How should spells spanning several periods be handled?
Comparing climpact2 with climix, the following difference emerged. If a spell spans more than one period, climpact2 treats it as a member of the last period. For example in the
cwd index a wet spell starts in December of 1997 and ends in January 1998, climpact2 then identifies this as the longest spell of 1998 with 34 days, whereas climix at the moment only counts the 14 days that are part of the period in question, 1998.
What behavior do we want? On the one hand it seems bad to not count particularly long spells. On the other hand that one is not part of the year and it would mean that one has to have access to at least one period before and after the time span of interest. Also, should that be counted for all periods, only the first, only the last, or the dominant one? What if a spell spans more than one full period, ie at least three periods? For annual indices that seams unlikely, but for monthly ones that is perhaps possible?