taylor13
commented
7 years ago Honestly, I think we should ask HighResMIP to reduce the number of years they request, but I don't want to spend the time trying to convince them. [They probably think they can learn something about extremes with a long record of high frequency data. If so, they should focus on just a few variables.]
If there were no drift in the piControl, then we could ask for an arbitrary 30-year period sample. Most long-term drift is only apparent at depth in the ocean; the surface doesn't drift much in a well spun-up model. For CMIP5 we said (on the 3hr table) that data should be archived for "30 years (ideally the years corresponding to years 121-150 of abrupt 4XCO2 run)". So we didn't expect strict conformance. I think this period was selected at the request of CFMIP. You could contact Mark Webb (UKMO) and see if this is a rigid requirement.
I don't think I care what 30 years are sampled, as long as the sample is taken after the spin-up period. I certainly don't think someone should provide 230 years of piControl 3hr data.
hope this helps.
martinjuckes
Hello Karl (@taylor13 ),
for the PiControl run you have asked for 3-hourly data for the years corresponding to the last 30 years of the abrupt4xCO2 run, and HighResMIP have asked for some of these variables for years corresponding to 1850-2049 in the historical + scenario runs. These two time slices could be overlapping or they could be disjoint (as far as I'm aware): how do you think we should deal with this?
Martin