CDO

Hello,
I am trying to create ensemble means with monthly mean CMIP5 RCP8.5 data for 2070-2099. Since not all models have consistent time steps during 2070-2099 (but I still want to use every available model), the "ensmean" function does not work. A work around I discovered is to "mergetime" all models and then "monmean" to calculate the mean for each month. For example, if I have 200 models and did a "mergetime", I would get 200 Januarys for 2070, 200 Februarys for 2070, ... , 200 Decembers for 2099. Then a "monmean" will calculate the average of all the Januarys for 2070, etc, giving me an ensemble mean that is not worried if there happened to be only 197 Decembers in 2099.

This worked perfectly find for historical data and RCP45 data, but for some reason for the RCP85 data the "monmean" is returning multiple yearly values at the. I am attaching the "merged" and "monmean" files below. The monthly mean file has several year 2070s, 2080s, and 2090s, only the years ending in 70, 80, or 90 (the in between years are fine).

Any idea what may be the issue, or a work around to solve this problem? Thank you very much in advance for any help!

Best wishes,
Kyle