Radio interference from man made sources is the biggest issue for radio telescopes in the traditional radio band. So it's not really the weather that decides, but the crappy radio skies of the UK.
That's a pity. From what I read, the Padova, Italy bid was better in several ways, so it came down to politics. Apparently, the UK threatened to pull out of the project entirely if their candidate was not chosen.
I'm a bit biased myself, living in Padova, but it's objectively a pretty good place for something like this, just like it was a good place for Galileo in his day.
Current systems for handling data would be unable to cope with the demands of the SKA and it is hoped technology will catch up by the time it is completed.
Several of my friends worked on the bid for south africa and the volume of data they they were talking about is astounding.
Another friend of mine is working on the forked nix os that they are building to run all this, very very interesting,lots of interesting computing. Lots of phd and post doc CS and EE work being done around this in South Africa.
Interestingly the algo which they use to process the data is very simple, several nested loops (trying to find the paper now one of my friends on the project sent me yielded no results), but this algo obivously has to run on masses of data and massively parallel.
The part that needs the most optimism is the provision of the connectivity to the section in South Africa, which is unfortunately a political challenge instead of an engineering challenge.
Relying on Moore's law-like as caling is becoming standard practice in big science. Someday it'll stop working, but there's plentry of evidence to suggest that tooling will continue to improve on timescales of a decade.
It's worth noting that data from existing radio telescopes are freely available. The National Radio Astronomy Observatory in the US operates an archive for their telescopes[0] as does the Atacama Large Millimeter/sub-millimeter Array[1], the Australia Telescope Compact Array[2], the Westerbork Synthesis Radio Telescope[3], and many others. Note however that much/most of this data is not the final data products but is raw and/or semi-calibrated.
How much data? Well, data rates have the potential to be multiple petabytes per second. The main concern is gathering, transmitting, filtering, processing, and storing this data (or enough of it) quickly enough while staying within budget.
That said, Jodrell Bank is a very interesting place. Good article: http://www.theregister.co.uk/2013/03/25/geeks_guide_jodrell_...