Why can't I aggregate two grid cells together with the probabilistic data?

It is not possible to aggregate two or more grid cells together because the values of change at a particular probability level (for example, 90%) for a number of grid squares cannot be averaged.

In detail

Many users of the UKCIP02 projections will have become used to looking at maps which show a snapshot of the distribution of changes in climate over the UK, for example those showing change in summer precipitation by the 2080s, for a High emissions scenario, at a resolution of 50 km. The UKCP09 maps have the same sort of appearance as those in UKCIP02, apart from the increased resolution of 25 km, but the nature of their content is quite different.

The maps in UKCP09 show changes at a particular probability level (e.g. 10%, 50%, 90%), taken from the cumulative distribution functions, CDF, at each 25 km square. The map below is generated by the User Interface and shows the projected changes in mean winter precipitation at the 90% probability level, over Wales. Values of percentage change are overprinted on each 25 km square; there is a 90% probability of the precipitation change being below this value. Shown alongside are the CDFs for two of the individual squares in that region, showing correspondence between the 90% probability level on the CDF and the value given on the map. So, for example, the upper CDF for an inland square shows that the projected change is very unlikely to be greater than 19% or, alternatively, very likely to be less than 19%. The lower CDF shows the very different change, 80%, at the same probability level, at a coastal grid square. The same principle will, of course, apply to maps showing projected changes at 10% and 50% probability levels.