Monday, January 31, 2011

Hype and Communication

I love a good winter storm. The kind that cripples an area with ridiculous amounts of snow. I grew up in them, I love to play in them, I love to walk, run, sled in them. They are not rare in the traditional sense but it takes a well timed storm of a certain duration with very heavy snowfall (or ice) accumulation. The combination of all these has not been rare this year. The storm before last in NY, CT, MA saw liquid equivalent snowfall rates (assuming they were accurate in 40-50 mph winds) of 6 mm/hour at their peak at multiple stations. The past storm which dumped even more snow had rates approaching 4-5 mm/hour at its peak. The timing was good with snow peaking in the middle of the night but still snowing through mid morning. I believe both storms lasted around 12-24 hours depending on your precise location.

That leads me to Oklahoma where the hype machine was in full effect, and weather impacts and hazards were being communicated as early as Saturday (watch) and early on Sunday (warning) for a Monday night-Tuesday afternoon storm. It is awesome being in the hype. Shelves at Walmart were bare and people were preparing for weeks without power ... though there must be a super secret handshake I must learn so that I can store that much food before it spoils without power to my fridge.

This is good. The models are doing their job, the forecasters are doing theirs, and TV meteorologists are getting the word, and the impacts, and the story out. I am just not sure that I have a lick about the uncertainty. The fact that weather models rarely converge is not well understood let alone efficiently and concisely communicated.

I looked at what was once a "next generation" tool ... the Short range ensemble forecast (SREF). Only in the last few cycles from the last 36 hrs has the precipitation for Tuesday morning begun to highlight the risk for our area. That risk includes the snow missing us to the east! Or giving us 8-12 inches of snow.

We assume that the ensemble (a collection of perturbed models) covers a sufficient range of realistic scenarios and from this we derive a more skillful forecast. But it is extremely difficult to predict these events precisely because of how localized they can be. A small shift east or west puts (possibly) different communities at risk. By small I could mean 6 inches of snow! These types of scenarios are common in winter storms as they are in summertime thunderstorm hazards.

But what do buy to prepare for a tornado? or a roof smashing hail storm? or a downed tree storm?

What kind of information do you need about uncertainty in the forecast for these situations? For your family? for the local hospital? for the local business?

In some sense the hype is nothing more than a teaching opportunity ... or a learning opportunity. Perhaps this is why we rely on experience ... not because it is always right ... but because we are under the impression we have already learned these lessons. And because those lessons are few and far between (misses being more common than hits) we are suckered into convincing ourselves that we have nothing more to learn. perhaps it is not what action must we invoke, but rather changing peoples minds that they have not learned all the lessons they need to make experience based decisions. Controversial? Probably. But only because it is difficult to convince anyone that they have had blinders on. Until they get run over. Then they have proof.