What is clipping and why has it been done? What are the implications for use?

The probabilistic projections sampled data in the netCDF files is clipped so that for each location, time period, averaging period, and variable, the sampled data is bound between the 1st and 99th probability levels.

In detail

Sampled data is designed to span the full range of uncertainty. But there are issues in the tails because a number of statistical processing steps are combined and the statistical approximations can have a large influence on the tails of the distribution.

It has been judged that clipping to the 1st and 99th probability levels will exclude data in the tails which are unreliable.

For these reasons the project team has narrowed the suggested robust range to the 10th–90th probability levels. Anything outside of this range can be taken into account but may not be considered robust in decision-making.

It should also be noted that the variance of daily precipitation, variance of daily temperature and skewness of daily temperature (three variables used by the UKCP09 Weather Generator) are clipped at the 5% probability level and the 95% probability level (as a result of addressing concerns related to the robustness of the projections of these variables).