Bigger Computers = Better Model Forecasts

May 15, 2013

Big weather news today!

The National Weather Service just announced they would be acquiring supercomputers with over 30 times the power  of the current generation within the next year!  This should level the playing field between the superior European ECMWF model and the Global Forecast System (GFS) models.  It took the Hurricane Sandy disaster on the East Coast to get the funding for the changes out of congress, but good news nonetheless.  It’s been well advertised not only in the meteorological community but in the media how terribly the American model performed versus the ECMWF not only with Sandy but on several other occasions.  The USA has just not been dedicating resources towards computing and it has finally caught up with us (and become a bit of a national embarrassment).


How about those supercomputers…what can they do?

Right now the NWS uses a system with 70 teraflops (70 trillion calculations per second).  Whoa…that’s fast, and it’s a room/rooms with many thousands of processors.  But within a year or two it should be up to 2600 teraflops!

I’m going to direct you straight to Professor Cliff Mass’s blog where he breaks it down in far more detail than I could:

Today was sure a chilly day, only 59 degrees in Portland which was our coolest day in over two weeks.  Not much rain though; no one in the metro area recorded more than .10″…my definition of a “drippy” day.

We should see some convection (rising motions leading to showers) over the hills/mountains and quite possibly over the  valleys tomorrow.  Probably not enough for hail/thunder, but there may be some brief downpours.  We’ll get these showers because we’ll see plenty of sunbreaks, in fact maybe just partly cloudy at times.  So even though it’ll be brighter with more sun and warmer temps on Thursday, it may end up being wetter in some locations.

I’ll admit I’m real bored with the weather pattern through Monday, very little going on with just a few showers or light rain here and there.

A little more interesting Tuesday and beyond next week as a cold upper-level trough settles over the Pacific Northwest for an extended stay (possibly through Memorial Day Weekend).  On the 00z GFS model you see it’s just about the coldest/deepest low in the whole western part of the northern hemisphere!


This setup in general would increase the chances for more active cold showers with hail/thunder.  It’ll also be quite chilly with 30s/40s at night and only upper 50s and lower 60s for daytime highs.

Models are in pretty good agreement on the cold upper level low next week, take a look at the latest 12z ECMWF and 12z GFS ensemble charts.



Well below average the middle of next week, then a slow recovery to normal near the end of the runs.

Chief Meteorologist Mark Nelsen