I know that a fair few current Metmen (I'm not!) read this forum, so I'm sure you'll get some input from them at some point.
http://www.metoffice.gov.uk/corporat...r20040524.html
From my days as a Met. Office observer on mainly mil airfields, I can safely say that on a good day, I would sit on my bum looking out the window watching the planes go by, and popping out once an hour to carry out an ob, and spending the next 15 mins submitting it. Not exactly rocket science.
However on a cr@ppy day (and being the UK, you have a fair few cr@ppy days!) both shift observers would be busy , keeping a constant watch, doing specials, updating computers, etc etc. Now I would guess that crappy met days also make for busy ATC days?
I'm not trying to claim a fully trained ATC observer is less capable than a Met. Office observer, however the Met. Office observers primary role is to provide Met. observations - the same cannot be said for Met. trained ATC staff. So how does that work on a busy day?
I'm not entirely convinced you can train to be an observer by 18 hour CBT either! Nothing can beat real-life training for a task that relies on physical observation of real world events - not computer generated poor quality images of indesinguishable clouds, visibility, cloud heights, etc etc. I dont want to suggest that there are any safety concerns - I'm sure there are enough safeguards in place anyway - however it really is a poor state of affairs for everyone!
Then there is the meteorological data aspect. A Met. Office observer provides more than just a METAR, they provide a very specific running meteorological record which feeds the Met. Office model. Over the last few years, the number of full time observation sites has reduced quite markedly, being replaced by part-time or automatic observations giving only partial information (when they work). Ask any Met. Office observer or forecaster what they think of this change. How can you expect forecast quality from the model output to get better if the information feeding the model is getting sparser and contains less information?