The importance of precipitation in marine boundary layer cloud Robert Wood, Atmospheric Sciences, University of Washington

Motivation Marine boundary layer (MBL) clouds cover about 1/3 of the world’s oceans and have an enormous impact on top-of-atmosphere (TOA) and surface radiation budgets the general circulation How clouds change remains one of the major uncertainties in future climate prediction Until recently, precipitation in MBL clouds was assumed to be of secondary importance – this view is changing

SST anomaly from zonal mean ERBE net cloud forcing

ISCCP inferred St/Sc amount

Tropical-subtropical general circulation from Randall et al., J. Atmos. Sci., 37, 125-130, 1980 warm SST cold SST

SST and wind stress coupled ocean-atmosphere GCM Prescribed ISCCP clouds SST and wind stress coupled ocean-atmosphere GCM Climatology Model clouds from Gordon et al. (2000)

Clouds in climate models - change in low cloud amount for 2CO2 GFDL Clouds in climate models - change in low cloud amount for 2CO2 CCM model number from Stephens (2005)

Precipitation in MBL clouds? Pioneering study by Albrecht (1989) importance of drizzle in cloud thermodynamics suggestion of microphysical controls upon cloud coverage/lifetime Early 1990s saw the development of sensitive radars that can detect even light drizzle (few tenths of a mm/day) Petty (1995) highlighted prevalence of drizzle in volunteer ship observer reports

Fraction of precipitation reports indicating “drizzle” Drizzle is prevalent form of precip. in MBL cloud regions 0% 10% 20% 30% 40% 50% >50%

Field campaigns with focus on low clouds ISCCP stratus/stratocumulus cloud amount

Low cloud amount (MODIS, Sep/Oct 2000) The southeast Pacific Low cloud amount (MODIS, Sep/Oct 2000)  Mean cloud fraction  Mean MBL depth

The EPIC Stratocumulus study Part of the East Pacific Investigation of Climate (EPIC) field program Ship cruise (NOAA R/V Ronald H Brown,10-25 October 2001) under the stratocumulus sheet Surface meteorological measurements, 3 hourly radiosondes, aerosols Suite of remote sensors: scanning C-band radar, 35 GHz profiling radar (MMCR), lidar, ceilometer, microwave radiometer Bretherton et al. (2004), BAMS

Drizzle challenges What is the frequency and strength of drizzle over the subtropical oceans? What are the structural properties of precipitating MBL cloud systems? Can drizzle affect cloud dynamics, structure and coverage - how does it do so? What controls drizzle production in MBL clouds?

EPIC Sc. visible reflectance (MODIS) SST (TMI) & winds (Quikscat) Wood et al. (2004)

Diurnal cycle and drizzle Surface-derived LCL Ceilometer cloud base

Quantification of drizzle

Quantifying drizzle Marshall-Palmer Z-R relationships derived using MMCR are then applied to the scanning C-band radar

Quantifying drizzle

Structural properties of precipitating stratocumulus

u 20 km 10 km

Mesoscale dynamics dBZ 0 10 20 30 [km] 23:09 UTC VRAD [m s-1] 1.5 km -10 -5 0 5 10 15 dBZ 23:09 UTC VRAD [m s-1] -3 -2 -1 0 1 2 3 1.5 km 23:18 UTC 0 10 20 30 [km]

Animation of scanning C-band radar 30 km mean wind

Echo Tracking Comstock et al. (2004)

Structure and evolution of drizzle cells Time to reflectivity peak (hours) Average cell reflectivity (dBZ) 15 10 5 -1.5 -1 -0.5 0 0.5 1 1.5 Drizzle cell lifetime 2+ hours Time to rain out < ~ 30 minutes Implies replenishing cloud water This plot shows average reflectivity with time for 8 example cells, plotted with respect to the time at which each cell reached its peak average reflectivity. We didn’t capture the entire lifecycle of any single drizzle cell, but we have parts of the life cycle from different cells From this plot, we can infer that the lifetime of the drizzle cell is a little over 2 hours. Shallow marine cumulus clouds form quickly, reach a peak reflectivity, and then rain themselves out within 10-15 minutes. If this were to occur in these clouds, the rain-out time would take about 30 minutes, due to their lower rain rates. In fact, these example cells are actually drizzling during the entire lifecycle captured here (if we can see them on the C-band radar, then they are drizzling). So they are continuing to grow even while they are losing water due to drizzle. This implies that the clouds are being replenished even as they drizzle. --------------------------------------------------------------------- (time to rain out = LWP/rain rate in kg/m^2 s, LWP ~ 300 g/m^2, R = 0.3-1 mm/hr = 0.3-1 kg/m^2 s, time = 20-60 min) Comstock et al. (2004)

Can drizzle affect MBL dynamics?

What controls drizzle production?

Pollution plumes in the SE Pacific Chile is world’s largest copper producer Copper smelting SO2 emissions from Chile (3 Tg yr-1) comparable to total SO2 emissions in Germany 90% of Chilean SO2 emissions from seven smelters! Andes mountains prevents eastward transport Put Seasonal Mean SON

How rapidly are CCN lost through coalescence?

Summary of drizzle observations from previous field programs

Open Cells Closed Cells Satellite Ship Radar

Drizzle and cloud macrostructure MODIS brightness temperature difference (3.7-11 mm), GOES thermal IR, scanning C-band radar

Summary Precipitation is common in MBL clouds The mean precipitation rates 1 mm day-1 are observed and can have significant thermodynamic impact upon the MBL Precipitating MBL clouds display interesting mesoscale dynamics that may influence their macroscopic properties Results suggest that drizzle is modulated by cloud LWP and by cloud droplet number

Future directions Broaden the scope of EPIC using a combination of satellite remote sensing, reanalysis, and buoy data (NSF funded, 2004-2007) Plan and participate in a more extensive field program in the SE Pacific (VOCALS 2007) Use Cloudsat (launch summer 2005) to begin to develop climatologies of precipitation in low cloud

Fraction of areal mean precipitation observed How long do we need to average?