Tropical cyclones and seasonal means in high-resolution CAM 5 runs. Julio Bacmeister ( NESL ) NCAR Earth System Laboratory Collaborators/contributors: Peter Lauritzen, Jerry Olson, John Truesdale +CAM-5 development team Workshop on High-Resolution Global Modeling June 15, 2010
CAM-5 Zhang-Macfarlane w/ Neale-Richter modifications UW shallow convection+pbl RRTM 2-moment Morrison-Gettelman microphysics ( prognostic aerosols (Ghan, Liu dycore...fv lat-lon
High resolution runs 0.23x0.31 lat-lon Five day forecasts: Initialized using MERRA reanalysis u,v,t,q on August 25 2005 12Z ( CAM-5(a -control: out-of-the-box ( remains -No ZM/NR deep con. (UW -RAS replaces ZM/NR 1997 and 2005 storm seasons: - out-of-the-box CAM-5(a) 1997 and 2005 -CAM-5(b) 2005 with and without ZM/NR -Comparison runs at 2.0x2.5 ZM/NR = Zhang-McFarlane with Neale-Richter modifications
ZM/NR = Zhang-McFarlane with Neale-Richter modifications
High resolution runs 0.23x0.31 lat-lon Five day forecasts: Initialized using MERRA reanalysis u,v,t,q on August 25 2005 12Z ( CAM-5(a -control: out-of-the-box ( remains -No ZM/NR deep con. (UW -RAS replaces ZM/NR ZM/NR = Zhang-McFarlane with Neale-Richter modifications
control No deep convection TRMM 3B42 MERRA
control No deep convection TRMM 3B42 MERRA
control No deep convection TRMM 3B42 MERRA
Minimum Surface Pressure CAM Track 5 control No Deep Conv 25 26 27 28 29 30 31 Day of Aug 2005 NHC
Precipitation types in control Track 5 forecast Total Deep+Shallow convection >120 mm/d Deep convection only Large-Scale only
Katrina forecasts: Surprisingly good intensity forecast both with and without ZM/NR deep convection. Storm cores dominated by grid-scale precipitation even with deep scheme on. -How does ZM/NR know to step aside?
1997 and 2005 seasons: Initialized using false-date reanalysis u,v,t,q over June 1 SSTs. Run until Oct. 15 1997: Strong El Nino, weak activity in Atlantic strong in W. Pacific ( CAM-5(a 2005: Extremely active Atlantic, somewhat below normal in W. Pacific ( CAM-5(a CAM-5(b), CAM-5(b)/no deep conv. param.
2005 2005 1997
2005 1997
East- west sec6on through storm close to peak on 8/8, P=922 hpa, winds~60 m/s Colors: V mag Lines: T Colors: U Inflow and ou*low Precipita6on mm/d Convec6ve ( Sh+Deep ) Large scale
Storm Days: P is -10 hpa below monthly mean.and. Maximum 900 hpa ( Cat3 ) wind within ~125 km is >30 m/s ( Cat1 ) or >50 m/s 1997 2005 Cat 1 Cat 1 Cat 3 Cat 3
1997 and 2005 seasons: Encouraging; W Pacific storms more intense, two over 70 m/s; variability in number of storms has correct sign in both basins, BUT W. Pacific much too sensitive IBTrACS 1997:2005 ratio * 19:15 CAM-5(a) 24:8!!!! Atlantic maybe not sensitive enough IBTrACS 1997:2005 ratio 3:15 CAM-5(a) 5:12 * Vmax>33 m/s Caveat: One run per season. Zhao et al (2009) find spread in numbers, e.g., N. Atl. 2005 (n~10-17) over 4 different ensemble members. CAM-5(b) has more W Pacific TC in 2005
Seasonal mean (JAS) 850-200 hpa wind shear difference between 2005 and 1997 Positive=> stronger shear in 2005 CAM MERRA m./s Wind shear does not seem to be the key
Other Seasonal Means Including comparisons with 2.0x2.5
Seasonal Mean precipitation JJA 1997 0.23x0.31 2.0x2.5 GPCP GPCP
Seasonal Mean precipitation JJA 1997 0.23x0.31 2.0x2.5 GPCP GPCP A real seasonal mean difference/improvement due to increased resolution!!!
non-storm precipitation JJA 1997 Here storm means daily mean PS 3 hpa lower than monthly mean and inst. winds at (day0,day+1) 12Z > ( km 20 m/s (within 125 0-40N means Tot. storm prec=0.81 LS =0.43 Storm precipitation JJA 1997
Total precipitation grid-scale precipitation
Two degree simulation Total precipitation grid-scale precipitation
CAM-5(a) and CAM-5(b) JAS 2005 precipitation raw 0.23x0.31 0.23x0.31 coarsened to 2x2.5 2x2.5 ( CAM-5(a ( CAM-5(b GPCP
CAM-5(a) and CAM-5(b) JAS 2005 precipitation raw 0.23x0.31 0.23x0.31 coarsened to 2x2.5 2x2.5 ( CAM-5(a ( CAM-5(b GPCP N Pacific warm pool ( CAM-5(b wetter in Physics differences? Ensemble variability?
CAM-5(b) and CAM-5(b) w/o deep convection param JAS 2005 raw 0.23x0.31 0.23x0.31 coarsened to 2x2.5 2x2.5 ( CAM-5(b CAM-5(b) no deep convection param GPCP
CAM-5(b) and CAM-5(b) w/o deep convection param JAS 2005 raw 0.23x0.31 0.23x0.31 coarsened to 2x2.5 2x2.5 ( CAM-5(b CAM-5(b) no deep convection param GPCP Spurious S Pacific ITCZ gets worse at high resolution and worse still without deep conv. param.
Propagating MCC s?
Hourly precip CAM-5(b) without deep convection parameterization Days since Aug 1 2005 NCAR Longitude NSF HQ
Hourly precip CAM-5(b) without deep convection parameterization Days since Aug 1 2005 NCAR Longitude NSF HQ
Hourly precip CAM-5(b) without deep convection parameterization Days since Aug 1 2005 Hurricanes ( recurving ) NCAR Longitude NSF HQ
Hourly precip CAM-5(b) default Days since Aug 1 2005 NCAR Longitude NSF HQ
Some statistics
TRMM 3B42 intenstity PDF, 5-days, 30S-30N Frequency Precipitation (mm d ( 1 -
TRMM 3B42 intenstity PDF, 5-days, 30S-30N vs Log-normal curves, power laws -1/2 to -1 Frequency -5/2 to -3 Closer to power law with scale break? Precipitation (mm d ( 1 -
TRMM 3B42 intenstity PDF, 5-days, 30S-30N vs Control ( ZM ) Frequency Precipitation (mm d ( 1 -
TRMM 3B42 intenstity PDF, 5-days, 30S-30N vs Control (ZM) RAS Frequency Precipitation (mm d ( 1 -
TRMM 3B42 intenstity PDF, 5-days, 30S-30N vs Control (ZM) RAS No-DeepCon Frequency Precipitation (mm d ( 1 - Note: No-deepCon still has UW shallow convection
TRMM 3B42 intenstity PDF, 5-days, 30S-30N vs Control (ZM) RAS No-DeepCon MERRA (0.66x0.5) Frequency Precipitation (mm d ( 1 - Note: No-deepCon still has UW shallow convection
TRMM 3B42 intenstity PDF, 5-days, 30S-30N vs Control (ZM) RAS No-DeepCon MERRA ( 0.66x0.5 ) - - - Control (ZM) deep convective precip only Frequency Precipitation (mm d ( 1 - Note: No-deepCon still has UW shallow convection
TRMM 3B42 intenstity PDF, 5-days, 30S-30N vs Control (ZM) RAS No-DeepCon MERRA ( 0.66x0.5 ) - - - Control (ZM) deep convective precip only Frequency Precipitation (mm d ( 1 - Note: No-deepCon still has UW shallow convection
No deep convection Precipitation on 2005-08-28 18Z RAS convection TRMM 3B42 Control CAM using ZM
Time evolution of precipitation intensity PDF hour hour 10 30 100 300 1000 5-day forecast run 5-day segment of seasonal run (days ~85-90, i.e., ( spun-up well 10 30 100 300 1000 Precipitation mm/d
Time evolution of precipitation intensity PDF hour hour Rapid spinup of precip stats 10 30 100 300 1000 5-day forecast run 5-day segment of seasonal run (days ~85-90, i.e., ( spun-up well 10 30 100 300 1000 Precipitation mm/d
Log-frequency ( 6-hrly ) Instantaneous low-level horizontal divergence Aug 1-7 12S to 12N ( CAM-5(b without param deep conv. log 10 f -1e-3 0.0 ( 1/s ) lowest level Divergence +1e-3
Joint PDFs of precipitation rate and wind speed at 900 hpa (instantaneous,once-daily for June 1 to Oct 15 1997) 2x2.5 0.25x0.31 coarse-grained to 2.0x2.5 V (m s -1 ) V (m s -1 ) Prec. (mm d -1 ) Prec. (mm d -1 )
Summary Some encouraging aspects in tropical cyclone simulations. ZMNR deep scheme allows TCs to deepen. (interferes with midwest (? MCCs High resolution improves JJA 1997 seasonal precipitation possibly due to resolved mesoscale dynamics. Precip intensity statistics in forecast and climate runs are similar ( hrs (and spin up within 24 Similar problems in fully parameterized runs. -weak extremes, excessive moderate rain Low level divergence has large extremes in run without deep conv. Param. Climate i.e. mean T q profiles (not shown) similar in 0.25 runs with and without deep conv. Param.. : Convection-free AMIPs?
THANK YOU The NESL Mission is: To advance understanding of weather, climate, atmospheric composition and processes; To provide facility support to the wider community; and, To apply the results to benefit society. NCAR is sponsored by the National Science Foundation