Forecast accuracy for March 2011
 

What I've done is look at the BBC website's 5 day forecast for my
postcode and compared them to each other. I haven't correlated
whether the weather they forecast actually happened, just how often
the forecasts changed (since a 5 day forecast that changes the
very next day can't be relied on, so isn't any use).

There was a marked improvement in forecast accuracy compared with
February (Correct 1 day ahead: 33%) and the month saw the first
successful 5 day forecast since Nov 8 - in fact there were 2 in a row,
which hasn't happened since October. It's no coincidence that this
month also had markedly fewer different types of weather forecast
than last month (9 forecast, vs. 11 in Feb. 6 occurred compared to 10).
So it's not really surprising that the more settled weather was
easier to forecast.


forecast forecast on: times num of days
-- date -- on day D-1 D-2 D-3 D-4 correct changes ahead
Wed 2 Mar WC* WC GC WC WC 3 2 1
Thu 3 Mar M WC SI WC SI 0 4 0 !
Fri 4 Mar SI SI SI WC WC 2 1 2
Sat 5 Mar M WC WC WC D 0 2 0 !
Sun 6 Mar WC SI WC WC SI 2 3 0
Mon 7 Mar S S S S S 4 0 4
Tue 8 Mar S* S S S S 4 0 4
Wed 9 Mar SI S WC LR WC 0 4 0 !
Thu 10 Mar SI* LRS WC LR S 0 4 0 !
Fri 11 Mar SI* SI LR SI SI 3 2 1
Sat 12 Mar SI* WC WC WC WC 0 1 0 !
Sun 13 Mar SI* SI LR LRS LRS 1 2 1
Mon 14 Mar SI* S SI WC HR 1 4 0
Tue 15 Mar M WC WC WC SI 0 2 0 !
Wed 16 Mar GC WC WC WC D 0 2 0 !
Thu 17 Mar WC WC WC WC WC 4 0 4
Fri 18 Mar LRS LRS LR WC SI 1 3 1
Sat 19 Mar S S S S SI 3 1 3
Sun 20 Mar WC WC WC WC SI 3 1 3
Mon 21 Mar WC* WC WC SI WC 3 2 2
Tue 22 Mar SI SI WC WC WC 1 1 1
Wed 23 Mar S S SI WC LR 1 3 1
Thu 24 Mar SI SI WC WC SI 2 2 1
Fri 25 Mar S S WC SI WC 1 3 1
Sat 26 Mar WC WC WC WC LR 3 1 3
Sun 27 Mar WC* WC S S S 1 1 1
Mon 28 Mar WC* SI LR WC SI 1 4 0
Tue 29 Mar SI SI LRS SI WC 2 3 1
Wed 30 Mar WC LR HR HR WC 1 3 0
Thu 31 Mar SI WC LR WC WC 0 3 0 !

For last 30 days: avg 2.13 different forecasts per day
On average forecasted 1.17 days ahead
Correct 4 days ahead: 3 times
Correct 3 days ahead: 6 times
Correct 2 days ahead: 8 times
Correct 1 day ahead: 18 times (60.0%)
Number of times the weather was same as yesterday(*): 10 times (33.3%)
Number of days every forecast was wrong(!): 8 times (26.7%)

Key:
D: Drizzle 0 occurrences, predicted 2 times over-forecast
GC: Grey Cloud 1 occurrences, predicted 1 times under-forecast
HR: Heavy Rain 0 occurrences, predicted 3 times over-forecast
LR: Light Rain 0 occurrences, predicted 10 times over-forecast
LRS: Light Rain Shower 1 occurrences, predicted 5 times about right
M: Mist 3 occurrences, predicted 0 times under-forecast
S: Sunny 5 occurrences, predicted 19 times about right
SI: Sunny Intervals 11 occurrences, predicted 25 times under-forecast
WC: White Cloud 9 occurrences, predicted 55 times over-forecast


As a sidenote, I now have a full year of forecasts. This is a summary of
the year, in total. I hope to find the time to look into the big picture
in more detail. But for now:

For last 365 days: avg 2.30 different forecasts per day
On average forecasted 0.99 days ahead
Correct 4 days ahead: 29 times
Correct 3 days ahead: 55 times
Correct 2 days ahead: 99 times
Correct 1 day ahead: 177 times (48.5%)
Number of times the weather was same as yesterday(*): 100 times (27.4%)
Number of days every forecast was wrong(!): 97 times (26.6%)

Key:
D: Drizzle 4 occurrences, predicted 20 times about right
F: Fog 9 occurrences, predicted 14 times under-forecast
GC: Grey Cloud 16 occurrences, predicted 74 times about right
HR: Heavy Rain 17 occurrences, predicted 90 times over-forecast
HRS: Heavy Rain Shower 17 occurrences, predicted 48 times under-forecast
HS: Heavy Snow 1 occurrences, predicted 2 times under-forecast
LR: Light Rain 27 occurrences, predicted 109 times about right
LRS: Light Rain Shower 26 occurrences, predicted 111 times about right
LS: Light Snow 2 occurrences, predicted 9 times about right
LSS: Light Snow Shower 1 occurrences, predicted 4 times about right
M: Mist 6 occurrences, predicted 10 times under-forecast
S: Sunny 57 occurrences, predicted 237 times about right
SI: Sunny Intervals 124 occurrences, predicted 490 times about right
SL: Sleet 0 occurrences, predicted 7 times over-forecast
TS: Thundery Shower 0 occurrences, predicted 2 times over-forecast
WC: White Cloud 58 occurrences, predicted 230 times about right

Forecast accuracy for March 2011
 

On Apr 1, 3:39*pm, root wrote:
What I've done is look at the BBC website's 5 day forecast for my
postcode and compared them to each other. *I haven't correlated
whether the weather they forecast actually happened, just how often
the forecasts changed (since a *5 day forecast that changes the
very next day can't be relied on, so isn't any use).

There was a marked improvement in forecast accuracy compared with
February (Correct 1 day ahead: 33%) and the month saw the first
successful 5 day forecast since Nov 8 - in fact there were 2 in a row,
which hasn't happened since October. It's no coincidence that this
month also had markedly fewer different types of weather forecast
than last month (9 forecast, vs. 11 in Feb. 6 occurred compared to 10).
So it's not really surprising that the more settled weather was
easier to forecast.

* * * * * forecast *forecast on: * *times * num of *days
-- date -- on day * D-1 D-2 D-3 D-4 correct changes ahead
Wed *2 Mar * WC* * *WC *GC *WC *WC * *3 * * * 2 * * *1
Thu *3 Mar * *M * * WC *SI *WC *SI * *0 * * * 4 * * *0 !
Fri *4 Mar * SI * * SI *SI *WC *WC * *2 * * * 1 * * *2
Sat *5 Mar * *M * * WC *WC *WC * D * *0 * * * 2 * * *0 !
Sun *6 Mar * WC * * SI *WC *WC *SI * *2 * * * 3 * * *0
Mon *7 Mar * *S * * *S * S * S * S * *4 * * * 0 * * *4
Tue *8 Mar * *S* * * S * S * S * S * *4 * * * 0 * * *4
Wed *9 Mar * SI * * *S *WC *LR *WC * *0 * * * 4 * * *0 !
Thu 10 Mar * SI* * LRS *WC *LR * S * *0 * * * 4 * * *0 !
Fri 11 Mar * SI* * *SI *LR *SI *SI * *3 * * * 2 * * *1
Sat 12 Mar * SI* * *WC *WC *WC *WC * *0 * * * 1 * * *0 !
Sun 13 Mar * SI* * *SI *LR LRS LRS * *1 * * * 2 * * *1
Mon 14 Mar * SI* * * S *SI *WC *HR * *1 * * * 4 * * *0
Tue 15 Mar * *M * * WC *WC *WC *SI * *0 * * * 2 * * *0 !
Wed 16 Mar * GC * * WC *WC *WC * D * *0 * * * 2 * * *0 !
Thu 17 Mar * WC * * WC *WC *WC *WC * *4 * * * 0 * * *4
Fri 18 Mar *LRS * *LRS *LR *WC *SI * *1 * * * 3 * * *1
Sat 19 Mar * *S * * *S * S * S *SI * *3 * * * 1 * * *3
Sun 20 Mar * WC * * WC *WC *WC *SI * *3 * * * 1 * * *3
Mon 21 Mar * WC* * *WC *WC *SI *WC * *3 * * * 2 * * *2
Tue 22 Mar * SI * * SI *WC *WC *WC * *1 * * * 1 * * *1
Wed 23 Mar * *S * * *S *SI *WC *LR * *1 * * * 3 * * *1
Thu 24 Mar * SI * * SI *WC *WC *SI * *2 * * * 2 * * *1
Fri 25 Mar * *S * * *S *WC *SI *WC * *1 * * * 3 * * *1
Sat 26 Mar * WC * * WC *WC *WC *LR * *3 * * * 1 * * *3
Sun 27 Mar * WC* * *WC * S * S * S * *1 * * * 1 * * *1
Mon 28 Mar * WC* * *SI *LR *WC *SI * *1 * * * 4 * * *0
Tue 29 Mar * SI * * SI LRS *SI *WC * *2 * * * 3 * * *1
Wed 30 Mar * WC * * LR *HR *HR *WC * *1 * * * 3 * * *0
Thu 31 Mar * SI * * WC *LR *WC *WC * *0 * * * 3 * * *0 !

For last 30 days: avg 2.13 different forecasts per day
On average forecasted 1.17 days ahead
Correct 4 days ahead: 3 times
Correct 3 days ahead: 6 times
Correct 2 days ahead: 8 times
Correct 1 day ahead: 18 times (60.0%)
Number of times the weather was same as yesterday(*): 10 times (33.3%)
Number of days every forecast was wrong(!): 8 times (26.7%)

Key:
* D: * * * * * * Drizzle *0 occurrences, predicted * 2 times over-forecast
*GC: * * * * *Grey Cloud *1 occurrences, predicted * 1 times under-forecast
*HR: * * * * *Heavy Rain *0 occurrences, predicted * 3 times over-forecast
*LR: * * * * *Light Rain *0 occurrences, predicted *10 times over-forecast
LRS: * Light Rain Shower *1 occurrences, predicted * 5 times about right
* M: * * * * * * * *Mist *3 occurrences, predicted * 0 times under-forecast
* S: * * * * * * * Sunny *5 occurrences, predicted *19 times about right
*SI: * * Sunny Intervals 11 occurrences, predicted *25 times under-forecast
*WC: * * * * White Cloud *9 occurrences, predicted *55 times over-forecast

As a sidenote, I now have a full year of forecasts. This is a summary of
the year, in total. I hope to find the time to look into the big picture
in more detail.

Look forward to it Pete.

But for now:

For last 365 days: avg 2.30 different forecasts per day
On average forecasted 0.99 days ahead
Correct 4 days ahead: 29 times
Correct 3 days ahead: 55 times
Correct 2 days ahead: 99 times
Correct 1 day ahead: 177 times (48.5%)
Number of times the weather was same as yesterday(*): 100 times (27.4%)
Number of days every forecast was wrong(!): 97 times (26.6%)

Key:
* D: * * * * * * Drizzle *4 occurrences, predicted *20 times about right
* F: * * * * * * * * Fog *9 occurrences, predicted *14 times under-forecast
*GC: * * * * *Grey Cloud 16 occurrences, predicted *74 times about right
*HR: * * * * *Heavy Rain 17 occurrences, predicted *90 times over-forecast
HRS: * Heavy Rain Shower 17 occurrences, predicted *48 times under-forecast
*HS: * * * * *Heavy Snow *1 occurrences, predicted * 2 times under-forecast
*LR: * * * * *Light Rain 27 occurrences, predicted 109 times about right
LRS: * Light Rain Shower 26 occurrences, predicted 111 times about right
*LS: * * * * *Light Snow *2 occurrences, predicted * 9 times about right
LSS: * Light Snow Shower *1 occurrences, predicted * 4 times about right
* M: * * * * * * * *Mist *6 occurrences, predicted *10 times under-forecast
* S: * * * * * * * Sunny 57 occurrences, predicted 237 times about right
*SI: * * Sunny Intervals 124 occurrences, predicted 490 times about right
*SL: * * * * * * * Sleet *0 occurrences, predicted * 7 times over-forecast
*TS: * * Thundery Shower *0 occurrences, predicted * 2 times over-forecast
*WC: * * * * White Cloud 58 occurrences, predicted 230 times about right

Thanks again, Paul