"Norman Lynagh" wrote in message
...
Will Hand wrote:
On 10 Feb 2017 19:03:11 GMT
"Norman Lynagh" wrote:
There's a yellow snow warning in force for here from 1800 this evening
till
1000 on Saturday. The current automated forecast for Tideswell gives
the
following:
Fri 10th
--------
1800: Cloudy. Probability of precipitation 10%
1900: Cloudy. PoP 10%
2000: Cloudy. PoP 10%
2100: Partly cloudy. PoP 5%
2200: Slight snow. PoP 50%
2300: Cloudy. PoP 10%
Sat 11th
--------
0000: Cloudy. PoP 10%
0100: Cloudy. PoP 10%
0200: Slight snow. PoP 50%
0300: Slight snow. PoP 60%
0400: Slight snow. PoP 70%
0500: Heavy snow. PoP 90%
0600: Heavy snow. PoP 80%
0700: Slight snow. PoP 60%
0800: Slight snow. PoP 60%
0900: Slight snow. PoP 60%
1000: Slight snow. PoP 60%
That takes us to the end of the period of validity of the yellow
warning.
Yet, after that the Tideswell forecast has slight snow continuing till
1300
then it predicts heavy snow from 1400 on Saturday right through to 1500
on
Sunday, with no current warning in force for this.
So, we have a warning for a 16-hour period during which not much snow
is
forecast to fall then a forecast of 24 hours of heavy snow for which no
warning has been issued. It doesn't hang together very well :-(
Produced by two different methods. I suspect the human Chief Forecaster
is not
aware of the automated forecast. I agree it's an utter shambles and a
disgrace.
Will
This then raises the question: If the Chief Forecaster does not slavishly
follow the machine output then why is the machine output considered to be
satisfactory end-user material? That brings me back to the opinion that I
have
aired on here recently i.e. that numerical model output is a tool to be
used by
experienced forecasters to assist them in their decision-making. It is not
reliable enough to be issued as an end-user product. It is not fit for
that
purpose. The sort of conflicts that I have highlighted above lay the Met
Office
wide open to possible legal action from someone who suffers a loss
resulting
from relying on a product which tells a different story from the 'party
line'
story. Getting the forecast wrong is one thing but having 2 conflicting
forecasts valid concurrently is quite another.
Oh dear where do we start. First of all the main driver for post-code
specific auto forecasts were the commercial people back in the early
noughties. This was backed up by a committee called the public weather
service customer group. NWP scientists were sceptical but were nevertheless
tasked with providing products. Around circa 2008 it was realised that
conflicts as you have highlighted could occur, so two strategic strands were
set in motion. One was full automation by 2020 (apart from warnings) and the
other was something called "Strategic Intervention" (SI). The idea of SI was
that the Chief forecasters would modify all NWP products once before they
left the building for customers. That way there would be one story and only
one story. But very soon massive technical problems became apparent. What do
we do with ensembles, how do you do the intervention? Do you modify surface
fields and use Omega equation ideas to do the rest automatically and how on
earth do you modify cloud/precip. etc etc. As well as these issues there
were also IT infrastructure problems, how do you interface the NWP with the
forecaster in a way that he could alter it quickly and efficiently? Then
there was the big issue of who was to pay for it? As far as I am aware SI
was scrapped. The big white hope is now high res. NWP and full automation
using a high resolution ensemble approach.
Will
--
" Some sects believe that the world was created 5000 years ago. Another sect
believes that it was created in 1910 "
http://www.lyneside.demon.co.uk/Hayt...antage_Pro.htm
Will Hand (Haytor, Devon, 1017 feet asl)
---------------------------------------------