Presentation is loading. Please wait.

Presentation is loading. Please wait.

Robert Wood, University of Washington

Similar presentations


Presentation on theme: "Robert Wood, University of Washington"— Presentation transcript:

1 Robert Wood, University of Washington
Understanding spatial and temporal variability in cloud droplet concentration Robert Wood, University of Washington with Ryan Eastman, Daniel McCoy, Daniel Grosvenor (U. Washington); Matt Lebsock (JPL)

2 “Background” (minimum imposed) cloud droplet concentration influences aerosol indirect effects
LAND OCEAN Forcing [W m-2] A  ln(Nperturbed/Nunpertubed) Low Nd background  strong Twomey effect High Nd background  weaker Twomey effect Quaas et al., AEROCOM (Atmos. Chem. Phys., 2009) Hoose et al. (GRL, 2009)

3 What controls CCN and cloud microphysical variability in the marine boundary layer? A simple CCN budget for the PBL Entrainment Surf. Source Precip. Sink Assume nucleation/secondary processes unimportant Dry deposition is negligible (Georgi 1990) Sea-spray formulation (e.g. Clarke et al. 2006) Ignore advection Precipitation sink primarily from accretion process Equivalency of CCN and cloud drop conc. Nd Wood et al. (2012, J. Geophys. Res.)

4 Steady-state CCN budget
Free Tropospheric CCN Sea-Spray Production Precip. Sink Concentration relaxes to FT concentration NFT + wind speed dependent surface contribution dependent upon subsidence rate (D zi) Precipitation sink controlled by precipitation rate at cloud base PCB. Use expression from Wood (2006).

5 Precipitation important in controlling gradient in Nd
Assume constant FT aerosol concentration Precipitation from CloudSat estimates from Lebsock and L’Ecuyer (2011) Observed surface winds Model Nd gradients mostly driven by precipitation sinks Wood et al. (J. Geophys. Res. 2012)

6 Precipitation is primary driver of geographical variability in mean Nd away from coasts
Model reproduces significant amount of variance in Nd over oceans  implications for interpretation of AOD vs re relationships Model (fixed FT aerosol) Wood et al. (J. Geophys. Res. 2012)

7 Does the precipitation sink drive seasonality in Nd?
Most subtropical stratocumulus regions exhibit significant seasonality in cloud droplet concentration that is anticorrelated with precipitation rate Droplet conc. r [Nd , Rcb] = -0.85 r [Nd , Rcb] = -0.84

8 Steady state CCN/Nd model prediction
Surface CCN flux Southeastern Pacific (10-30oS, oW) Free-tropospheric CCN Model predicted (NFT=125 cm-3) seasonality driven by precip only Entrainment rate 𝑁= 𝑁𝐹𝑇+ 𝐹 0 𝑤 𝑒 1+ 𝑆 precip MODIS observed Steady state CCN conc in MBL Non dimensional precip sink Adapted from Wood et al. (2012) Steady state CCN/Nd budget shows skill in predicting SE Pacific Nd assuming seasonally invariant FT aerosols. Application to other regions challenging Unknown FT CCN seasonality constraints Problems with mixed phase precipitation

9 Steady state CCN/Nd model prediction
Surface CCN flux Southeastern Pacific (10-30oS, oW) Free-tropospheric CCN Model predicted (NFT=125 cm-3) seasonality driven by precip only Entrainment rate 𝑁= 𝑁𝐹𝑇+ 𝐹 0 𝑤 𝑒 1+ 𝑆 precip MODIS observed Steady state CCN conc in MBL Non dimensional precip sink Adapted from Wood et al. (2012) Steady state CCN/Nd budget shows skill in predicting SE Pacific Nd assuming seasonally invariant FT aerosols. Application to other regions challenging Unknown FT CCN seasonality constraints Problems with mixed phase precipitation

10 Strong seasonal cycle of aerosols and cloud droplet concentration Nd over the Southern Ocean
Marked annual cycle of Nd in low clouds over Southern Ocean Summer Nd maximum hypothesized to be biogenic (DMS, organics) In situ and satellite observations consistent Summertime albedo enhancement (Twomey) of 25% Figure by R. Wood, SOCRATES White Paper (2014)

11 CloudSat precipitation (2c-precip-column)
Wintertime maximum for low cloud precipitation likely, but annual cycle not hugely strong (range: mm day-1) Outstanding retrieval problems over Southern Ocean Cold-topped clouds, radar echoes below 800 m altitude contaminated by ground clutter

12 Lagrangian framework 𝜏=−𝑇/ ln 𝑟  30 hours
Treat anomalies of cloud droplet concentration Nd as a red noise process where 𝜏 is the Lagrangian decorrelation timescale. Slope of initial (t = 0) vs final (t = T = 24 hr) anomalies (right) provides the value of r, then 𝜏 for low cloud cover is much shorter ( hours). 50 40 30 20 10 -10 T=24 hr (24 hour trajectories) 𝑟[𝑁 𝑑 ′ 𝑡+𝑇 , 𝑁 𝑑 ′ (𝑡)]= 𝑒 −𝑇/𝜏 1:1 slope (r = 1   = ) Note: linear slope  linear decay process (mean rate of decay of initial signal does not depend on the amplitude) (r = 0.45  𝜏 = 30 hr) Observed slope Nd anomaly at t = 24 hr [cm-3] Zero slope (r = 0   = 0) 𝜏=−𝑇/ ln 𝑟  30 hours Nd anomaly at t = 0 [cm-3] Eastman et al. (2015) Clouds therefore decorrelate faster than the aerosol they are forming on

13 Timescale for precipitation removal
(Wood 2006, J. Geophys. Res.) CALIOP Cloud top height (Muhlbauer et al. 2014) 𝜏 𝑐𝑜𝑎𝑙 = 𝑁 𝑁 = 𝑧 𝑖 𝐾ℎ 𝑃 𝐶𝐵  2/PCB 𝜏 𝑐𝑜𝑎𝑙  2 days for PCB = 1 mm day-1 Values: K = 2.25 m2 kg-1 (Wood 2006); zi = 1500 m; h = 350 m (typical values) Cloud top height [km]

14 Take-home points Light precipitation from low clouds exerts major control on CCN and cloud droplet concentration Drives geographical variation of the annual mean Nd away from coastal zones Drives Nd seasonal cycle in some regions (e.g. SE Pacific, possibly S. Ocean) Lagrangian decorrelation timescale for Nd (30 hours) is substantially longer than for cloud cover Clouds decorrelate faster than the underlying aerosol that they ingest Decorrelation timescale comparable with timescale for precipitation removal but next steps will apply Lagrangian framework analysis to explore connection between precipitation and Nd

15

16 Additional slides

17 Correcting solar zenith angle biases in MODIS-derived Nd
Grosvenor and Wood (Atmos. Chem. Phys. 2014)

18 Open cells drizzle harder, but more intermittently
Muhlbauer et al. (2014)

19 What factors control the magnitude and uncertainty of the global first aerosol indirect effect?
Ghan et al. (J.Geophys. Res., 2013)

20 Natural emissions contribute half of AIE uncertainty
(a) Seasonal cycle; (b) contribution from natural, anthropogenic emissions and aerosol processes; (c) uncertainty ranges from different perturbed parameters Volcanic and DMS produced SO2 are major natural sources of uncertainty Anthropogenic SO2 key anth. Source Aerosol processes are not major sources of uncertainty in this analysis Carslaw et al. (Nature, 2013)

21 Does the precipitation sink drive seasonality in Nd?
Surface CCN flux Steady state CCN/Nd budget shows skill in predicting SE Pacific Nd (Wood et al. 2012) Application to other regions challenging Poor FT CCN constraints Aforementioned issues with warm rain estimates Free-tropospheric CCN Entrainment rate 𝑁= 𝑁𝐹𝑇+ 𝐹 0 𝑤 𝑒 1+ 𝑆 precip Steady state CCN conc in MBL Non dimensional precip sink Adapted from Wood et al. (2012) MODIS/CloudSat observations Key result is that Sprecip is 1-2 for mean drizzle rate of 1 mm day-1


Download ppt "Robert Wood, University of Washington"

Similar presentations


Ads by Google