240 likes | 380 Views
CAM3 composite cyclone comparison. Paul Field, A. Gettelman, R. Neale, P. Rasch, R. Wood. Satellite Observations. All data over ocean AMSR – WVP, Rainfall, SST Quikscat – 10m winds NCEP/NCAR – mslp, T-profile 2003, 2004, 1500 midlatitude cyclones from NP, NA, SP, SA
E N D
CAM3 composite cyclone comparison Paul Field, A. Gettelman, R. Neale, P. Rasch, R. Wood
Satellite Observations • All data over ocean • AMSR – WVP, Rainfall, SST • Quikscat – 10m winds • NCEP/NCAR – mslp, T-profile • 2003, 2004, 1500 midlatitude cyclones from NP, NA, SP, SA • Field and Wood JCLI 2007
Compositing cyclone center
Compositing • Locate low center onto 4000x4000 km grid • 100x100 km ‘pixels’ • Subsample composites based on mean windspeed and mean wvp
Models • 10 yr runs, analyze last 3 years • Control: CAM3 0.9x1.25deg Finite Volume • SSAT: Modify Slingo cloud fraction curve • 4x5 : 4 by 5 degree grid spacing 1 CF 0 1.0 RHice
Satellite Control High Cloud All show positive correlation with cyclone strength Control ~+3% bias in high cloud fraction Ssat ~no bias, slightly more dependence on wvp that obs 4x5 ~+10% bias ssat 4x5
Satellite Control Rainfall All show positive correlation with cyclone strength All show greater dependency of rainfall on cyclone strength than obs ssat 4x5
Satellite Control RHcol Obs shows no correlation with cyclone strength or wvp Model runs show positive correlation with cyclone strength Control & ssat ~ -7% bias 4x5 less bias ssat 4x5
Satellite Control High Cloud Composite (max strength, med wvp) All show high cloud to east and se Model runs greater means and extent Ssat is better than control 4x5 is worse than control ssat 4x5
Satellite Control Rainfall composite (max strength, med wvp) All show max rain to east and se Model runs greater means and extent ssat 4x5
Warm Conveyor Belt thermodynamic dynamic R = WVPV W V S q0 = WVP / S
Satellite Control Warm Conveyor Belt All show positive correlation Models show greater dependency of rainfall on cyclone strength than obs ssat 4x5
No Convection • No deep or shallow or midlevel convection • Unbalanced • 5 year run – analyse last 3 years
Satellite Control High Cloud Composite (max strength, med wvp) No conv: much less high cloud No conv
Control (NP) Obs (NA) Cloud top temperature pdf slices No convection: much less high cloud reduction in mid-level cloud better agreement with obs No conv (NP)
Satellite Control Warm Conveyor Belt No convection: Still shows more dependence on cyclone strength Still RHcol correlation with strength (not shown) No conv
Quikscat wind divergence (10m) Increasing moisture Increasing cyclone strength
Horizontal wind divergence (10m) Satellite Control Obs: width of pdf increases with strength Model: no obvious increase in pdf width with strength weak med strong 4x5 ssat
Conclusions (so far) • Rainfall behavior largely unaffected by differences in models • RHcol dependence on strength in model – not obs (related to div/dynamics?) • High cloud is better in ssat run. Introduction of scheme that allows supersat should be improvement • No conv – improves high cloud representation (fix midlat cyc clouds with improved conv)
To do list • Provide physical arguments for observed variations • Compare results from • New microphysics • New conv. schemes