Monday, September 2, 2013

More on the EF-Scale controversy

I've already had to answer several questions regarding the EF-Scale, so I feel the need to say more on this topic.

It is true that mobile Doppler radar measurements typically are at heights well above the 'standard' anemometer height of 10 m.  In the case of hurricane recon flights, their winds are not measured at 10 m, either, but there is an 'established' procedure for converting those measurements to the standard height.  For tornadic winds, there is as yet no consensus for such a conversion.  There are some indications that in some cases, wind speeds might actually increase downward from where the radars are measuring the wind speed!  And there's little reason to believe that the winds in a tornado follow something like the conventional 'logarithmic law'.  Real tornadic winds are virtually certain to be quite complicated, with enormous changes in both space and time.  The notion of a tornado as a Rankine vortex is typically a grotesque oversimplification of what's going on.  Numerical model simulations, mobile Doppler radar observations, film/videos of tornadoes,  and laboratory vortex models have indicated that many tornadoes are much more complicated than a simple translating symmetric vortex.  Although we don't yet have the capability to map in great detail the winds in real tornadoes over the lifetime of the tornado, it's evident that the picture is mostly much more complex than any simple model can describe.

In the case of the El Reno tornado of 31 May 2013, it's my understanding that the individual subvortices within the large tornado were observed by mobile radars to be rotating at about 75 m s-1 (more than 150 mph!  If the individual intravortex flow adds only 50 mph to that, the result would be 200 mph:  the threshold windspeed for EF-5.  This information doesn't require any extrapolation to a height of 10 m.

I note that a publication exists regarding the Spencer, SD tornado of 30 May 1998 and the relationship between radar-observed winds and damage.  It's only one study, but among the conclusions was that the radar-measured winds converted to F-scale (not EF-scale!) ratings typically exceeded the actual damage at the ground.  A number of hypotheses were offered to explain the discrepancies.  Clearly, much more such work needs to be done and perhaps a consensus may emerge on how to convert Doppler-measured winds to EF-scale ratings.

Another part of the criteria for wind measurements (besides reduction to a height of 10 m) is the use of a 3-second average.  This is a standard favored by engineers for reasons they might want to chime in and explain.  But consider the aforementioned subvortex moving at 75 m s-1- such a vortex that is, say, 75 m in diameter would pass by a house in about one second.  Whatever damage such a vortex would cause to the home would be done mostly within that one second, not by a 3 s average wind!  Doppler radar wind velocities are quasi-instantaneous and extremely rapid changes in those velocities are seen even at sampling intervals of 2-s!  What meaning does the 3-s criterion have in the context of such rapid time changes in the wind speed?

After the Jarrell, TX tornado of 27 May 1997, which was rated F-5, some engineers argued that the slow movement of a large tornado magnified the damage beyond what one would expect from the winds alone - that is, winds blowing for a long time would be more damaging than winds sustained only briefly.  This seems reasonable, but what about a wind that accelerates extremely rapidly?  Could not that also enhance the damage potential?  The duration of the wind likely has some impact on the damage, but the real relationship of wind duration and damage isn't necessarily simple.  The aforementioned paper on the Spencer tornado discussed this, among other factors regarding the complex relationship between wind and damage.

Since I mentioned the diameter of a subvortex - just how does one measure the width of a tornado?  The El Reno tornado was claimed to be 2.6 miles in diameter, supposedly a record surpassing that of the tornado that hit Hallam, NE after sunset on the evening of 22 May 2004.  But how does one define the width of a tornado?  The damage doesn't have a hard edge to it, so even if you're driving at right angles to the track, how can you tell where tornado damage begins/ends?  Try it sometime.  It's not so easy as you might think!!  Debris is often centrifuged out of the tornado, so the presence of debris doesn't define the edge of the track.  Insofar as I can tell, tornado width estimates in Storm Data are probably even more inconsistent than F/EF-scale ratings.

So fast forward to a day when we can have continuous wind speed information all along the track of a tornado (not in my lifetime!) - imagine we can have a complete picture of the time-space history of the wind.  Still, where does the tornado begin and end?  Is there an arbitrary wind speed that defines a 'tornadic' wind?  How does one distinguish between winds in the rear flank downdraft (which usually is adjacent to the tornado) from the 'tornadic' winds?   One thing is for certain:  the edge of a tornado is not at the edge of the condensation funnel!  Remember - the tornado is the (invisible) wind, not the cloud. 

Given the complexity of comparing winds to damage, it seems to me that if we can obtain wind speeds from measurements, we should seek to find ways to use them, rather than to ignore them.

One final word:  there's a question about who 'owns' the EF-scale.  Who has the right to modify the rating criteria in light of new observational capability and/or new science?  At the moment, the EF-scale is something that was created by a process involving scientists and engineers, resulting in a document that forms the basis for how the scale is implemented within the National Weather Service (NWS).  But the NWS doesn't claim to 'own' the EF-scale and they should not.  No one does, at present.  It properly belongs to the whole scientific and engineering community.  Efforts are underway to establish a systematic, inclusive process for modifying the criteria as new science is available.  I can't say much about it, as it's still in the formative stages, but I sincerely hope it eventually can be recognized as the place wherein the EF-scale can become a 'living' process, rather than a set of criteria frozen into a document.  Hopefully, within such a process, many of these issues will be resolved and we can move forward to take advantage of new capabilities and new science/engineering.