School closings in tornado hazard situations, Part 1

Recently, my colleague Dr. J Marshall Shepherd, has raised questions about policies regarding school operations during tornado situations.  This issue is far from simple.  It involves many complex topics and I want to discuss at least some of those topics.  This will be somewhat longer than my typical blog, but it necessitates some detail.  My bottom line is that any particular "answer" to Marshall's questions is extremely unlikely to be appropriate in all possible circumstances.  In other words, I doubt seriously there's any "one size fits all" procedure for deciding what to do.  Therefore, any decision regarding school responses to a threatening tornado situation depends strongly on the circumstantial details.  And that includes not only the choices by the school administration, but also all the individual families including the schoolchildren.

First, a review of products from the National Weather Service [NWS]:  the Storm Prediction Center (SPC) in Norman, Oklahoma issues various severe weather forecast products from a few hours in advance of a hazardous storm event, to several days ahead.  The quality and accuracy of their forecasts have been improving over the decades.  Their forecasts are not perfect but they can provide substantial value for decision-makers when used properly.

That raises an interesting point - at least interesting to me and some others - just how do decision-makers use a weather forecast that is inevitably uncertain, especially in terms of intensity, and temporal/spatial specificity (i.e., exactly when, where, and how strong will the event be?).  It isn't possible for any forecast by anyone to provide such detail accurately and consistently.  How does a user of this information make use of forecasts if it's known (and it is) that all forecasts have greater or lesser uncertainty.  If the forecasts were perfectly accurate in all details, then the decision-maker's decision is made by the forecaster!  The user then would know exactly what will happen and can make decisions easily.  I understand why most users want this to be the case, even though they know better.  People in Hell want a glass of ice water, too!  No forecaster can do this, so it's illogical to expect that it is possible.

Moreover, a decision-maker must incorporate more information than just the weather forecast in making a decision -- factors that public sector (i.e., NWS) forecasters in general know little or nothing about.  Some decision-makers, because of their circumstances, need a lot of advance warning in order to take appropriate action.  Others can get by with much less lead time.  Some have prepared shelter positions at immediate hand, others do not.  Every user has unique circumstances.  There's no way for NWS forecasters to know all the external factors that govern a weather forecast user's needs, so the forecasts simply can't be used as if they were somehow perfect.  Perhaps a few user/decision-makers might be able to pay for the services of a private sector forecaster to make their weather decisions for them based on shared information so the forecasts know precisely what are their needs, but that's just not possible for NWS forecasters.  Instead, NWS forecasters provide a forecast that is their best estimate of what will happen, and - ideally - supply some understandable information about the uncertainties of their forecast.  A user then - ideally - merges that forecast with all the other information needed to make a decision.

Consider the implications of a Severe Storm Outlook issued by the SPC:  this product delineates the area expected to experience severe weather a day or more in advance, to allow users to begin to prepare for the possibility of experiencing a hazardous storm.  For any given location within the area designated, during the time when the forecast is valid, there is some generally unknown probability of experiencing that hazard, but the SPC seeks to assign a probability based on their understanding of the specific weather situation.  In general, it is quite far from a time/space/intensity-specific prediction, of course.  Even a 5% probability at this point in the weather situation is actually a relatively high value.  On any given date, the probability of a severe weather event on the average is far below that 5% value!  In most circumstances, the odds of any given location within the outlook area experiencing a tornado hazard is too low to take any actions, but users might best be served by preparing to take action when/if a hazardous situation arises.

When the weather situation evolves toward the imminent development of severe weather, the SPC usually issues a Severe Storm Watch that includes some information about the specific probability of a tornado within the space-time volume of the watch, and also some indication of the expected intensity of any tornadoes that might occur.  This can include what is described as a Particularly Dangerous Situation (a so-called PDS Watch) that includes the potential for long-track, violent tornadoes.  In the usual PDS tornado watch, the probability of having a significant tornado somewhere within the watch increases to some value, perhaps as high as 80%.  With such a high probability, this might be sufficiently threatening for some users (those whose protective actions require extra time), but certainly not all, to commence their tornado precautions.

Finally, if a tornado has been detected in some way (often based on radar information), local NWS offices may issue a tornado Warning.  Even in such cases, any specific location ahead of the tornado may or may not be hit.  The typical size of a tornado-warned area is considerably less than that of the typical watch and the existence of a tornado is of far less uncertainty than that associated with a watch, so most people within the warned area are well-advised to take tornado precautions, but even in this dangerous situation, there is no guarantee of anything:  a tornado could change intensity, dissipate, or swerve off in a new direction.  Most long-track violent tornadoes roll along a more or less straight path for many minutes (up to an hour or more), but each storm case is different and not all tornadoes are "typical".

... to be continued