During the recent unsettled spell I tabulated the grid-point rainfall
predictions from the MRF model representative of Luton for each
successive model run for complete rainfall events (i.e. the rain
resulting from the passage of a depression or frontal system,
rather than discrete 6 hr periods):

Event
28/29Oct 30-31Oct 2Nov 3/4Nov
Run
T-96 - - 8.5 0.0
T-90 - - 3.5 0.5
T-84 - 0.0 6.5 0.0
T-78 - 0.0 3.5 0.5
T-72 - 0.0 18.0 0.5
T-66 - 0.5 8.5 0.5
T-60 - 9.0 5.5 0.5
T-54 - 6.5 7.0 0.0
T-48 - 27.0 3.5 0.5
T-42 25.5 21.5 7.5 1.0
T-36 17.5 24.0 13.5 0.5
T-30 12.0 25.5 10.5 0.5
T-24 7.5 28.5 11.5 0.5
T-18 11.0 32.5 11.5 0.5
T-12 8.5 21.5 8.0 0.5
T-6 8.0 28.0 7.5 0.5
T-0 7.5 24.0 8.0 0.5

Actual 9.9 20.7 11.1 0.2

(T-0 is the last run before the rain began; T-96 is four days
before the rain began)

Having grown up in an era when trying to predict rainfall
amounts even 24 hours ahead was regarded as laughable,
I find these really quite impressive.

Philip Eden

Having grown up in an era when trying to predict rainfall
amounts even 24 hours ahead was regarded as laughable,
I find these really quite impressive

I agree. This is quite a laborious exercise to carry out, which explains
why I never do it. Is this accuracy typical of such predictions? My own
suspicion is that it is not but I don't have figures to support this view. I
don't think the same success would be found in non-frontal situations,
particularly in summer, but that is a notoriously difficult problem and again I
can't prove my suspicions.

Tudor Hughes, Warlingham, Surrey.

"Philip Eden" philipATweatherHYPHENukDOTcom

Having grown up in an era when trying to predict rainfall
amounts even 24 hours ahead was regarded as laughable,
I find these really quite impressive.

Those figures are certainly impressively accurate, but as Tudor Hughes has
pointed out, quite laborious to produce the predications.

Philip. Are you able to explain how you made the calculations and from what
raw data? And presumably the whole thing could in theory be computerised
(automated) so that I could simply call up a location (in my case, almost
exactly 52N 000E/W) and see some predications?

And as Tudor further pointed out, with convective shower rain prediction - a
huge bonus - it is difficult to imagine how any level of accuracy would be
possible. Just as an example, I was out for a walk last Sunday afternoon
and could see a distant shower exactly upwind of me (I was watching the
movement as it got closer - coming directly towards me. Was I going to get
wet?) But by the time it had "reached" me, it had fizzled out completely so
little more than a decayed shower cloud was apparent. Now could anyone
predict with accuracy that I would not get wet? Yes, knowing exact location
of the shower, its expected lifespan, etc, would have worked on the day, but
at T-24? That must remain a dream.

Jack

"Jack Harrison" wrote in message
...


Those figures are certainly impressively accurate, but as Tudor Hughes
has
pointed out, quite laborious to produce the predications.

Philip. Are you able to explain how you made the calculations and
from what
raw data? And presumably the whole thing could in theory be
computerised
(automated) so that I could simply call up a location (in my case,
almost
exactly 52N 000E/W) and see some predications?

Jack (and all)
Not sure if this is the source Philip is using, but working through the
Meteograms section of this site, you can tabulate PPN accumulations from
the GFS output ...

http://www.arl.noaa.gov/ready/cmet.html

Martin.

"martin rowley"

Jack (and all)
Not sure if this is the source Philip is using, but working through the
Meteograms section of this site, you can tabulate PPN accumulations from
the GFS output ...

http://www.arl.noaa.gov/ready/cmet.html

Yes. It does work (but of course no real rain here for next week!) I tried
it for a rainy area (the rains in Spain....)
and it is certainly possible, but far too tedious for everyday use. I
suppose as a one-off you might want to see how wet it will be for your
birthday......(or wedding, funeral, etc)

Jack

Now could anyone
predict with accuracy that I would not get wet? Yes, knowing exact location
of the shower, its expected lifespan, etc, would have worked on the day, but
at T-24? That must remain a dream.

Not a dream, Jack, but a nightmare, though I see your point of view. The
day the weather becomes that predictable at the meso- or micro-scale will be
the day to stop taking any interest in it and frankly I hope it never happens.
For instance, today I recorded my highest November max (17.8) in the 21 yrs I
have been recording. From the forecast I did not expect it but it all adds to
the interest. Maybe a professional would think differently.

Tudor Hughes, Warlingham, Surrey.

"TudorHgh" wrote in message
...
Now could anyone
predict with accuracy that I would not get wet? Yes, knowing exact
location
of the shower, its expected lifespan, etc, would have worked on the day,
but
at T-24? That must remain a dream.

Not a dream, Jack, but a nightmare, though I see your point of view.
The
day the weather becomes that predictable at the meso- or micro-scale will
be
the day to stop taking any interest in it and frankly I hope it never
happens.
For instance, today I recorded my highest November max (17.8) in the 21
yrs I
have been recording. From the forecast I did not expect it but it all
adds to
the interest. Maybe a professional would think differently.

Some might, but not the ones that come here, I reckon. What sort of
professional would it be, anyway, whose sole job was to disseminate
these 100% accurate computer products? Not a job for anyone the
least bit interested in the weather.

Getting back to Jack's point, I quite agree. My point in posting was
not to suggest that the model was approaching infallibility, merely to point
out that, on those recent occasions, it did rather well. I do recall that
rainfall event in late-August -- the only significant rainfall of the
month hereabouts -- which brought 0.9mm intermittent rain straddled
over a six hour period to Luton, and 2 or 3 days in advance the same
model was predicting about 75mm over a 48 hour period.

Philip Eden