Weather Banter - View Single Post

Doorlish · November 20th 09, 09:28 PM posted to uk.sci.weather

what utter self indulging bull****

Dawlish wrote:

Have a look at the stats, David. Start in 1988. That gives a
reasonable 21-year dataset. 75%+ accuracy (actually 17/21 years were
warmer than their respective 30-year average = 81%) with saying
"warmer than average" every year since then. I've never found anything
better than this for predicting anything about the UK (OK, Central
England) winter. You can almost go back to 1983, but the accuracy
drops to slightly below 75%, but stays above 70% (19/26 winters warmer
than average = 73%). Amazing stats really, aren't they?

I'm not at all sure Joe *******i understands what he's up against with
that prediction of his. 3 months all below the 1971-2000 average in a
UKGW winter (OK, my phrase, but it does acknowledge the UK following
the global trend) is very unlikely to happen - unless the GW trend has
ended. He'd have to be very, very, lucky to nail that prediction.

http://www.metoffice.gov.uk/climate/...ics/uktemp.txt

The 21-year hindsight odds would be about 1/4 a warmer than average
winter, but I'm the bookie, so I'd only offer 1/3. Thus; 3/1 a colder
than average winter this year. Any individual month = 3/1 against
having a temp below the 30-year monthly average (should be 4/1
against, but again, I'm the bookie!). The odds for getting 3
consecutive months below average should equal 3x3x3/1 =27/1 (really it
should be 4x4x4/1 = 64/1). However there is an element of persistence
in that (if a month is colder than average, the next month is more
likely to be as the cold may well persist into the next month and
reduce the monthly mean, so 5/2 the second consecutive colder month
would be my odds. If two months are below average, we may well have an
unusual blocking situation and the persistence factor would increase
the chances of the third month being below average, IMO, so the odds
for a 3rd consecutive month would fall to 6/4.

Thus the odds I'd offer on 3 months being colder than average would be
3/1 x 5/2 x 6/4 = 3 x 2.5 x 1.5 = 11/1. The *actual* hindsight odds,
from the Hadley stats would be 4/1 x 3/1 x 2/1 = 24/1

So; 24/1 against are the odds that I think Joe Bastadi is facing in
saying that all three months of this winter would be below average.
Good luck to him. As the bookie, I'd offer 11/1 against his forecast
being spot on - if you'd like to place a bet!