Arctic Mixed-Phase Clouds and Arctic Mixed-Phase Clouds and Their Simulations in Climate Their Simulations in Climate
Models Models
Shaocheng Xie
Atmospheric, Earth and Energy DivisionLawrence Livermore National Laboratory
My Background My Background
Atmospheric scientist at LLNL Have been working on climate model, climate model
evaluation, cloud parameterization development, and field data analysis in the past 20 years
Have some knowledge on cloud microphysics, but not much
OutlinesOutlines
OutlinesOutlines
Model evaluation: how to test microphysical parameterizations used in climate models?• Field measurements (M-PACE)
• Modeling approaches (CAPT – run climate model in short-range weather forecast)
• Model vs. Data
Sensitivity of climate simulations to ice nucleation schemes
Summary
A Little Background: Climate A Little Background: Climate Models Models
Climate Models are systems of differential equations based on the basic laws of physics, fluid motion, and chemistry.•Momentum (u, v)•Continuity (w)•Thermodynamic (T)•Moisture (q)
Model dynamicsp, T, u, v, q…
Model physicsP(M, R, S, T)
Horizontal resolution: 100-200km
Vertical resolution: ~50 hPa
A Little Background: CloudsA Little Background: Clouds
Clouds Impact on Radiations, Clouds Impact on Radiations, Hydrological Cycle, and more …Hydrological Cycle, and more …
• Global cloud cover: 60%• Two competing effects:
• Reflect solar radiation back to space ~ 60 W/m2
• Trap infrared radiation emitted by the surface and low troposphere ~ 30 W/m2
• Depend on macrophysical and microphysical properties
• Type, location, altitude, amount
• Water content and Phase: ice or liquid? -> effective radius and optical depth
• A net cooling effect
Clouds in Climate Models - What are Clouds in Climate Models - What are the problems ?the problems ?
Many of the observed clouds and especially the processes within them are subgrid-scale processes (both horizontally and vertically)
GCM Grid cell 100-200km
Parameterization is needed
Slide from Joyce Penner and Adrian Tompkins (modified by Xie)
Clouds in GCM - What are the problems Clouds in GCM - What are the problems ??
convection
Clouds are the result of complex interactions complex interactions between a large number of processes
turbulence
Large scale dynamics
microphysics
radiation
Slide from Joyce Penner and Adrian Tompkins (modified by Xie)
Cloud Schemes - A Brief History
Slide from Joyce Penner
Cloud Schemes - A Brief History
Slide from Joyce Penner
60s
Condensation (non-convective)
qv > qs
Radiation effects
Prescribed zonal mean albedo and emissivity
Microphysics none
60s 70s
Condensation (non-convective)
qv > qs qv > qs
Radiation effects
Prescribed zonal mean albedo and emissivity
a diagnostic [usually f(RH)] ql prescribed
Microphysics none none
Cloud Schemes - A Brief History
Slide from Joyce Penner
Cloud Schemes - A Brief History
Slide from Joyce Penner
60s
Condensation (non-convective)
qv > qs
Radiation effects
Prescribed zonal mean albedo and emissivity
Microphysics none
60s 70s
Condensation (non-convective)
qv > qs qv > qs
Radiation effects
Prescribed zonal mean albedo and emissivity
a diagnostic [usually f(RH)] ql prescribed
Microphysics none none
60s 70s 80s
Condensation (non-convective)
qv > qs qv > qs ql prognostic a diagnostic
Radiation effects
Prescribed zonal mean albedo and emissivity
a diagnostic [usually f(RH)] q l prescribed
a = as cloud scheme
Microphysics none none Simple bulk microphysics
Cloud Schemes - A Brief History
Slide from Joyce Penner
Clouds: Still A Major Source of Clouds: Still A Major Source of Uncertainty in Climate ModelsUncertainty in Climate Models
Cloud Radiative Forcing: RAD_cld - RAD_clr at TOA
Figure shows globally averaged cloud radiative forcing changes for 2080-2090 under the A1B scenario for individual models.
A1B: one emission scenarios defined by IPCC (Intergovernmental Panel on Climate Change)
Cloud Radiative Forcing
Model ID Number
W/m2
Figure from IPCC Fourth Assessment Report (2007)
This Talk Focuses on This Talk Focuses on Arctic Mixed-Phase CloudsArctic Mixed-Phase Clouds
Arctic is experiencing the most Arctic is experiencing the most rapid changes in climaterapid changes in climate
Sea ice is declining faster than most IPCC models predict.
Stroeve et al. (2008)
Slide from Tony Del Genio
Why Mixed-Phase Clouds? Why Mixed-Phase Clouds?
Why Arctic Mixed-Phase Clouds?Why Arctic Mixed-Phase Clouds?
Klein et al. (2009)
Cloud phase is a major source of Cloud phase is a major source of uncertainty in modelsuncertainty in models
OBS
CAM3 – an earlier version of the NCAR Community Atmospheric Model (used before April 2010)
Rasch & Kristjansson (1998) single-moment to predict only mixing ratio of cloud condensate, liq/ice fraction determined by T
All ice when T < - 40C, all liq when T > -10C
AM2 – the climate model developed by GFDL (Geophysical Fluid Dynamics Laboratory)
Rotstayn (1997) and Rotstayn et al. (2000)
Single-moment, liq/ice fraction determined by the Bergeron process -- the ice crystal growth by vapor deposition at the expense of coexisting liquid water
CAm3Liu: An improved scheme for CAM3 (Liu et al., 2007) – part of the scheme being used in CAM4/CAM5
Double-moment to predict both mixing ratio and number density, liq/ice fraction determined by the Bergeron process (Rotstayn et al. 2000)
How Do Climate Models How Do Climate Models Determine the Cloud Phase?Determine the Cloud Phase?
Single-Moment vs. Double MomentSingle-moment: q
Double-moment : both q and N
Single-moment cannot represent aerosol-cloud coupling
The coupling requires a prognostic equation for the number concentration of cloud droplets so that the impact of aerosols on the cloud droplet number can be realistically represented
Aerosol-Cloud-Radiation interaction is one the key processes missing in many current climate models!
More on Cloud MicrophysicsMore on Cloud Microphysics
Bergeron (or Bergeron-Findeisen) ProcessA process that describes the formation of precipitation in
the cold clouds by ice crystal growth.
More on Cloud MicrophysicsMore on Cloud Microphysics
Water vapor, ice and liquid coexist in the mixed-phase clouds
esw > esi
In mixed-phase clouds, the air is saturation wrt the liquid droplets, but it is supersaturated wrt the ice crystals ==> water vapor will deposit on the ice crystals ==> leads to unsaturated air with respect to liquid ==> the liquid droplets will evaporate until the air once again reaches saturation. The process then continues.
In short summary, the ice crystal grows by vapor deposition at the expense of liquid water
How is the Bergeron Process How is the Bergeron Process Parameterized in Climate Models? Parameterized in Climate Models?
Bergeron process is parameterized based on Rotstayn et al. (2000)
dqi /dt ~ Ni , (esw – esi)/esi
AM2AM2: 1) Ni is diagnosed following Meyers et al. (1992)Ni = exp[12.96(esl - esi)/esi - 0.639]
2) Assume that the saturation vapor pressure is with respect to liquid esw=esl
CAM3LIU:
1) Ni is predicted by considering the processes of advection, convective transport, ice nucleation, droplet freezing, etc.
2) assume that the saturation vapor pressure is weighted by the proportions of ice and liquid water mass for mixed-phase clouds, ew = fl*esl + (1-fl)*esi
15-min Break15-min Break
A Schematic of the Model A Schematic of the Model Development ProcessDevelopment Process
Jakob, 2010
Climate Model EvaluationClimate Model Evaluation
Observational data is neededImproving mixed-phase cloud parameterizations
requires an advanced understanding of cloud and cloud microphysics through carefully planed field studies
Appropriate modeling approach is neededHow to link field data to global climate model evaluation and development?
The Mixed-Phase Arctic Cloud The Mixed-Phase Arctic Cloud Experiment (M-PACE)Experiment (M-PACE)
The DOE Atmospheric Radiation Measurement (ARM) program conducted a campaign at its North Slope of Alaska site to study the properties of mixed-phase clouds (10/5/04 – 10/22/04)
Barrow
More on M-PACEMore on M-PACE
Cloud Measurements
• Millimeter-wavelength cloud radar
• Micropulse Lidars
• Laser Ceilometers
• Aircraft
• Microwave Radiometers
M-PACE provides extremely valuable information to assess and improve model mixed-phase cloud parameterizations
How to link field data to model evaluations and developments?
The U.S. DOE CCPP-ARM Parameterization Testbed (CAPT) Project
CCPP (Climate Change Prediction Program)– developing, testing, and applying coupled-model for climate predictions
ARM (Atmospheric Radiation Measurement)– collecting field data for testing and improving model cloud and radiation parameterizations
CCPP+ARM Model +Data
CAPTCAPT
CAPT provides a flexible user environment for running climate models in NWP ‘forecast’ mode:
Climate models initialized with analysis data from NWP center’s data assimilation systems
A series of short-range weather forecasts performed The detailed evolution of parameterized variables compared
with field data link model deficiencies to specific atmospheric processes Evaluate the nature of parameterization errors before longer-
time scale feedbacks develop
What does CAPT do?What does CAPT do?
NCAR CAM3 FV 1.9x2.5 L26
GFDL AM2 2.0x2.5 L24
ModelsModels
• A series of 3-day forecasts with CAM3 and AM2 were initialized with the NASA Data Assimilation Office (DAO) analysis every day at 00Z for M-PACE.
• 12-36 hour forecasts near the Barrow site are analyzed
Barrow
Radar Clouds at BarrowRadar Clouds at Barrow
A: Multi-layer cloudsB: Persistent mixed-phase boundary layer cloudsC: Deep frontal clouds
B
CA
For mixed-phase clouds, the range of cloud temp is from -5 C ~ -20 C
A strong liquid layer occurred near cloud top at 1300m
Ice crystals in the liquid cloud layer and precipitating ice crystals beneath
Liq
Oct. 10, 2004
Aircraft Measured CWCAircraft Measured CWC
Ice
Simulated CloudsSimulated Clouds
(d)
• Cloud types• Cloud fraction
Simulated Cloud Liquid Simulated Cloud Liquid
Water Mixing RatioWater Mixing Ratio
(d)
AM2 clouds contain much less liquid than CAM models. Why?
Compared with CAM3Liu, CAM3 produces similar amount of liquid even though its cloud fraction is much lower than CAM3LIU
(d)
Simulated Cloud Ice Simulated Cloud Ice Water Mixing RatioWater Mixing Ratio
AM2: less ice for BLC
CAM3LIU more ice than CAM3
Liquid Water PathLiquid Water Path
AM2 contains much less liquid compared to CAM3-- AM2 may have a faster conversion rate of liquid to ice?
ice crystal number concentrationrelative humidity in mixed-phase clouds
Why does AM2 have less Why does AM2 have less
liquid than CAM3Liu? liquid than CAM3Liu?
Revisit the schemes
Bergeron process is parameterized based on Rotstayn et al. (2000)
dqi /dt ~ Ni , (esw – esi)/esi
AM2 assumes that the saturation vapor pressure is with respect to liquid esw=esl
CAM3LIU assumes that the saturation vapor pressure is weighted by the proportions of ice and liquid water mass for mixed-phase clouds, ew = fl*esl + (1-fl)*esi
esw in AM2 is larger than that in CAM3Liu leads to a faster conversion rate from liquid to ice through the Bergeron process
Liquid fraction as a function Liquid fraction as a function
of cloud heightof cloud height
Flights on 9-10 October for the single-layer mixed-phase clouds
Aircraft data: liquid dominates, fliq increases with height, ice seen in the lower half of clouds
• Snow component is added to the total cloud condensate to be consistent with aircraft data
• Normalized cloud height
Liquid fraction as a function Liquid fraction as a function
of temperatureof temperature
Flights on 9-10 October for the single-layer mixed-phase clouds
Aircraft data: no clear relationship, liquid and ice coexist within this temp range
The Bergeron process is critical for models to capture the observed characters of mixed-phase clouds
Surface and TOA LWSurface and TOA LW
(d)
Sensitivity Tests on Ice Sensitivity Tests on Ice Nucleation ParameterizationsNucleation Parameterizations
IN Parameterizations Largely Depend IN Parameterizations Largely Depend on Observationson Observations
IN measured in midlatitudes usually much larger than that observed in Arctic regions
Figure adapted from Prenni et al. (2007) BAMS paper
Prenni et al. (2007)
Meyers et al. (1992)
M-PACE
• Meyers et al (1992) produces significantly larger IN concentration than what observed during M-PACE
• Prenni et al. (2007) modified the Meyers et al. (1992) scheme to best fit the M-PACE data.
Meyers et al. (1992):
Ni = exp[12.96(esl - esi)/esi -
0.639]
AM2N90N – uses the Prenni et al (2007) scheme, which leads to a smaller ice nuclei number concentration
Sensitivity Test on IN – AM2Sensitivity Test on IN – AM2
Clouds Are SensitiveClouds Are Sensitive
Smaller IN Leads to Larger LWPSmaller IN Leads to Larger LWP
Sensitivity of Climate Sensitivity of Climate Simulations to IN SchemesSimulations to IN Schemes
CAM5 is usedCAM5 is used
More IN Schemes TestedMore IN Schemes Tested
Meyers et al. (1992): widely used in current climate models, an empirical formulation developed based on midlatitude measurements of ice nuclei concentrations, which are generally much larger than Arctic IN concentration.
Phillips et al. (2008): more physically based; link IN to aerosol (dust and soot) surface area, which generally gives much lower IN number concentrations than Meyers et al. (1992).
DeMott et al. (2010): link IN to aerosol particles (dust) large than 0.5 um based on more than 14-year observations over many regions of globe, which generally gives much lower IN number concentrations than Meyers et al. (1992).
CAM5 Climate Simulation
IN Concentration in Mixed-Phase Clouds
Meyers et al. (1992) Phillips et al (2008)
DeMott et al. (2010)Meyers et al. (1992) produces significantly larger IN number concentration than the other two schemes
Courtesy of Dr. Xiaohong Liu (PNNL)
LWP IWP
CAM5 Climate Simulation
Courtesy of Dr. Xiaohong Liu (PNNL)
Global Annual Means
-1.4 W/m2 (a cooling effect)
Courtesy of Dr. Xiaohong Liu (PNNL)
SummarySummary
Cloud microphysical processes need to be accurately represented in climate models, but this is a very challenging task
More physical based cloud microphysical schemes results in more accurate cloud simulations
Model simulated climate is sensitive to cloud microphysical schemes
Detailed observations and appropriate modeling approaches are needed to further improve our knowledge of cloud microphysics and their treatments in climate models