CAM-UW update
description
Transcript of CAM-UW update
CAM-UW update
Christopher S. Bretherton and Sungsu Park
Dept. of Atmospheric Sciences, Univ. of Washington, Seattle
Compare CAM-UW (UW moist turb+ShCu, fv2x2.5, L30) CAM-UW-nodeep (as above but no ZM) (5yr)with CAM3-fv (2x2.5L26).These runs are with CAM3.1, but we have now migrated CAM-UW to CAM3_3_45 development branch with little apparent change in climo.
CERES
RESTOA LWCF
27.1
SWCF
49.2
CAM3-fv -0.2 30.6 -55.1
CAM-UW (S026) -0.3 31.2 -57.1
Nodeep (S027) 3.0 34.2 -56.5
CAM-UW
CAM-UW-nodeep
CAM3-fv
ERA40 CAM3 CAM-UW
SE Pacific cross-section
SON climo
No stratofogulus in
CAM3-UW
CAM-UW
CAM-UW-nodeep
CAM3-fv
CAM3-fv CAM3-UW CAM3-UW-nodeep
CAM-UW
CAM-UW-nodeep
CAM3-fv
DJF low cloud
CAM3-fv
CAM-UW-nodeep
CAM-UW
DJF SAT and surface LWCF
RMS error table (all gridpoints, 4 seasons)
Field Data RMSE ratio vs. CAM3.0 3.3.45fvctrl S026 S027
Sea Level Pressure ERA40 1.01 1.33 1.08Ocean Surface Stress ERS 0.96 1.02 0.83Surf. Air Temp (land) L-W 1.11 1.14 1.14Rainfall Xie-Arkin 1.12 1.09 1.04Trop. Land Rainfall Xie-Arkin 1.06 1.02 1.12Net LW (TOA) CERES 1.19 1.20 1.00Net SW (TOA) CERES 1.06 0.99 0.98U (300 hPa) ERA40 0.93 1.15 0.85T (lat-p xsect) ERA40 0.93 0.96 0.98RH (lat-p xsect) ERA40 0.84 0.89 1.02Climate Bias Index 1.02 1.08 1.00
Synthesis so far
• CAM-UW bias patterns are very similar to CAM3 overall, except for accentuated biases of SLP and jets.
• Biases in boundary layer SWCF in trade Cu regimes are highly correlated with precipitation biases.
• Removing ZM deep convection by using only the UW shallow Cu scheme for all convection has larger impact on tropical biases, improves SLP, wind stress and overall skill, though not as much as switching to the two new candidate deep convection schemes.
• Excessive high latitude winter low cloud/SAT in CAM3, even worse in CAM3-UW (but see Sungsu’s talk tomorrow!), not affected by deep convection scheme.
SCAM GEWEX intercomparison results
• At UW and with C. Lappen of CSU, we have initiated SCAM3/SCAM-UW participation in international boundary-layer SCM/LES intercomparison studies.
Here, we discuss interesting insights from: • GABLS-1 idealized stable boundary layer case:
How might PBL contribute to hi-lat cloud/TS biases?• RICO precipitating shallow Cu (latest GCSS case):
Exposes issues with cloud fraction, LWP, precip.
GEWEX Atmospheric Boundary Layer Study (GABLS)
Case 1: Idealized stable boundary layer (Beare et al, Cuxart et al 2006)
- Stratified initial sounding, no moisture
- Surface cooling of 0.25K/hr for 9 hrs
- 8 m/s geostrophic wind, 1 cm surface roughness
- Good consensus among LES simulations with z = 2 m.
- Requires several mods to SCAM to set up.
High res: ~10 m vertical resolution
Low res: L30 as used in CAM-UW, grid levels at 60, 180, … m,
CAM3: K-profile; diagnosed PBL depth, no Ri cutoff.
CAM-UW: K(Ri) from Galperin (1988) in stable PBL, cuts off if Ri>0.19
Goals: Compare HR, LR PBL depth, surface downward heat flux.
9 hour time-heightsHi-res results
CAM3 – deeper PBL
CAMUW – shallower PBL
8-9 hr mean
Surface sensible heat flux: parameterization & resolution
• Downward heat flux larger in CAM3 than in LES, CAM-UW.
• Only slight change for both params at L30 vs. hi-res, even though there is only one grid layer within the CAMUW PBL!
• Both params work respectably for this stable PBL case.
GCSS-BLCWG RICO shallow cu intercomparison• VanZanten, Siebesma et al. –
ongoing. • Deep, weakly capped trade Cu
with some showers• Based on composite conditions for
16 Dec. 2004-5 Jan. 2005• Metrics: - Radar rain rate ~1 mm/d. - T, q profiles should be quasisteady
given forcings. - LES simulations.• SCAM results have ZM turned off
to avoid spurious deep convection.
B. Stevens photo
SCAM RICO results
• CAM3 –develops spurious inversion because Cu goes too deep.
• CAM3 and CAM-UW both have reasonable mean rainfall rates
34-36 hour profiles
LES
Cumulus profiles
• CAM3 shallow Cu fraction too large and top-heavy, with excess LWP and numerical oscillations.
• CAMUW better overall.
• LES shows rainfall highest at 2 km, not Cu base (life cycle).
LES Condensate from ShCu scheme
Summary
• The UW moist turbulence and shallow Cu schemes have a modest effect on CAM climatology, despite improving single-column performance in GCSS cases.
• Interactions with stratiform cloud and deep convection parameterizations have big impacts on PBL biases.
• Both current CAM and CAM-UW do a respectable job on the GABLS dry stable PBL case even at L30 resolution, bracketing the LES ‘truth’.
• Tomorrow Sungsu will show you a disturbing but legitimate way to reduce Arctic low cloud/TS in CAM that dwarfs effects of changing PBL scheme.
A curiosity - Leads and SHF through sea ice
• CAM includes a geographically and seasonally varying climatological open water fraction in sea-ice regions.
• Although small, this often produces upward sensible heat flux even in the highly stable winter PBL over the Arctic Ocean, making diagnosis of PBL type ambiguous, and has a 5-10 W m-2 effect in surface energy budget.