0-1 km storm-relative helicity/SRH < 100 m2/s2). This is why SPC issued only a severe thunderstorm watch for northeast Kansas. But by dark, low-level winds had strengthened and backed more than expected, causing much larger shear values, both in low-levels and through a deeper layer. This made STP values sky rocket as the eastern KS environment improved dramatically for supporting tornadoes. See the graphics above comparing STP over Kansas at late afternoon (2200 UTC) with just after dark (0300 UTC), and notice the change in wind profiles from late afternoon to after dark on local RUC analysis soundings at Topeka and Emporia.

The tornado struck Reading around 9:15 pm CDT (0215 UTC), and other tornadoes from the same complex continued to be reported during the following 90 minutes after dark. Notice the RUC 0-1 km EHI and low-level CAPE forecasts above for 0300 UTC, indicating strong CAPE/SRH combinations _and_ a surface-based setting after dark, important ingredients for nighttime tornadoes.

Shawna and I stayed in the local area north of Kansas City watching storms on Saturday evening, and were very saddened to hear about the strong tornado that hit Reading. After an enjoyable week of storm chasing with no deadly severe weather in the plains, last night brought things back to sober reality.

- Jon Davies 5/22/11

Saturday, May 21, 2011

A look at 3 plains severe weather settings 5/17/11-5/19/11

Shawna and I are part of "The Great Tornado Hunt" on The Weather Channel with Mike Bettes and other Severe Studios chasers this month. This blog post looks briefly at the 3 severe weather settings we chased this past week.

The first day involved a landspout tornado in Washington County, Colorado on Tuesday 5/17 west of Cope. We left Kansas City late morning and arrived in Burlington, Colorado late afternoon as a cell blew up west of Akron. Realizing we probably couldn't catch this cell as it headed northward too fast, we drove toward the dryline on I-70 and headed north at Siebert, not expecting much from the mushy looking towers to our west and northwest. One cell did develop a sharp high base, and as we approached Cope, Shawna spotted a dust whirl under it, 10-15 miles to our west. This became a landspout for a few minutes over open country (see photo above). We were lucky to catch it, but a later look back at the SPC mesoanalysis 0-3 km lapse rates and surface-based CAPE fields (see 2nd image above) at 2300 UTC showed the cell we were watching to be in a favorable location where a nose of steep low-level lapse rates (8.0-9.0 deg C per km) and decent SBCAPE values (500-1000 J/kg) were co-located with it on the dryline wind shift. Stretching on the boundary with these ingredients apparently set the stage for a mesoscale accident that we literally bumbled into. The original cell farther to our north was in cooler surface air and less steep low-level lapse rates. Colorado "magic" over elevated terrain is sometimes amazing, with dew points this day only in the low 40s F east of the dryline, but surface heating along the dryline helping to do something interesting!

The next day (Wednesday 5/18) we were tempted to head from Burlington to western Oklahoma where much stronger instability than in eastern Colorado was located, but we weren't excited about the long drive. Looking at model forecasts of the pattern, I noticed that mid and upper level dynamics for lifting were forecast to be much stronger over Colorado, while Oklahoma was near a shortwave ridge at 500 mb / roughly 18,000 ft MSL (see maps in the 3rd image above). Much of that state was also located beneath or south of the southern branch of the polar jet, so I wondered if, even with all the instability, subsidence in that region might keep Oklahoma "capped", or at least keep storms there from intiating until after dark. Because Shawna and I don't get much opportunity to observe upslope storms, we decided to stay in Colorado, and were rewarded with a beautiful supercell that tracked from near LaJunta to Kit Carson, again over gorgeous wide open country (see photos in the 3rd image above). I have to admit we were both surprised that no storms at all developed in Oklahoma, even after dark.

Our final chase day of the week was Thursday 5/19, with a potentially dangerous setting over southern and central Kansas. We started in thick fog that morning at Colby, Kansas, but drove to sunshine in Great Bend at mid afternoon near the intersection of the warm front and the dryline. Signficant tornado parameter values in this area were sizable (> 2.0-3.0, see 4th image above) as storms began to develop just north of Great Bend. But during our afternoon update online with The Weather Channel, both Shawna and Dr. Forbes expressed concerns about the rapid low-level transition in temperatures across the warm front in central Kansas, going from low 80s F to only near 70 F in about 30 miles or so. Shawna also felt strongly that supercells might become high-precipitation (HP) as they moved northeastward into the cooler air north of the front.

I thought the tornado potential was significant, but once again my wife was right! As the cell we followed moved toward I-70, temperatures cooled and the atmosphere became more hazy, reducing visibility. By the time the storm approached Lake Wilson, it was HP in character and hard to see what was going on in the often rain-wrapped area of low-level rotation (see 4th image above). There were possibly some very brief tornadoes in there, but we could not see them, and I have not yet heard any reports of tornado damage from this cell. The contrasting RUC analysis soundings above (last image) at Great Bend and Salina on opposing sides of the warm front showed the issue: low-level convective inhibition (MLCIN) and low-level stability increased rapidly across the front, with large MLCIN making storms increasingly elevated the farther north they moved, even with large low-level shear (storm-relative helicity 300-400 m2/s2). With such a cool low-level thermodynamic environment north of the front, it turned out to be not a strongly favorable set up for tornadoes.

When storm chasing, on most days there are things to learn and refine your knowledge with, if you pay attention :-). And thankfully, there were no strong/dangerous tornadoes in populated areas of the plains last week. Early this next week, Shawna and I will be heading out into the central and southern plains as potential for severe weather and tornadoes picks up again.

- Jon Davies 5/21/11

Sunday, May 15, 2011

Cold-core tornadoes in central Nebraska 12 May 2011

Two weak tornadoes in Nebraska were the first I've intercepted on a "cold-core" storm chase in 5 years, and the first cold-core tornadoes my wife Shawna has seen. Cold-core systems are closed mid-level lows (at around the 500 mb level) that are pools of very cold air aloft, with just enough low-level moisture below them flowing in from the southeast to make the atmosphere notably unstable, although the surface moisture (dew points in the low to mid 50s F) may not appear all that impressive for tornadoes. This event was interesting in that it featured both a landspout tornado near Aurora NE, and a brief supercell tornado near Stromsburg NE.

The surface map and satellite image above at mid-afternoon on Thursday showed a fairly typical cold-core setup (see this paper and this paper for more information about cold-core systems and tornadoes), with a surface low in south-central Nebraska just northeast of the mid-level low, and a boundary intersection area northeast of the surface low in the Grand Island-Aurora-York NE area. This boundary intersection area is often a favored location for tornado development, with a surface heat axis typically protruding in from the south for localized instability enhancement, and low-level shear and helicity often maximized in this area east or northeast of the surface low.

The landspout occurred on a sharp dryline boundary visible on radar as a blue fine line (see third image above) near the boundary intersection area, via non-supercell/non-mesocyclone processes (see this paper). A high-based cell that developed and stetched vorticity on this boundary around 2130 UTC produced a sharp condensation funnel well above ground around 2150 UTC (4:50 pm CDT) that eventually made contact with the surface (see my photos above) with a dust column visible.

A later brief tornado (see last image above) developed with a supercell and mesocyclone featuring much lower cloud bases and cooler surface temperatures farther northeast near Stromsburg around 2340 UTC (6:40 pm CDT) as the boundary intersection area and midlevel cold core system worked eastward. Video of this tornado by Dustin Wilcox can also be found here. This supercell was interesting in that the inflow temperatures were only around 60 F, and the cell structure was rotated 90 degrees counter-clockwise (inflow coming in from the north rather than the east, and the flanking line extending eastward instead of southward from the cell). This made for some spotting confusion at times as cell elements moved northwestward in upper flow ahead of the mid-level low, while new right flank development and eastward evolution of the surface boundary intersection area caused some eastward movement.

NWS Hastings also has some information about the 5/12/11 events here. Thankfully, most tornadoes with cold-core systems are relatively weak (EF0 and EF1) in intensity, as was the case on Thursday with little or no damage reported.

Tornadoes with cold-core systems are rather difficult to intercept, because if there are several cells in the boundary intersection area east or northeast of the surface low, it is hard to know which cell to follow and observe. There are also many issues that can keep cold core systems from producing tornadoes. For example, this same system the prior day on Wednesday produced supercells in northeast Colorado and northwest Kansas, but there were no tornadoes because all cells were occurring north of a sharp stationary boundary where the surface air (40s and low 50s F) was too cold for tornado development.

Computer models are suggesting that yet another cold core system will affect the central plains around the period 5/18/11-5/20/11.

- Jon Davies 5/15/11

Sunday, May 1, 2011

27 April 2011: An extraordinary environment for tornadoes.

Last Wednesday's tragic and historic tornado outbreak (see photos above, and my prior blog post) was unprecedented in the past 75 years of U.S. history, topping even the 1974 "Super Outbreak." The number of deaths are now close to 340 people. It will take a while to sort out the actual number of tornadoes, with NWS surveys ongoing. At least one tornado has been rated EF5 so far (Sunday a.m. 5/1/11).

Why were the tornadoes so strong, so numerous, and on the ground so long? Comparing the 27 April 2011 storm environment to years of tornado database parameters that I've kept, the setting for Wednesday's tornadoes was rare and quite extraordinary.

The scatterdiagram (3rd graphic above) is from research I did with Bob Johns in the early 1990s (Bob is now retired from SPC). We looked at observed soundings associated with 242 F2-F5 tornadoes from the 1980s, and found combinations of storm-relative helicity (SRH, a new parameter at the time that quantitized low-level shear, developed by Robert Davies-Jones at NSSL) and CAPE to be useful in assessing environments with potential for significant tornadoes. Using that data, and with the help of John Hart at SPC, we came up with a parameter called the Energy-Helicity Index (EHI) using the diagram above. EHI is still widely used in tornado forecasting today to assess areas and combinations of SRH and CAPE that can support low-level mesocyclones and, hence, supercell tornadoes.

For reference, on this same diagram above, I've plotted the estimated SRH/CAPE combination points for the deadly 1999 Moore OK and 2007 Greensburg KS tornadoes. I've also plotted the SRH/CAPE environment corresponding to last Wednesday's violent Tuscaloosa AL tornado that killed nearly 40 people (also see the skewT diagram above).

The Tuscaloosa environment location (large red dot) on the scatterdiagram is quite remarkable! It falls in the center and slightly toward the upper right in the area of most optimum SRH/CAPE combinations for tornadoes. Most telling, there are no points on the diagram to the upper right of the Tuscaloosa point, which makes it rather unique. I've looked through my own databases of tornado cases over the past decade, and can't find any tornado environment that even approaches this one (roughly 600 m2/s2 0-1 km SRH _and_ 3000 J/kg MLCAPE). Typically, if SRH is greater than 500, CAPE is less than 2000 (still a very good environment for tornadoes), or, if CAPE is greater than 3000, SRH is only around 200-300 (yet another very good environment for tornadoes). To have both the SRH and CAPE so large, during daytime heating no less (SRH values are usually largest at night with the low-level jet, when CAPE is not so strong), is quite rare and unusual.

Given this information, and the fact that mid-level/500 mb winds were approaching 100 kts on Wedensday (excellent deep-layer shear in addition to low-level wind shear), it is no wonder that tornadoes were strong to violent and on the ground so long. The environment conditions where the storms formed and propagated were probably as optimum as we will ever see!

The final graphic above shows 500 mb maps and 0-1 km EHI conditions approximated/forecast from the RUC model for Monday 4/25/11 (a notable tornado day in TX and AR) and Wednesday 4/27/11 (the historic outbreak). Notice how the midlevel trough stayed in place over the southern U.S. over several days (there were tornado episodes in the South from Sunday 4/24 through 4/27), with the strongest winds peaking on Wednesday (> 100 kts) in association with a strong negative tilt shortwave moving through the trough. Notice, too, from the EHI panels, how much SRH and CAPE were present already on 4/25, setting the stage for the peak combinations on 4/27 with the strong short wave and wind max moving through. The forecast EHI values on Wednesday p.m. 4/27/11 were very large over the MS/AL area, signaling the outbreak to come, though no one could know how historically deadly it would be.

Even with the huge death toll, it must be said again that the National Weather Service and local media did a superior job on Wednesday with outlooks, alerts, and warnings. My heart truly goes out to those who have had their lives destroyed by these tornadoes. I hope anyone reading this will at least consider making a donation to help. See:

- Jon Davies 5/1/11