CMIP3 / CMIP5 temperature comparisons

Some basic initial comparisons of CMIP3 and CMIP5 global climate model (GCM) downscaled outputs from SNAP’s data repository. Specifically, comparisons of temperature time series from each phase, for various geographic regions and 12-month and four 3-month seasonal periods [precipitation plots can be found here]. These climate variables were downscaled to 2-km and 771-m resolution using PRISM climatologies for Alaska and western Canada. Some comparisons of regional averages of the 2-km resolution outputs are presented here.

Although the scenarios from CMIP3 and the RCPs from CMIP5 are different by design, there is an expected inclination for some to want to compare them one to one. We have the b2, a1b, and a2 scenarios from CMIP3 and RCP 4.5, 6.0, and 8.5 from CMIP5 at SNAP. Although the latter are not meant to substitute directly for the former in respective order, we have some basic plots of this anyway to get a peak at how they differ. The time series plots are coarse, consisting of decadal averages of a five model average from each phase; a very rough first glance at some of our downscaled CMIP5 projections vis-a-vis CMIP3 projected outputs, aggregated over time, space, and models.

There is much more to come, including model-specific comparisons, comparisons at an annual time step, inter- and intra-annual, seasonal and decadal variability, etc. In the meantime, some people are interested in these. Clocking in at a whole nine time steps (gasp!), these are not the most exciting time series plots ever put on display. In my opinion they mostly serve the purpose of yielding more questions, which is an excellent purpose and why I began with something coarse and simple, and demonstrating a need for deeper insight into the data, in addition to demonstrating that at this coarse scale we can see that nothing totally unexpected happened; the newer models I downscaled are not completely at odds with the old ones at an aggregate level. Don’t worry, I’ve already anticipated what people will find lacking here, because it’s so obvious. More fine-scale graphs of various kinds will be available in upcoming posts.

And as usual, everything was done in R, start to finish. The region abbreviations refer to SNAP’s full Alaska/western Canada domain, Alaska only, Alberta, and British Columbia. DJF, MAM, JJA, and SON refer to 3-month seasonal periods. In the plots where CMIP3 scenarios and CMIP5 RCPs are combined pairwise, differences are used for temperature and ratios, or proportional changes, are used for precipitation.




12-month differences


December, January, February


December, January, February differences


March, April, May


March, April, May differences


June, July, August


June, July, August differences


September, October, November


September, October, November differences

This entry was posted by Matt Leonawicz.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: