Friday, March 12, 2010

"Cold Core" discussion and comments 3/8-3/10/10

I've gotten lots of requests to discuss the two "cold core" systems that came through the central U.S. this week. Was the Hammon OK tornadic storm a "true cold-core-type" supercell? Why did the Monday 3/8 system generate tornadoes in western OK, but no severe weather near Kansas City the next day (3/9)? Why didn't the next system on Wednesday 3/10 produce tornadoes in Oklahoma or Kansas? There are not easy, concrete answers to any of these questions. It goes without saying that the atmosphere is enormously complex and something we'll continue to struggle with understanding. I'll try to offer a few thoughts and opinions here in this rather long discussion.

First, labels like "cold core" (coming from Ernest Goetsch's 1988 SLS conference paper, and Mike McDonald's 2000 Canadian paper) are clearly problematic. Labels may help get us in the general "ball park" so that we can begin discussing settings and issues without having to go into long preliminary descriptions each time we mention them. But labels mean different things to different people, and come with very different assumptions depending on the person and their experience. Certainly, the atmosphere doesn't recognize labels or behave according to them.

In my work on the subject (e.g., this 2004 paper with SPC's Jared Guyer, and this 2006 paper), I've stuck with "cold core" as a loose way to refer to tornado activity near 500 mb closed lows where surface dew points may look deceptively small for tornadoes to occur. People can argue what's "near" and what's "not near" 500 mb closed lows (Jared and I used roughly 200 miles as a subjective but practical range for most of the cases we examined, including several "missed" tornado forecasts), and they can argue whether a particular case is a "true cold core case" (whatever that is). But I do think that misses the point. Significant tornadoes with surface dew points below 60 deg F are rather rare (in my database of 2001-2005 events, only 17 of 255 significant tornado occurred with a dew point below that!). The EF2 Hammon OK tornado occurred in an area where surface dew points were 52 deg F at best (see 3/8/10 surface map above), near the northwest edge/tip of the warm sector. It's clear that the associated cold air advection aloft (see -20 C 500 mb isotherm drawn over 3/8/10 surface map above) was a key component in generating the instability needed for that tornadic storm. So, if the "cold core" label does nothing more than serve as a flag or reminder for forecasters to at least consider the possibility for tornadoes within roughly 200 miles of closed upper lows when surface dew points appear "marginal" (e.g., upper 40s-low 50s F), avoiding a "surprise" when reports initially come in, then it has done it's job. If others want to do studies and papers to better define what a "true cold-core" tornado event is, I encourage that. But beware getting caught up in labels and definitions at the expense of practical meteorological assessment and experience.

Several people have asked me why the Hammon supercell appeared so much larger horizontally with more "typical" structure (e.g., large RFD etc., see Charles Edward's excellent photo above) than many low-topped supercells near 500 mb lows. It's important to note that the western Oklahoma late afternoon environment on 3/8/10 had plenty of clockwise curvature to the wind profiles, with 0-1 km storm-relative helicity (SRH) at least around 200 m2/s2 (not shown). Add to that around 1000 J/kg of surface-based CAPE (not shown), and the atmosphere doesn't care whether the storm has a "cold core" look to it (whatever that is). The shear/CAPE combinations were certainly large enough to support a significant tornadic supercell oer western OK in this situation. For what it is worth, cells occurring much closer to the midlevel low will tend to look "smaller", both vertically and horizontally, but there are certainly no hard and fast rules.

Why didn't the same system produce severe weather and tornadoes/funnels in the Kansas City area the next day? No one can say for sure, but there was almost a complete absence of sunlight with clouds surrounding both the surface and broadening/distancing midlevel low (see 3/9/10 satellite image above). This lack of heating for instability, along with with less focus and definition to both the surface and upper systems, might account for the lack of severe reports. There was one possible funnel report near Cameron MO, and the southern cloud base of a cell passing over my house north of Kansas City had some good motion in it. But the lack of sun and a well-defined dryline/clear slot kept Shawna and me from chasing that day. I will observe that, if "cold core" systems are tornadic near the 500 mb low on day 1, day 2 rarely results in tornadoes within roughly 200 miles of the midlevel low. This probably has to do with how difficult it is for such systems to maintain a favorable focus, juxtaposition, and organization near the midlevel low as they evolve over a couple days time.

Finally, several people have asked me why there weren't tornadoes in northern OK and southeast KS on Wednesday 3/10/10, (the Arkansas tornadoes after dark were deep in the warm sector away from the 500 mb low and Pacific front/dryline, in a good shear/CAPE environment with surface dew points around 60 F). Shawna and I chased that day (see my blog on 3/11/10), and noted several things seeming to go wrong as the day evolved. First, the surface moisture was having trouble returning, which resulted in initial warm advection storms at early afternoon being very high-based over southeast KS way out ahead of the surface low. Enhanced evaporative cooling below these high cloud bases generated outflow and westerly winds in the Coffeyville KS area (see 3/10/10 surface map above), where one would have otherwise expected a broad fetch of east or east-southeast winds spiraling into the northeastward-moving surface low over northern OK and then southern KS. This reduced SRH and low-level shear, retarded surface heating and surface moisture, and just generally "messed up" what might have been a well-organized "spiraling" surface pattern. This midlevel system was so strong that it still generated a photogenic arc of low-topped convection with sunshine behind it (see 3/10/10 satellite photo above) that moved into Chautauqua and Cowley counties of KS, producing several brief funnels from the cold air aloft and lots of pea-sized hail. But the outflow air from these cells (we measured 38 deg F!) seemed far too cold for supporting tornadoes.

As I mentioned in a talk at Denver a few weeks ago, there's much more that can go wrong than can go right with so-called "cold-core" systems and their ability to produce tornadoes. At best during spring or fall, maybe 1 in 10 will produce tornadoes within 200 miles of the midlevel low, and one often won't understand very well what's evolving in real time. I guess that's what makes these systems so interesting and such a challenge to forecast.

- Jon Davies 3/12/10


Adam L said...

Excellent write up Jon! I like your take on things. I always tell people the best forecast is still just an educated guess. We like to think we know what will happen but we don't and that is why we are out there studying and observing!

Ted Keller said...

As always, a good explanation with time taken to save and annotate the relevant maps. We went out on the 9th and 10th with of course not much to talk about. Thanks Jon!

Jon Davies said...

Adam and Ted:

Thanks for your comments and encouragement. I'm always trying to learn something useful from these events, but this past week made for some very difficult forecasts!


Lanny said...

Jon, another great write up for sure! I find myself checking your blog just about after every event with the hopes that I can learn something and apply it to this crazy hobby and the field. I encourage and ask that you keep posting your thoughts and certainly helps me to better understand things and I appreciate your time!