Tuesday, September 17, 2013

Follow-up on the Colorado Floods

It's come to my attention that the NWS forecasts for the major flash flood event in Colorado failed to anticipate the disaster that was about to unfold.  And it seems the numerical weather prediction models did a pretty decent job in forecasting substantial precipitation in the area.  Thus, this seems to be a case where the human forecasters subtracted value from the model output, rather than adding value. Learning of this outcome is extremely discouraging to me; this kind of outcome in the face of a severe weather event is precisely what I've spent the last 30 years of my life trying to prevent with forecaster training courses based heavily on what we were trying to do in the Flash Flood Forecasting Course (FFFC).

When the NWS drops the ball by not anticipating hazardous events, the whole chain of responses to those events is threatened.  The core mission of the NWS is to protect lives and property, so by failing to anticipate damaging, life-threatening events, the NWS fails in its most basic responsibility.  Good weather forecasts make a positive difference, but poor forecasts have negative impacts.  

In my last blog, I went on at some length about the origins and content of the FFFC, which was discontinued in the late 1990s.  The original motivation was to prevent more tragedies like the Big Thompson flood of 1976.  Research studies by Bob Maddox, Charlie Chappell, Mike Fritsch, Fernando Caracena, Richard Grumm, Wes Junker, and others provided not only physical understanding of the processes by which flash flooding occurs, but also a historical perspective on the distribution and frequency of flash floods.  We learned the ingredients for a flash flood, which is truly a hydrometeorological event, involving both meteorological and hydrological aspects.  The knowledge is there, ready to be applied in operations.  "Surprises" of this sort simply should not happen.

During our 2-day lectures (as part of the 2-week FFFC), we had very little time to accomplish much in the way of advanced concepts, but the insights and instructional skill of Bob Maddox meant we packed a lot of content into that short time.  We had to do remedial education in the basics of convection, as well as training in the application of physical concepts to real-world forecasting practice.  Most of the forecasters were ill-prepared to move into more advanced topics, so we had to stick to the basics.  Even then, we were constantly challenging forecasters to re-think their understanding.  I still do these things today when I do a training course.

After the end of the FFFC, I was asked to help develop a distance learning module for COMET and the module was supposed to 'duplicate the outcome of the FFFC'.  When I asked what that outcome actually was, since I was unaware of any effort to investigate that topic in any meaningful way, I was greeted with a stony silence.  I declined the opportunity to contribute, on principle.  I didn't believe that any distance learning module could duplicate what we did in our two days.  And I still don't believe that it's possible.  Events in Colorado this year seem to confirm my opinion of distance learning.  What the NWS does for forecaster training isn't even remotely adequate and this tragedy has underscored my concerns.

In the time since our last participation in the FFFC, it's pretty likely that most of the folks we had in the FFFC are no longer sitting at operational forecast desks.  I'm deeply disappointed to learn that the 'corporate memory' within the NWS evidently no longer includes what we tried to impart during the FFFC.  In a world where NWP models are increasingly important to the forecast, it seems that at least some NWS forecasters are no longer capable of using models and the science of meteorology to produce a forecast superior to the models alone.  I've written extensively about where this trend is likely to lead us - see here and here and here and here, just for starters.  It's not something one can contemplate with much confidence that something is being done to prevent the trend from taking us where we don't want to see it go.

Those of us who spent so much time and effort trying to do something about flash flood forecasting are angry and frustrated about this massive flash flood forecast failure.  The forecasters must deal with having a 'defining moment' characterized by failure, but NWS management has to accept its responsibility for this case and do some internal soul-searching to seek a meaningful solution to the problem it represents.

Monday, September 16, 2013

Thoughts on the Colorado floods

In the late summer/fall of 1976, I had graduated with my doctorate and was working at the National Severe Storms Forecast Center (an earlier incarnation of what is now the Storm Prediction Center), in Kansas City, MO.  A colleague was on a late summer vacation in the Colorado Rockies, when on 31 July, the Big Thompson Canyon in Colorado was swept by a flash flood that killed 143 people.  My friend escaped and returned home safely, but a post-event investigation revealed it had not been well-forecast.

After the Big Thompson debacle, the National Weather Service instituted a Flash Flood Forecasting Course (FFFC) at the NWS Training Center, taught by Drs. R. A. Maddox and C. F. Chappell, who had studied the Big Thompson event, as well as many others.  Their established expertise in flash flood events made them logical instructor choices.  In the fall of 1982, I transferred from NSSFC to the Weather Research Program in Boulder, CO, with Dr. Maddox as my boss.  I was reluctant at first to get involved with the FFFC but eventually complied with his request, mainly because he was being swamped with duties.  I eventually came to embrace the course and its goals - our participation was for only 2 days out of a 2-week course, but since it focused on basic principles and application of them to forecasting, it was a great way to learn how to do effective forecaster training.  Dr. Maddox was a great instructor and, while I had to do things to match my ways, I openly copied many things I saw him doing, because they made perfect sense.  I'm still training forecasters in short courses the same way.

I learned many things about flash floods during the years spent teaching the FFFC (which eventually was discontinued), and on the occasion of the 20th Anniversary of the Big Thompson event, a symposium was held in Fort Collins, CO, that included a tour of the canyon.  The tour was a very humbling and sobering experience, and it gave me a great deal of respect for the massive power of moving water.  And added a powerful sense of urgency to my training lectures.  Since then, I continue to be troubled by our society's collective inability to grasp the significance of flash floods at a personal level.  Whereas tornadoes are relatively exotic and scary, everyone has experienced heavy rainfalls and modest flooding, so it seems that many have no respect for what moving water can do when the rainfall is sustained and the flooding becomes massive.  We've made some progress since that fateful day in 1976, but we have a way to go, yet.  Communities need to plan for high-intensity flash floods, such as what to do when roads are washed away and power is disrupted by the floods.  Regulations need to be developed and enforced to mitigate factors that serve to increase the threat, such as washed-away propane tanks.  Building on floodplains should be forbidden.  And so on.  We are not yet a 'weather ready' nation - not even close!!

One thing was clear at the 20th Anniversary Symposium:  the Big Thompson Canyon flood was not a "freak" event.  Something of the sort happens nearly every year somewhere in the high terrain of the western USA, differing only in the magnitude of the event.  People in the know have been saying for decades that Boulder, CO was long overdue for a big event ... and this year, the time had come.  Although not the same in detail as the 1976 Big Thompson flood, the ingredients were there for an event and it has come to pass with devastating effects.   As I write this, there are more than 1200 people yet unaccounted for well after the floods, and at least 5 known fatalities.

This event was foreseen decades ago - although it wasn't known precisely when and how it would happen, of course - but people this week are saying things like "We've never seen anything like this!  We had no idea things could get this bad here!"  This betrays a significant shortfall in our communication of risk to the public.  No one should be so ignorant as to believe that what they've seen in their short lives is the worst the planet can produce.  No one should be so ignorant of the illusory nature of their security when confronted by the forces of the natural world.  As I write this shortly after the 2013 CO floods, it's virtually inevitable that an event much worse than anything yet recorded in all of human history will happen someday, somewhere.  What is written in that history shows that major damaging events have happened in the past, and the lesson to be learned is they will happen again, and might well be even more devastating than anything we've experienced before!  It's not 'scare tactics' to convey an accurate risk assessment.  We owe the public an accurate understanding of the threats they might have to confront.  Evidently, we have yet to accomplish that.