Ensembles Rock! Literally...Added : Wednesday at 13:50 Any meteorologist will tell you that if you take the control runs of the GFS ► model too literally then that is the fastest route to greying hair and busted forecasts imaginable. The reason is quite simple, remember those machines where you drop 10p in at the top and it bounces all the way to the bottom? What happens when you do it 100 times? That's right, it changes.
Weather models are of course slightly different. The physics in the model are better than simply random fluctuations, but they work on the principle that the data being fed in at the start may not be 100% accurate and so they change the starting conditions slightly for each run using Stochastic Perturbation.
Now, take at look at the GFS ► chart for next Friday :-
You can see high pressure to the North of the UK with a pool of colder air to the Northeast which is just ready and waiting to come South across Europe. But, the overnight runs back away from this. And now, for the 6z run it's back again?!
The charts you look at on the site are the control suite. If you take a look at the 850mb temperatures from the ensembles you can see this :-
See the black line? This is the control run and you can see the output of all the other runs on the same page. Looking at the chart above meteorologists would call the control run a "Weak Cold Outlier" but not by much.
You see, if you look at the variations you can see approx 35% of the other runs showing a warmer outlook, whilst the main "clustering" of output is pretty close to the control run.
When this happens, meteorologists will start putting a little more faith in the control run and as a result we inform our longer range clients that the risk of colder weather is now increasing .
So, remember that the control run is not the be all and end all and you should never take the longer charts as gospel. They should be used against the ensembles to gauge whether or not the control run has much support from the ensembles.
That is unless you prefer having grey hair and busted forecasts all the time.
METEOROLOGIST : DOESNTNEEDJUSTFORMEN |