I have redone this placemark collection with the following enhancements and updated the above technical discussion to reflect the current situation.

The NPI released the 2006-2007 data on 31 March 2008. This is now incorporated into the placemark collection and the 2005-2006 data has been archived.

After some further reading and consideration I decided that using a normal distribution to compare sites was problematic as there are some very small emissions and very large ones. In order to linearise the data I took the natural logarithm off each emission, calculated the Z-scores then summed exp(Z) for each site and medium (air, water, land). I believe this provides a much more accurate means of comparing sites.

My other big post on the European Pollution Emission Register hasn't had any trouble with formatting in the community collection so I have included the full data in the top post. I'll keep the sister thread going for a while yet to avoid dead links.

Thanks to everyone who has offered suggestions, criticism and encouragement.
Andrew.