Ah right. I'm in the queue at RS. Looking forward to see what it will do.

OK I went to degreedays.net (thanks Ed!), searched for Aberdeen, an option came up to search by map around Aberdeen. Surprisingly there are HDD sets from lots of local weather stations available, not just the met office ones and airports but also council buildings, amateur met stations etc etc

I downloaded the HDD data for every station within about 20 miles of me. The difference between the highest and lowest was 18%. Surprisingly the station nearest the sea was not the lowest HDD. Two stations a mile apart were different by 6%.

From this I agree with Seret that the available HDD data is not very precise and I shouldnt sweat over making corrections for my own site location.

I think local variation of 20% is pretty good on the Celsius scale as it assumes that 0°C is in median value (possibly) and the resolution is 0.1°C (probably) and the precision is about 1% (across the temperature range). The accuracy is probably about 5%. Then there is rounding errors.
The way to get around this is to calculate the Standard Error of the Mean SD/N^0.5, this will allow you to set the upper and lower bounds, and as long as 95% of your data is within these, you have nothing to worry about.
Alternatively you you can go down the ensemble route where you average all the results from the different sources during the same time period. I would use mode rather than mean if there are enough data points as that automatically gets rid of the extreme reading (these are the ones that are most likely to be wrong). Think of this as a consensus method of looking at data. Look at what they have in common, not what they have that is different.
Statistics is your friend here and your data will not lie to you.

Good man