One of the most significant tornado outbreaks of the modern era occurred on Wednesday in much of the East, particularly in Mississippi, Alabama, Georgia, and Tennessee. The number of fatalities has exceeded 300, with still hundreds unaccounted for. Such a tragically high number of deaths has sobered the meteorology community, particularly given the relatively excellent forecast of the event by today's standards.
My thoughts on the outbreak, in general, have been proffered by others. However, at the very least, they deserve repeating, as the last thing the meteorology community and the public can afford is a slow dwindling of the memory and a repression of the tragedy. Complacency is unacceptable, given the appalling number of fatalities from Wednesday's event.
Although it is obviously premature to propose the reasons (and their relative importance) behind the number of fatalities, it is fairly likely the reasons were in the communication arena rather than the meteorological. Many, if not most, of the tornadoes had substantial lead-times, given each storm's ease with which it produced tornadoes. The Storm Prediction Center (SPC) had heightened awareness of the event, even out to five days in advance, and by the 0600 UTC Day-1 convective outlook, had issued a high risk. Particularly dangerous situation (PDS) tornado watches soon followed, and NWS warnings were inundated with "heightened" text. News media were covering the event very well. So what went wrong?
I offer a few thoughts...
1) It is clear to me that antecedent convection played a role in one almost undeniable way and one potential way. First, much of northern Alabama was without power after early morning convection from a severe mesoscale convective system (MCS). The MCS produced substantial wind damage and even a few tornadoes that morning in the Tennessee Valley, contributing to widespread power outages in much of rural northern Alabama. Thus, there was little if any means of communication with many of those affected by the subsequent tornadic supercells. It is very likely that these power outages contributed to the loss of life totals, particularly in northern Alabama. The second way, I suspect, is that many people have the "one-and-done" philosophy with severe convection. Once severe thunderstorms affected northern Alabama that morning, I suspect there was reduced interest in impending weather and heightened interest in recovery. This may stem from preconceived notions that multiple rounds of severe weather are very uncommon within a 24-h period. I am somewhat unclear on how this possibility was conveyed to the public at large, but it was reasonably well forecast by the SPC and NWS offices in the affected regions -- nevertheless, the frequency with which multiple rounds of high-impact weather affects the same location in a 24-h period is possibly low enough to contribute to reduced awareness, particularly when there is limited means of increasing that awareness thanks to the power outages.
Because power outages almost undoubtedly contributed to loss of life in this event, we must find a means of conveying the necessary information to the public without the luxury of electricity. This potentially excludes (outdoor) sirens, which likely depend on available power. (More on sirens in a later blog post.) Cell service is potentially helpful, but given the potential for cell service to be disrupted and the questionable availability of cell service in particularly rural portions of the region, this is only one potential option. Note that NOAA weather radios can (and should) be battery powered, but these are not immune to power outages either, as the Huntsville NWS has discovered. Clearly, this should be a primary focus in the communication arena to prevent a repeat of this event.
2) It is downright irresponsible to claim that it is impossible to survive an EF4/EF5 tornado, given the imagery of the tornadoes affecting Smithville, MS; Tuscaloosa, AL; and Birmingham, AL. Although there is some evidence that entire housing was essentially completely destroyed and displaced from foundations, even with reasonably strong grounding, making this claim can lead to the unwise decision simply to refuse to take cover in subsequent scenarios. The "intervention of destiny" syndrome is NOT what the public should be led to believe when tornadoes occur. Although being underground may have been the only way to survive in exceedingly rare circumstances of these tornadoes, the public should be encouraged to take cover in the best way they possibly can.
On a side note, it is my belief that the best way to survive a tornado is to get out of its path. In this manner, this means driving in a direction perpendicular to the track of the tornado well in advance of the tornado's approach. As a meteorologist, this is relatively simple knowledge to obtain. It certainly could have led to a better result than the notorious Andover overpass incident, for example, which subsequently led to loss of life in the 3 May 1999 tornado outbreak. However, I worry when some bring up this prospect when urban areas are in the path of tornadoes. I suspect that mass panic may result, with a large number of drivers erratically and irresponsibly taking risks to avoid the tornado, which they may not know how to do anyway. The last place people want to be is in a traffic jam as the tornado approaches. I caution people who propose this idea, at least until more research on this potential shift in philosophy is conducted and completed.
3) It is time to increase research regarding increased population and potential impacts from tornadoes. As most in the severe weather community are aware, nonmeteorological artifacts inundate the severe weather reports archive. A relatively untested artifact is the magnitude and trend of population density -- would this tornado outbreak have resulted in such substantial loss of life in the Great Plains, or if the tornadoes were displaced x number of miles north/south of their actual tracks? Such research can provide insight on the uncertainty regarding the overall impacts from a meteorologically significant severe weather outbreak (or other type of hazard, such as a hurricane or winter storm), and also may provide an improved means of comparing this event to those of the past (e.g., 3 April 1974).
4) Although there is some statistical evidence that there is at least locational dependence of major tornado days to ENSO, it is scientifically (and certainly statistically) questionable to associate individual severe weather outbreaks, and even collective significant severe weather outbreaks, to this weather-climate phenomenon. The sample size of the outbreaks is too small, and their dependence on so many other variables, some of which we still are not completely aware of, too great to make such a daring conclusion. My thoughts, then, on associations with global climate change should go without saying (and thus, will go without saying).
5) It is absolutely imperative that the meteorology and social science community make a considerable effort to identify the breakdowns that led to the calamitous casualty count, and propose solutions to these breakdowns. There is no excuse! It is our responsibility as scientists and as forecasters. Increased effort to research, rather than speculate, on these causes is of utmost importance. Tornado outbreaks of this magnitude will happen again, and if current trends continue, even more people may be affected in the future. Although loss of life in these types of events is virtually certain, we can and must do more to ensure that we can limit, if not completely eradicate, that number.
Such effort includes breaking down and solving the inherent complexities of such interdisciplinary research, eliminating science-vs.-social science nonsense, and refusing to fall into the trap of assuming we know the answers without actually identifying them scientifically. Our attempts to bridge these gaps so far have not been adequate. We can and must do more.
Although the science questions of this case are plentiful (e.g., what processes on the mesoscale occur to promote nearly universal development of tornadoes with supercells on days like 27 April versus a much smaller ratio on others, what antecedent environments appear to exist with tornado outbreaks and are these environments relatively few and predictable, etc.), the social science aspects appear to be just as important, if not more so. The tragic events of 27 April 2011 should be a call to arms in the severe weather community!