1. Many models show a substantial drop off in droplet number away from the coast as is observed. Models’ vertical distribution of droplet number is surprisingly diverse. A Comparison of Regional and Global Models during VOCALS Matthew C. Wyant, Robert Wood, Christopher S. Bretherton, Department of Atmospheric Sciences, Seattle, Washington, USA and VOCA Modeling Groups Acknowledgements This work is only possible through the collaborative efforts of all VOCA modeling participants. The emissions data set for VOCA was provided by Scott Spak of the University of Iowa. This work was supported by NSF Grant ATM-0745702. Introduction Email: mwyant@atmos.washington.edu References Allen, G. and co-authors, 2011: South East Pacific atmospheric composition and variability sampled along 20S during VOCALS-Rex, Atmos. Chem. Phys., 11, 5237-5262. Bretherton, C. S., R. Wood, R. C. George, D. Leon, G. Allen, and X. Zheng, 2010: Atmos. Chem. Phys., 10, 10639-10654. 1x10 -10 kg/kg Mean Low Cloud Fraction Mean Liquid Water Path Cloud Liquid Water at 20S MODIS Total Cloud Fraction AMSR LWP [kg/m 2 ] Solid colors are MODIS, symbols are aircraft measurements. See Bretherton et al. (2010) Observed Droplet Number 20S CCN (950 hPa) Observed Cloud Top and Base Bretherton et al. (2010) Participating Institution Model NameModel TypeHorizontal Resolution [km] inner/outer Vertical levels below 700hPa Aerosol Advection Aerosol- cloud Interaction VOCA Participants COLARSMRegional259J. Manganello IPRC/SOESTIRAM 1.2Regional3012Yes A. Lauer NRLCOAMPS 5Regional15/457X. Hong PNNLWRF-Chem 3.2.1 Regional951Yes Q. Yang ScrippsSCOARRegional20/100D. Putrasahan UCLAWRF 3.1.1Regional15/4531A.Hall U. IowaWRF-Chem UIOWA 3.3 Regional1247Yes P. Saide U. WashingtonCOSMORegional25/10021A. Muhlbauer U. WashingtonWRF-Chem 3.2.1 Regional2515Yes R. George U. Wisconsin Milwaukee WRF-CLUBB (WRF 3.1.1) Regional2526L. Grant ECMWF36R1Global Forecast2518YesJ.-J.Morcrette UKMOMetUM 7.5Global Forecast5020Yes J. Mulcahy GFDLAM3Global Climate20012Yes Y. Lin LMDLMDZ 5v2Global Climate5911F. Codron NCARCAM4 / CAM5Global Climate2005 / 9Yes C. Hannay UCLAAGCM v7.1Global Climate2003H. Xiao PreVOCA compared 15 regional, weather forecast, and climate models (in forecast mode) for October 2006 in the VOCALS region. Many models had large errors in distribution of low cloud cover, though ECMWF and UKMO performed well. Most models produced a marine BL too shallow near the coast at 20S. However, most models qualitatively captured diurnal and day-to-day variability of the cloud and BL despite mean biases. In general, global models outperformed regional models. The VOCA experiment is designed to compare models in the VOCALS region for the REx period of intensive airplane and ship observations. VOCA has more of a focus on aerosols, aerosol and chemical transport, and aerosol-cloud interactions than PreVOCA. Some of the main research questions for VOCA are: Do models simulate the variation of droplet concentration N d along 20S? Is anthropogenic sulfate the main contributor to geographic N d variation? What controls N d in remote ocean regions? What is the simulated indirect effect due to anthropogenic aerosols perturbing clouds and net TOA radiative flux in the VOCALS domain? Simulations cover the REx field project period: 15 Oct. -15 Nov. 2008. Most models are run in short (1-3 day) forecast mode. Horizontally gridded data was submitted from a coarse larger domain or a fine inner domain or both domains. Capable models include aerosol species: SO 4, sea salt, dust, black carbon, organic carbon Emissions of aerosol and gas species are specified in a standard protocol for regional models. Many participating models use their own surface emissions formulas. VOCA Experiment Setup PreVOCA Fine Domain Coarse Domain VOCA Goals VOCA Participating Models Low cloud fraction varies strongly between models. Some models under- predict coastal low cloud cover. Many models over-predict clouds in the tropics away from the South American coast. Liquid water path also strongly varies between models. Few models show good agreement with observations. At 20S, boundary layer depth away from the coast also differs strongly between models. Conclusions Mean model biases in CCN dominate the comparison. Observed excursions from coast of higher droplet number concentrations are matched by higher than average modeled CCN concentrations for many models. Observed Droplet Number Observed DMS Large model biases in cloud fraction, liquid water path, and boundary layer depth are present in most models, similar to the PreVOCA study. Models show large biases in aerosol and chemical concentrations. Models can reproduce timing of excursions of continental air into the offshore boundary layer. We will be analyzing a geo-engeering component of the VOCA experiment with cloud droplet number artificially adjusted to 375 cm -3. Mean 20S profiles Sulfate Aerosol Mass at 950 hPa Pressure [hPa] g/kg Longitude Pressure [hPa] Longitude Pressure [hPa] 750 800 900 1000 Most models over-predict sulfate mass. Most models under-predict CCN concentrations (0.1% supersaturation) both in the boundary layer and above the boundary layer. Many models’ DMS concentrations are much higher than observed (note that observations don’t cover the full diurnal cycle). Bretherton et al. (2010)