Hey Matthew,
> -----Original Message----- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of Matthew > Somerville > Sent: 19 October 2006 12:46 > To: [email protected] > Subject: Re: [backstage] Official BBC Weather feeds ahoy! > My first comment is that, whereas the mass of feeds is fine > if you just want to pick one and subscribe to it or put it on > your web site, there seems to be a lot of duplication overall > (I thought I remembered reading a post about the mapping of > postcodes to forecasts, but I can't find it now :-/ ). For It was on a different mailing list, that's why. I'll repeat it now, with some bonus info for the curious. The BBC Weather Centre receives global weather model output from the Met Office twice a day. The output from the midnight run arrives at around 0800GMT and the noon run output arrives around 2000GMT. The TV graphics team use a bunch of data files with sufficient resolution to produce the 3D graphics that we all know and love|hate (delete as appropriate). On the new media side we use a data file that contains 5day forecast information for 4300 weather observations stations throughout the world. Of those 4300 obs stations, approximately 300 are in the UK. As Matthew has very astutely observed, we appear to provide forecasts for over 3000 places in the UK, but in fact each one of these UK locations uses data from the most appropriate obs station of the 300. 'Appropriate' is often, but not always, the closest obs station geographically. There are factors (meteorological, topological) that sometimes result in the most appropriate obs station being one that is not the closest one to a given location. We rely on the Met Office to make these judgements, but as you can probably imagine, with over 3000 locations to assign, the process is prone to error. I am in the middle of auditing the UK locations -> obs station mapping, which involves making lots of maps similar to the one you've done, and compiling a list of mappings that look suspicious. I will then be sending that off to the Met Office so that they can fix or explain. > I've noddily plotted these on a Google Map: > http://www.dracos.co.uk/play/bbc-weather-map/ > from which I work out that the world feed's location (the one > in yellow) is presumably the location of the weather station, > not Aberdeen. :-) Indeed. > Searching for a Aberdeen postcode gives one of the above > feeds (though see below), searching for Aberdeen gives > id/1000, whereas going through the UK map - > http://www.bbc.co.uk/weather/ukweather/sscotland/ - gives world/0014. > Is there a reason that the postcode searches don't just map > to a single feed for Aberdeen, e.g. the world one? I guess > the current answer for a hypothetical backstage developer is > just to use the world feed, but then it's hard to work > out/get e.g. just the Scottish feeds. I should be able to provide a list of world feeds by nation for the UK. > Secondly, still on Aberdeen, there are two locations with > (currently) different forecasts with exactly the same co-ordinates: > AB14 (Aberdeen/Peterculter): > http://feeds.bbc.co.uk/weather/feeds/rss/5day/id/1005.xml > http://feeds.bbc.co.uk/weather/feeds/rss/5day/id/1006.xml > and AB13 (Aberdeen/Milltimber): > http://feeds.bbc.co.uk/weather/feeds/rss/5day/id/1003.xml > http://feeds.bbc.co.uk/weather/feeds/rss/5day/id/1004.xml > I presume that local changes are made to the non-Aberdeen > locations because of some other information, but this does > make it look a bit odd. :) I'll put this on my list of queried mappings. Thanks, Kass Kathryn Schmitt Senior Developer BBC Weather Centre 2026 Television Centre T: 020 82259448 M: 0771 7582482 www.bbc.co.uk/weather www.bbc.co.uk/climate - Sent via the backstage.bbc.co.uk discussion group. To unsubscribe, please visit http://backstage.bbc.co.uk/archives/2005/01/mailing_list.html. Unofficial list archive: http://www.mail-archive.com/[email protected]/

