Investigation of Microphysical Parameterizations of Snow and Ice in Arctic Clouds During M-PACE through Model- Observation Comparisons Amy Solomon 12 In collaboration with: Hugh Morrison 3, Ola Persson 12, Matt Shupe 12, Jian-Wen Bao 1 1 NOAA/Earth System Research Laboratory, Boulder, Colorado 2 CIRES/University of Colorado, Boulder, Colorado 3 National Center for Atmospheric Research, Boulder, Colorado Investigation of Microphysical Parameterizations of Snow and Ice in Arctic Clouds During M-PACE through Model- Observation Comparisons Amy Solomon 12 In collaboration with: Hugh Morrison 3, Ola Persson 12, Matt Shupe 12, Jian-Wen Bao 1 1 NOAA/Earth System Research Laboratory, Boulder, Colorado 2 CIRES/University of Colorado, Boulder, Colorado 3 National Center for Atmospheric Research, Boulder, Colorado

Models that include prognostic equations for cloud water, rain, ice, and snow mixing ratio and number concentration (two- moment) successfully model horizontally extensive, persistent mixed-phase Arctic stratocumulus Observations Two-Moment One-Moment ( After Morrison and Pinto 2006) Impact of Microphysical Parameterization on Simulations of Arctic Stratocumulus Climate models fail to produce Arctic clouds that maintain liquid water at low temperatures (e.g., Prenni et al. 2006, Inoue et al. 2006, Sandvik et al. 2007, Klein et al. 2008)  Since cloud liquid water causes an increase in the downwelling longwave radiation and a decrease in the incoming shortwave radiation  Inadequate simulations of Arctic clouds causes significant errors in the modeled surface energy budget

ObjectivesObjectives  Process-oriented validation of microphysical parameterizations Specifically, in this study:  Verify the simulations of mixed-phase clouds and boundary layer structure during M-PACE with the WRF V2.2  Comparisons of LWP/IWP  Comparisons of SWD/LWD  Vertical structure and temporal variability of LWC and IWC  Variability of vertical velocities  Variability of LWP/IWP  Investigate the sensitivity of the model simulations to the parameterization of snow/ice  Sensitivity to mid-latitude N 0s  Sensitivity to two-moment N 0s  Sensitivity to approximate observed N 0i

Prognostic variables include mixing ratios and number concentrations of cloud ice, cloud droplets, snow, graupel, and rain Hydrometeors have the form of a complete gamma size distribution:  (D) = N 0 D Pc e - D =  cN  (P c +d+1)  1/d  q  (P c + 1)  = the Slope Parameter N 0 = N Pc+1 is the Intercept Parameter Pc is the Spectral Parameter (Pc=0 for ice, snow, rain, graupel) Bulk density of ice = 0.5 g cm -3 Bulk density of snow = 0.1 g cm -3 (Now available as a standard option in WRF V3) Microphysical Scheme

Control Set-up Weather Research Forecast Model V2.2  Nested 18/6/1 km horizontal grids  50 vertical levels (20 levels below 800 hPa)  Morrison Two-Moment microphysics  CAM Radiation  3D PBL mixing in the 1km grid (YSU in 18/6 km)  NOAH LSM

Control Set-up and Horizontal Structure in 1 km Domain Weather Research Forecast Model V2.2  Nested 18/6/1 km horizontal grids  50 vertical levels (20 levels below 800 hPa)  Morrison Two-Moment microphysics  CAM Radiation  3D PBL mixing in the 1km grid (YSU in 18/6 km)  NOAH LSM

Impact of Horizontal Resolution on the Maintenance of Liquid Water 1 KM 50 KM 24-hour average 1-hour average

Vertical Velocity and Growth of Cloud Droplets  Hourly-averaged 2M vertical velocity at Barrow from Oct Z to 10 11Z, in units of cm s -1 Hourly-averaged 2M vapor deposition with LWC is shown with dashed blue contours, in units of g m -3 W (cm/s)Vapor Droplets (g/m3/hour)

STD(W) at different averaging periods Retrievals Two-Moment Validation of Dominant Timescales for Vertical Motions and Ice/Liquid Interactions Correlation between IWP and LWP at different averaging periods  Model underestimates variability on timescales shorter than 2 minutes  And overestimates variability on timescales between 2-30 minutes  An aliasing of fast processes to longer timescales?

STD(W) at different averaging periods Retrievals Two-Moment Validation of Dominant Timescales for Vertical Motions and Ice/Liquid Interactions Correlation between IWP and LWP at different averaging periods  Observations show significant correlations between IWP and LWP for timescales shorter than 12 minutes  The model shows significant correlations between IWP and LWP on all timescales  Another indication that the model is aliasing fast processes to longer timescales?

Validation of Liquid and Ice Water Content Retrievals Morrison 2M Microphysics IWC LWC

Validation of Water Paths and Surface Radiation SW LWP IWP LW Observations 2M Morrison 1M Morrison

Validation of Cloud Droplet, Snow, and Ice Size Distributions Two-moment scheme does a good job of simulating the observed cloud droplet and snow size distributions In general, N0 is not known apiori and varies in space and time! N0 for ice is underestimated by an order of magnitude… Aircraft measurements Two-Moment (After McFarquhar et al. 2007) N 0i ? N 0s ?

Mixing Ratio of Snow Mixing Ratio of Cloud Droplets Two-moment N 0s (solid) One-moment N 0s =8.e5 m -4 (dash) Averaged for LWC>0.01 g m -3 at Barrow, Alaska Impact of Specifying Two-Moment N 0s

Two-moment N 0i (solid) One-moment N 0i =5.e7 m -4 (dash) at Barrow, Alaska Aircraft Measurements of ice number concentration (d >53 microns): Mean (white line) and +/- one standard deviation (grey shading) Impact of Specifying Observed N 0i 24-hour mean ice number concentration24-hour mean snow number concentration

FindingsFindings 1)Morrison two-moment microphysics does a good job of simulating the LWC in Arctic stratocumulus observed during M-PACE, as well as, the dominant initiation and growth mechanisms for cloud water, snow, and ice. 2)Cloud water forms at the cloud base and grows in updrafts. Coarser resolution runs (>18 km) do not resolve this process because the vertical velocity needed to lift the cloud water is underestimated, resulting in clouds that form closer to the ground. 3)Specifying N 0s to typical mid-latitude values depletes the cloud liquid water--The two- moment scheme is able to reproduce the observed LWC because it does a good job of predicting N 0s, which is advantageous since N0 is poorly constrained by observations and varies in space and time. 4)Ice number concentration is underestimate by an order of magnitude in the two- moment simulation and there is an indication that variability is being aliased to longer (> 2 minutes) timescales. 5)Specifying the ice size distribution observed during this period in a one-moment simulation acts to deplete the liquid water, indicating that the vapor deposition to ice may be occurring too rapidly in the model. 6)Specifying the two-moment snow size distribution in a one-moment simulation results in similar snow mixing ratio and number concentration but a depletion of cloud water due to the coupling between mixing ratio and number concentration in the one- moment scheme.

Structure of Arctic clouds  Mixed-phase clouds dominate the low- cloud fraction within the Arctic during the colder three-quarters of the year (Curry et al. 2000; Intrieri et al. 2002; Uttal et al. 2002; Wang et al. 2005)  Arctic low-level mixed-phase clouds tend to be long-lived and are not observed to glaciate quickly due to the Bergeron process (Pinto 1998; Hobbs and Rangno 1998; Curry et al. 2000)  Liquid layer were observed in the M-PACE experiment at temperatures down to -34°C (Verlinde et al. 2007)  Mixed-phase clouds are observed to occur in regions of both strong and weak surface forcing, indicating that the cause of these cold, ice-precipitating clouds is microphysical in nature (Harrington et al. 1999; Morrison et al. 2005) (After Shupe et al. 2006) Observations of Mixed-Phase Arctic Stratocumulus

Case Studied: Mixed-Phase Arctic Cloud Experiment Intensive Observations Oct : Measurements at DOE ARM NSA Site + High Spectral Resolution Lidar + Atmospheric Emitted Radiance Interferometer + Radiosonde launches + Two Instrumented Aircraft with a Compliment of Cloud Physics Probes  A strengthening high-pressure system north of Alaska  Caused air to flow from pack ice over the open Beaufort Sea to the North Slope of Alaska  Forcing roll clouds that extended from the pack ice to the North Slope of Alaska  These clouds are aligned closely to the direction of the boundary layer winds  With wavelength of 10-15km and a PBL 1km over the ocean and <1km inland  These clouds were mixed phase with total water content dominated by liquid hydrometeors throughout the cloud layer MODIS visible image Oct 9 0Z 2004