Satellite analysis of aerosol indirect effect on stratocumulus clouds over South-East Atlantic

Transcription

1 1/23 Remote sensing of atmospheric aerosol, clouds and aerosol-cloud interactions. Bremen, December 2013 Satellite analysis of aerosol indirect effect on stratocumulus clouds over South-East Atlantic Lorenzo Costantino(1) and François-Marie Bréon(2) (1) CEA, DAM, DIF, Arpajon (2) Laboratoire des Sciences du Climat et de l'environnement (LSCE), Saclay France, mixed laboratory CEA CNRS UVSQ Costantino and Bréon (2011), Geo. Res. Lett. Costantino and Bréon (2013), ACP Costantino and Bréon (2013), ACPD

2 2/23 Introduction

3 3/23 Aerosol AEROSOL: complex and dynamic mixture of tiny solid and liquid particles that float in the atmosphere VOC from vegetation Desert dust Volcanic ash Smoke from fires Industrial pollution Sea salt Can be transported by wind to very long distance Strong temporal and spatial variability, variability over both land and ocean. Size down to 10 ¹ µm

4 4/23 Aerosol Impact on Climate SMOKE and INDUSTRIAL POLLUTION can interact with solar radiation in two ways: INDIRECT EFFECT: Acting as CCN aerosol can 1/ modify cloud microphysics and, in turn, cloud reflectivity (INDIRECT EFFECT #1) 2/ affect cloud structure and life cycle and, in turn, cloud reflectivity + cloud cover (INDIRECT EFFECT #2) DIRECT EFFECT: Scattering and absorption of solar radiation

5 5/23 Radiative Forcing Aerosol impact on climate change: GHG Results of IPCC report 2007 Aerosol 1/ SECOND most impostant ATHROPOGENIC FORCING after GHG (opposite in sign) 2/ The consequence of aerosol-cloud interaction is the PRIMARY UNCERTAINTY

6 6/23 Aqua/MODIS 09/01/ :45 UTC Fires in Central and Southern Africa Aqua/MODIS 04/01/2011 One of the 14:05 bestutc regions to observe cloud-aerosol Dust and firesinteraction... across Sahel Emission - Transport Dec-Feb Fires in West Africa Fires in Southern Africa Emission - Transport July-Sept Terra/MODIS 26/09/ :20 UTC Smoke and clouds over Southern Africa Aqua/MODIS 03/08/ :20 UTC

7 7/23 Fire occurrence Wind speed 0.5 km Wind and Fires 1.5 km 2.5 km Aerosol Optical Depth Efficient trasport of aerosol particles over the ocean!!! Costantino and Bréon, 2013

8 8/23 Biomass Burning (BB) aerosol Produced by: Savanna and cropland fires. Contain: OC (Organic Carbon): major component (~50% soluble). BC (Black Carbon): insoluble dust, ash, soluble salt. Primary emitted in efficient flaming fires (a more efficient combustion increases soot surface oxidation, that leads to a stronger chemical reactivity and water uptake). Large fraction of soluble material already very efficient CCN immediately after the fire.

9 9/23 Aerosol Transport over SE-Atlantic 2300 km North-East South-West Altitude [km] 8 6 Aerosol 4 Cloud 2 0 Longitude Land Costantino and Bréon, 2010 Ocean

10 10/23 Multisensor Monitoring MODIS L2 CLOUD product (1-5 km resolution) MODIS L2 AEROSOL product (10 km resolution) Aerosol Optical Depth Cloud Optical Thickness 00 01' 15'' September, 14, 2010, from 13:55 to 14:04 UTC Broken cloud layer Cloud Aerosol Cloud 20 k m CALIPSO L2 CLOUD-AEROSOL product (5 km resolution) Aerosol Well separated d > 250 m Cloud Mixed layers d < 100 m Aerosol Cloud

11 11/23 INDIRECT EFFECT #1: Impact on Cloud Microphysics

12 12/23 INDIRECT EFFECT #1: Impact on Cloud Microphysics Assuming a constant liquid water content N c N 0.7 a Satellite-derived relationship (Kaufman et al., 1997) Δ log N c Δ log r e= 3 N a AI Δ log r e = 0.23 Δ log AI

13 13/23 June 2006 Decembre 2010 (~ coincidences) CDR AI relationship Mixed Number of retrievals (2 grid box) Unmixed 4N -2N -15N -30N Mixed Unmixed Δ log r e 0.15 Δ log AI Aerosol Not interacting Interacting CDR = µm (-31%) <CDR> = 14.5 µm Costantino and Bréon, 2013 d > 750 m Aerosol R = R = d > 250 m Cloud CDR AI theoretical realtionship: Δ log r e 0.03 Δ log AI Δ log r e = 0.23 Δ log AI

14 14/23 CDR AI relationship PARASOL (CDR) MODIS (AI) CALIPSO (vertical position ) Costantino and Bréon, 2010 Not interacting Interacting June 2006 Decembre 2008 (~ coincidences)

15 15/23 LWP AI Relationship INDIRECT EFFECT #2: Impact on Cloud Water Content Simple idea (Albrecht's hypothesys): More aerosol smaller particles less collisioncoalescence efficiency less rain more water in the cloud

16 16/23 LWP AI relationship MODIS-CALIPSO concidences (33000) LWP = g/m² (-37%) LWP = 90/80 g/m² Costantino and Bréon, 2013 Not interacting Δ log LWP 0.16 Δ log AI Interacting Δ log LWP 0.04 Δ log AI LWP decreases (drying) with increasing aerosol concentration!! Opposite result with respect to Albrecht's hypothesis (moistening effect) Is this an aerosol-induced effect?? I am positive, but...

17 17/23 Droplet Evaporation Boundary layer height Drying from increased entrainment of dry air (cloud top) Extremely dry air Easterly trade wind ( HPa) Entrainment of dry air increases with increasing Nc (Ackerman, 2004) verify with WRF-Chem??? Moistening from decreased precipitation (cloud base) Leading factor of LWP response to aerosol enhancement: humidity above the inversion

18 18/23 COT AI Relationship Aerosol effect on of cloud reflectance Costantino and Bréon, 2013 No evident correlation between COT and AI AER-CLD interaction: weak radiative impact!!! 2 LWP τ = c 3ρ r w e Δ ln τ c Δ ln LWP Δ ln r e = Δ ln AI Δ ln AI Δ ln AI = 0.01

19 19/23 Aerosol impact on Cloud Fraction and Precipitation (... a more difficult issue.. )

20 20/23 INDIRECT EFFECT #2: Impact on Cloud Lifetime Aerosol effect? AI from 0.1 to 0.5 CLF increase of ~ 55 % Aerosol contamination (for AOD > 0.7) Adjacent (blueing) effect Humidity (Swelling effect) Surface wind Meteorology? Low troposferic stability High pressure systems Other types of meteorologically driven co-variation of CLF and AI Vertical developpement Artifact? Horizontal extension Overestimated (wrt MODELS) Increasing aerosol index Aerosol effect on of cloud fraction (CLF AI) AI= Vertical developpement AI= Horizontal extension Costantino and Bréon, 2013

21 21/23 CLF AI relationship CONSTANT Cloud Top Pressure Costantino and Bréon, 2013 Cloud Fraction Sensitivity : Δ ln CLF Low cloud cover increases with increasing low tropospheric stability (Klein and Hartmann, 1993) Δ ln AI Aerosol above cloud: increasing with decreasing altitude WARMING Up to 3.5 Kd(at 700 hpa) Mixed case: small but CONSTANT Data sorted by CTP, from 1000 to 600 HPa, by step of 15 HPa Cloud cover response to aerosol invigoration seems to depend on aerosol vertical position (and radiative effect)

22 22/23 Inhibition of Precipitation PRECIPITATION OCCURRENCE (Lohmann et al., 2000): change in sign of CDR COT relationship slope from POSITIVE to NEGATIVE Thin clouds Clean clouds carry more water (15%) Non precipitating (Adiabatic assumpion) 0.2 c re τ N 0.5 Precipitating (Constant LWP with increasing COT) re τ 1 c Thick clouds Polluted clouds carry more water (15%) r e τ 0.11 c Less precipitating: polluted clouds (mixed case with AI > 0.09). r e τ 0.43 c More precipitating: clean clouds. Costantino and Bréon, 2013

23 23/23 Summary and Conclusions - We used satellite data to analyze aerosol-cloud interaction - and CALIPSO information to distinguish between mixed (interacting) and unmixed (non interaction) layers - Large impact of Aerosol on CDR, in-line with theoretical expectations - No evident Aerosol impact on cloud reflectance (albedo) SMALL RADIATIVE EFFECT - Strong correlation between CLF and AI, but further analysis indicates this is probably not an effect due to aerosol-cloud interaction - Aerosol seems to induces a decrease in precipitation efficiency only in optically thick clouds (tau > 10)

24 24/23 Thank you for your attention! QUESTIONS?? Aerosol Cloud For more details: Costantino and Bréon (2011), Geo. Res. Lett. Costantino and Bréon (2013), ACP Costantino and Bréon (2013), ACPD mail to:

25 25/23 Impact of meteorology For very low AI, CDR LWP, COT converge to the same values. In particular mixed and unmixed CTP (the cloud parameter mostly linked to background meteorology) are quite close, for every aerosol regime. Costantino and Bréon, 2013 This result suggests a uniform impact of meteorology on both populations: changes in cloud properties (when aerosol and clouds intermingle) are most due to aerosol-cloud interaction.

26 26/23 Impact of aerosol above clouds Error is may be wavelength dependent: Haywood et al. (2004): using the 0.86/2.1 µm couple of wavelengths, CDR is very little underestimated (< 1 µm), COT is underestimated by 10-20%. Cloud dependent: Coddington et al. (2010): the error in CDR is less than 1 µm and that in COT is within the uncertainties of the instrument (MODIS and SSFS, on board of a airplane flying between the aerosol layer and the cloud top) in regions with small cloud variability (as S-E Atlantic). Errors are much larger in case of strong cloud heterogeneity (up to 10 µm and 10). Pollution dependent: Meyer et al. (2013): the error in CDR and COT for polluted clouds is 6% and 18%, and 2.6% and 11% for clean + polluted. In the present study we use 0.86/2.1 µm, cloud field is supposed to be quite homogeneous (confirmed by PARASOL measurements), and unmixed case is composed only of clean clouds (no multi-layer cloud scenes): in case of aerosol above clouds CDR, LWP, COT seems to be almost insensitive to large AI variations (while we should expect a decrease in CDR and COT), while mixed CDR variation is about 30%.