HiGEM High Resolution Global Environmental Modelling

Status of HiGEM integrations

Expid Length
Description Version Major Changes Notes Archived at BADC
xboej1 YearInitial test runvn5.5NoneFirst attempt at a N144 1/3x1/3 HadGEM1. Problems with initial land surface temps resulted in large spin-up. Concern about performance of JJA Arctic sea-ice, oceanic diffusion and speed of model integration. No
xboem5 YearsInitial test run with better configured atmos dump
and with revised sea-ice parameters
vn5.5New atmos dump
WME scaling factor 0.75->0.55
P_star 2000->6000
Ice drag 5.50e-03->1.50e-02
Spin-up problems resolved. Still too little JJA Arctic sea-ice.No
xbrxb 5 YearsSensitivity test with reduced Visbeck diffusionvn5.5Visbeck parameter reduced to 2.50e6 globally
Fix to prevent inland sea surface height drift introduced
Ocean eddy variability increased but model crashed after 5 years with the CFL criterion violated in the sea-ice codeNo
xbrxd8 YearsSensitivity test with increased sea-ice strength parameter (P_star)vn5.5P_star set to 40000Increased P_star doesn't resolve the lack of JJA Arctic sea-ice. Model crashes with the CFL criterion violated in the sea-ice codeNo
xbpjt5 YearsInitial vn6.1 test run with HadGAM1a revisionsvn6.1HadGEM1a revisions included. Ice advection substepping included. Fix to oceanic vertical velocities introduced. Visbeck parameter set to zero. Seasonal vegetation included. Optimisations to free surface solver included. New orography around South Pole. Land point at NP removed in atmos model. New atmos polar filter included. Ocean timestep 15mins->20minsModel crashed after 2 years with atmos noise around South Pole. Resolved using new atmos start dump. Model crashed in Nares Straits after 5 years with unrealistically large salinity values. HadGAM1a revisions to sea-ice albedos, removal of NP land point and sea-ice melting bugfixes improve JJA Arctic sea-ice. Optimisations triple speed of the model.Yes
xbyla2 yearsIntegration with unfiltered barotropic velocitiesvn6.1No Fourier filtering of oceanic barotropic velocities to fix spurious convergence and drift of tracers near the pole. MODIS soil albedos introduced. Atmos polar filtering of increments removed. Run-off over frozen soil fix introduced.Stopped to make changes to ancils Yes
xbyld12 years 8 monthsIntegration without snow pillars in ancilsvn6.1No Tibetan snow pillars (isolated ice points over Tibet) that caused problems in some Earth Simulator runs. DMS aerosol ancil changed after four years 3 months to one with constant values around North Pole. Restart after 12 years 3 months after archive problems. Net TOA drifts after 5 years to 1 W/m2 (downward positive). SSTs cool over much of the globe but warm near the Antarctic leading the Antarctic sea-ice loss. Yes
xbylk5 years Integration with reduced upper ocean vertical diffusion and no atmospheric diffusionvn6.1Including the atmospheric diffusion resulted in a substantial degradation in the fidelity of tropical windstress, so it has been removed. The upper ocean vertical diffusion was reduced to values consistent with observations. Changes in the vertical diffusion had little impact on the TOA drift but removing the atmospheric diffusion improved the representation of tropical windstress (and so SST).Yes
xbylp2 years Integration with altered mixed layer parameter (delta=50m lamda=0.55)vn6.1 The TOA drift and melting of Antarctic sea-ice were thought be associated with incorrect mixed layer depths. These changes in the mixed layer parameters resulted in too shallow (and thus too warm) summertime mixed layer depths around Antarctica.Yes
xbylrCurrently 62 years Integration with mixed layer parameters delta=50m lambda=0.7 vn6.1The mixed layer parameters were changed to more consistent with values constrained by observations.The changes in the mixed layer parameters slowed down the rate of melting Antarctic sea-ice but do not prevent it.Yes
xciel50 years As similar as possible as xbylr but at HadGEM resolutionvn6.1Some changes to facilitate the HadGEM resolution, such as the reinclusion of GM mixing, changes in timestep and changes in filtering. Other parameters are the same as in xbylr.There are a large number of changes, too many to go into here. See Norton et al 2008 for details.Yes


Read the HiGEM mid-term report produced Jan 2006.

Download here