Wednesday, February 20, 2013

Mid Range Generalities

One of the caveats of discussing mid range forecasts -- those between seven and fourteen days out -- is that specifics are not usually gospel truth or even consistently close but you can talk about the general idea of storm threats or that there's a chance of a storm system moving through.   For the modeling followers out there, worrying about detail or "zomg, it's going to snow because the long range GFS shows it" when modeling error spread past seven days can average over 250 miles isn't worth the exercise of losing sleep at night or stressing over track and detail.  Those finer points can be sweated over if that storm shows up in the near range.

I'll use the example of the two most recent forecast runs of the European computer model, showing the very far end of the forecast range, which is March 1st.  The first map, below, shows a large storm system 250 or so miles out to our east with us on the breezy, but dry, side of the storm as the track was out into the Atlantic and a near miss for us.

Contrast it to twelve hours prior and the midday run yesterday, the Euro track for the low was from south to north through our region with windswept rain across the Delaware Valley and snow for Cleveland, an arguable shift of 250-300 miles in one run.  I posted this map on our Facebook page yesterday to show you one of the solutions that was on the table and to not stress over details.

Those of you who follow the weather and modeling year round see similar "inconsistencies" with tropical systems in the summer.  One computer model run might send a hurricane up the East Coast, another may send the storm out near Bermuda, and reality ends up being in between (a storm that passes about 200-300 miles east of Cape Hatteras)...or even something completely different (going into the Caribbean or a storm fizzling in the open Atlantic due to wind shear).  The farther out from an initial starting point a computer model is, the larger the errors in forecast typically become.  Modeling can occasionally hit a home run in the mid or even 14+ day range -- the Euro was quite reliable with Sandy on most of its forecast runs from the mid range on in although individual runs waffled on the center's track and specifics -- it still had Sandy as a threat to the Mid Atlantic and Northeast more often than not.

That said, if you see other modeling showing something in the "ballpark" (within 250-300 miles either direction of us) in the same range, it is a decent indicator that computer modeling may be onto something for that timeframe.  The GFS from midday yesterday showed a storm out in the Atlantic in a similar position to last night's Euro.  One would not want to get hung up on the specific track this far out but it's certainly ok to think there may be a chance of a storm system in the timeframe of ten days out.  Chance does not mean likely and it's completely plausible that this model hype could end up as a weak clipper or a mere southern storm that misses us by quite a bit by the time reality hits.