Hi Frank!
It's good to hear from you, even if I don't agree with all that you
say :-)
The lovely satellite views of the weather are of course useful, but
only for the next few hours. When we have a weather system over the
UK, we use the MOMIDS terminal (military dissemination system, I
think) to watch its progress and how it will affect us. But I am
talking about planning for the next day, and perhaps a day or 2 after
that.
I do think that our Met Office discourages individual met officers
from making their own interpretations of a synoptic situation. Our
planning forecast for the next day looks remarkably similar to the one
produced by the Met Office on their website, the F215 (forecast
weather below 10,000ft). This leads me to believe that there is
centralised forecasting based largely on computer modelling.
I understand how convenient it is to have an automatic system
processing the vast amount of data that is now available, Frank, but
we need humans to interpret for the nuances of each region, people who
have experience of the oddities of each region. I have been watching
forecasting now for over 30 years and, although we pilots have always
been rude (in a friendly fashion) about forecasters and their
products, the results now are significantly worse than about 150 to 20
years ago. This coincides with the comments by my chief forecaster
(when I had a team of 6 or 7 working for me) who told me that the Met
Office wanted to reduce costs by reducing the number of people working
there ...
Chris
On 21 Sep 2008, at 17:28, Frank@xxxxxxxxxxxxx wrote:
> Yes, we had this discussions (not rant) before. Being a weather man
> myself, I feel I have to put a few words in here - and stop lurking
> for a moment ;-)
>
> I agree which you Chris that you can make very good weather
> predictions yourself now, short term ones that is. But do you (like
> me) not use those automate products you despise of yourself then?
> That is what good weather services nowadays do too: the man-machine
> mix is very very powerful, much more than either the man or the
> machine alone.
>
> I have seen it all happen. I used to be an old-fashioned
> meteorologist, being educated and trained in the late sixties and the
> early seventies. Models just started being developed, and were very
> poor at that. And observations (made by humans) from all over the
> globe were very important as the only direct input - both for the
> models and for the old meteorologists drawing their weather maps. I
> remember myself cursing when a particular ship observation somewhere
> in the north Atlantic was missing on one of our 3-hourly maps, and we
> had to wait for the next one... remember we didn't have any real time
> satellite imagery then. And we sometimes asked pilots, who had just
> crossed that ocean, what weather had been during their flights - to
> have some extra information on top of the regular one.
>
> But the satellite imagery getting better and almost real time has made
> the big difference. That and the use of highly sophisticated weather
> radar (Doppler type). Together those have made a lot of the human
> obervations redundant. And both are so easily accessible nowadays,
> that you can indeed make your own weather prediction for a few hours
> ahead, just as well as a professional forecaster. Of course you will
> need some skills to interpret all these images, but a pilot like you
> certainly will have had the training to build those skills.
>
> Could you do without a web site (one of the many) like this one:
> http://www.meteox.com/h.aspx?r=&jaar=-3&soort=satradar
> and this one
> http://www.sat24.com/frame.php?html=view&country=gb
> ?
> If you need more examples I will happily oblige!
>
>
> For one to five days ahead it is a bit different. You (or I) would not
> be able to make any better guess than "tomorrow will be like today"
> with slight modifications for the days after that. But this is where
> the numerical models come in to do the work for us. They have been
> developed to do the main process of thinking what possible evaluations
> can be made of the observations (satellite, radar, human, and
> automated ground stations) of the actual weather. And they give us
> basic predictions which are so good, that we weather persons usually
> only have to make a few futile choices during and in particular at the
> end of the numerical process. These choices are important, of course,
> and their value should not be neglected. But none, neither
> professional nor amateur meteorologist, nowadays could do a proper job
> without the results of the numerical models.
>
> Here in Holland the success rate for a one-day forecast used to be 78%
> back in the sixties. And remember that flipping a coin for head or
> tail already gives you 50%, and plainly deciding that the "weather
> tomorrow will be the same as today" will yield a 65-70% score! All the
> effort we made was for that extra 10-15%. But now, thanks to the
> machinery doing the bulk of the work for us, the succes rate has risen
> to 85%.
> I don't have the actual figures for 5-day forecasts, but they are in
> the same order.
>
> And even our 10-day forecasts (have a look here:
> http://www.knmi.nl/waarschuwingen_en_verwachtingen/ensemble.html
> and pardon the Dutch) are pretty good.
>
> When you know what important long term decisions are taken nowadays
> with those outlooks as a basis you would be astonished, I think.
> Building contractors, breweries, the national energy grid - just to
> name a few: they all think it a big nuisance if the daily update of
> those forecasts is one or two hours late. Their decisions depend on
> it, and the deciders have learned to see its value.
>
> I agree strongly with you on one issue: the local conditions being
> very important. Topography indeed. But unless things are very
> different in the UK from here in the Netherlands, no
> micro-meteorological forecasts (as we call them) are ever made on
> basis of a machine prediction only.
> I agree that in this case the interpreting human work will always be
> necessary - simply because numerical weather models with a resolution
> grid fine enough for that purpose do not exist. Or perhaps the models
> do, but not the machines to give a timely result...
==============================================
List usage info: http://www.zuikoholic.com
List nannies: olympusadmin@xxxxxxxxxx
==============================================
|