xcpsa/xdbub extra
Since the publication of the control climate for these versions of FAMOUS in GMD a few extra bits of analysis have been done by me and others. As I do/find out about useful things in the FAMOUS climate, I’ll put them here.
Global average CO2 sensitivity has already been found to be similar to that of HadCM3, but I’ve done a brief look at the spatial patterns of climate change under C02 forcing in FAMOUS, compared to those in HadCM3.
ENSO is briefly mentioned in the GMD paper, but Thomas Toniazzo at NCAS-Climate, Reading has done a proper job of analysing the variability in the tropical Pacific in FAMOUS. You’ll notice I didn’t say ENSO that time. It’s complex.
Zhaomin Wang did a short analysis of the climatology of the Southern Ocean in FAMOUS last summer. His presentation from the FAMOUS meeting 2008 can be found here
The GMD paper quotes a TOA imbalance of 0.08W/m^2 into the atmosphere for xdbua, after about 4000 years of spinup. This was run on the QUEST cluster at Bristol. Salvatore Pascale at Reading found that his modified version of FAMOUS on HECToR was drifting more than this, so we looked into things a little. Even without Salvatore’s mods, Jonathan found his “standard” HECToR job has a different imbalance to the QUEST spinup. There a few possible reasons, including different compilers (QUEST jobs use pgi, HECToR can use pgi or pathscale) and the inclusion or not of HadOCC (which can affect heat uptake at the ocean surface).
It’s wasn’t simple to find strictly comparable runs in the morass of test jobs we have, but I pulled some numbers from different runs, calculated in a (hopefully) more model-compatible way. All of these numbers are pretty small
| platform | notes | TOA imbalance (W/m^2) |
| QUEST | as xdbua (pgi) | −0.08 |
| HECToR | (pathscale, no HadoCC) | +0.02 |
| HECToR | (pathscale, with Had OCC, older run) | +0.03 |
| HECToR | (pgi, with Had OCC, older run) | +0.02 |
There’s not very much data to do a direct, clean comparison of the impact of Had OCC on the TOA trend. What I’ve got suggests that turning it off might well have a small +ve effect.
Overall, not very scientific, but it looks like platform/compiler differences are probably not causing differences in model drift. The exclusion of HadOCC might have a small warming effect on the climate.
Observational records show a correlation between the long Central England Temperature record and SSTs, particularly in the North Atlantic. This could help with prediction, since it’s thought that SSTs are potentially more predictable than some other climate variables. Ed Hawkins at NCAS-Climate, Reading has been looking at this correlation in different climate models - the correlation plots for FAMOUS are pretty good really! Top left is obs., top right is HadCM3, bottom left is an adtan/xbyvs, bottom right is xcpsa/xdbub (links to large EPS file).
