Currently the histograms are plotted as normalised density plots (e.g. https://github.com/MetOffice/CSET/blob/main/src/CSET/operators/plot.py#L72). However, these are less flexible, and have a subtler interpretation than PDFs (which most users would be familiar with). It also leads to misleading values on the y axis (i.e. values above one) which whilst mathematically accurate can cause interpretation problems.

The idea is to make the histograms more user friendly, and more flexible, by converting them into PDFs by using the counts for each bin and then normalising by the number of elements used to make the histogram.