Changed lines 1-2 from:
(back up to CurrentWorkRobin)
to:
(back up to ScienceWorkRobin)
Added lines 5-10:
'''version XFHCY'''
!!!Free Drifting Ice
Up to and including XFXWB FAMOUS, sea ice is simply advected with the surface currents. Amongst other things, this stops the ice from drifting north off the coast of Antarctica like it should, which we thought might be important in correcting the Southern Ocean warm bias and lack of AABW. UM4.5 does include a subroutine for advecting the ice independently of the surface currents that was intended for [=HadCM4=] but never used(?). The code's clearly a bit flakey, as it initially failed on the QUEST cluster until I fixed a memory allocation error, but it appears to run fine on other platforms. This is being tuned up for the MOSES2.2 release of FAMOUS
Changed lines 13-14 from:
Much of the northern latitude cold bias is linked to excessive ice production. Encouraging the melting of the ice thus improves the high latitude surface temperatures considerably. Sea-ice albedo is temperature dependent, between a low “melting” value and a high “frozen” value. For FAMOUS releases up to XDBUA, the GMD paper reference version, the melting value has been lowered from 0.5 to 0.2. This is a bit extreme, although individual meltponds may have albedoes this low. This value was found to help temperature and ice fraction the most, through a short-ish series of sea-ice parameter tests. The overall Northern Hemisphere ice albedo in summer is now a little unrealistically low, but it’s actually better than it was for the rest of the year, and temperatures, year-round, are much improved. There’s also a slight average improvement in the Southern Hemisphere ice coverage. There’s still too much ice cover for much of the year in the Northern Hemisphere, then too little after the summer melt. Interestingly, the excessive ice in the Pacific really isn’t impacted much by this - really, it’s good for the temperatures via the ice albedo change, rather than by the actual removal of ice fraction. In addition, the tunable parameter of the minimum ice thickness of 0.5m is also reduced, to 0.25m.
to:
The earlier low-albedo fix for the cold bias isn't really required once the excessive atmospheric filtering is reduced, so for XFHCY I've gone back to the higher sea-ice albedoes/thicknesses as for HadCM3. These may change during the tuning exercise, however.
---
'''versions XFXWB and earlier'''
!!!Albedo
Much of the northern high-latitude cold bias is linked to excessive ice production. Encouraging the melting of the ice thus improves the high latitude surface temperatures considerably. Sea-ice albedo is temperature dependent, between a low “melting” value and a high “frozen” value. For FAMOUS releases up to XDBUA, the GMD paper reference version, the melting value has been lowered from 0.5 to 0.2. This is a bit extreme, although individual meltponds may have albedoes this low. This value was found to help temperature and ice fraction the most, through a short-ish series of sea-ice parameter tests. The overall Northern Hemisphere ice albedo in summer is now a little unrealistically low, but it’s actually better than it was for the rest of the year, and temperatures, year-round, are much improved. There’s also a slight average improvement in the Southern Hemisphere ice coverage. There’s still too much ice cover for much of the year in the Northern Hemisphere, then too little after the summer melt. Interestingly, the excessive ice in the Pacific really isn’t impacted much by this - really, it’s good for the temperatures via the ice albedo change, rather than by the actual removal of ice fraction. In addition, the tunable parameter of the minimum ice thickness of 0.5m is also reduced, to 0.25m.
Deleted lines 25-29:
Since XDBUA, the use of 2 dynamic sweeps per timestep (see CurrentWorkTimestepping) has increased atmospheric heat transport, and use of the new albedoes noted above leads to an excessive polar warm bias during winter. The original ice albeoes are thus being used for the moment with 2 sweeps, but further tuning will probably occur at some point.
!!!Free Drifting Ice
Up to and including XDBUA FAMOUS, sea ice is simply advected with the surface currents. Amongst other things, this stops the ice from drifting north off the coast of Antarctica like it should, which we thought might be important in correcting the Southern Ocean warm bias and lack of AABW. UM4.5 does include a subroutine for advecting the ice independently of the surface currents that was intended for [=HadCM4=] but never used(?). The code's clearly a bit flakey, as it fails on the QUEST cluster with a puzzling memory allocation error, but it appears to run fine on HECToR. Free drifting ice does improve the SO ice situation, but can lead to excessive export of ice into the North Atlantic. Use in combination with lower isopycnal diffusion in the ocean (see CurrentWorkLowIsopyc) seems to have a general beneficial effect.