Atmospheric Dust Sources

Transcription

1 Dust/Climate Interactions Atmospheric Dust Sources Wind, precipitation, vegetation CLIMATE DUST Ina Tegen Institute for Tropospheric Research Leipzig, Germany Direct and indirect radiative forcing SOLAS Summer School 2005, Cargese September 7, 2005 CO 2 drawdown OCEAN Fe input 1 2 Dust Particle Images Saharan dust storm from Space Shuttle, 1992 Global soil dust aerosol: Dust particle sizes typically 1-10µm clay FeO(OH) Global annual dust emissions: Mt/yr Quartz Atmospheric lifetime: ~hours to 2-3 weeks SeaWifs 3 Busek and Posfai,

2 SeaWifs,, China Sahara Shanton, TX, Big Spring, TX, Seawifs,, Modis, Takla Makan Red Sea Dead Sea Midura, Australia, Modis,, Aerosol Signature from Terra Satellite Satellite Retrieval of Absorbing Aerosol TOMS Absorbing Aerosol Index

3 Satellite Retrieval of Dust Aerosol Advantage: -Global coverage Dust Deposition Data Mass Accumulation Rate = Dust Fraction * Density * Sedimentation Rate DIRTMAP Late Holocene Accumulation Rates Problems: -Aerosol mixture- -size(?) -refractive index? -shape?? -Vertical distribution (TOMS) -Cloud cover New satellite instruments and retrieval methods are being developed, remote sensing of aerosol properties is improving. But: Only airborne dust retrieved, not deposition flux g/m2 /y 9 Problems: Temporal resolution, gaps (Kohfeld et al.) 10 Estimates of Global Annual Dust Emissions Goal: Determine fluxes of soil-dust iron (bioavailable) into the world s oceans for modern and past time periods. Problem: Measurements are sparse and dust loads too variable regionally and temporally => problematic to obtain global estimates from measurements alone. Needed: Model of the global dust cycle, simulating dust sources, transport and deposition, then: Fe fluxes=dust deposition fluxes Fe content Fe solubility Duce (1995), IPCC (2001) 11 modified from Duce (1995) 12

4 Dust Sources Principal Dust Source Regions ¾ Sahara ¾ Sahel ¾ Arabian Peninsula ¾ Thar desert (Middle East) ¾ Aral Sea (Central Asia) ¾ Taklamakan desert (China) ¾ Gobi Desert (China/Mongolia) ¾ Lake Eyre Basin, Australia ¾ North America Global Distribution of Unvegetated Areas Modelling Dust Sources Required for occurrence of dust emission: Unvegetated, dry soil Annual dust sources No dust emission Seasonal dust sources 15 (Data from AVHRR NDVI, ) 16

5 Modelling Dust Sources Global Distribution of 10m Wind Speed Required for occurrence of dust emission: Unvegetated, dry soil High surface wind speeds, vertical mixing 17 (Data from ECMWF ERA15 reanalysis) 18 TOMS Absorbing Aerosol Index Dust sources Required for occurrence of dust emission: Unvegetated, dry soil High surface wind speeds, vertical mixing Loose, fine sediment, smooth surface If considering only wind speed and vegetation, hot spots in dust observations cannot be adequately reproduced 19 20

6 Preferential Source Areas (Topographic depressions) Tegen et al., (2002) Global Dust Source Areas Processes of Soil Particle Movement Annual dust sources Seasonal dust sources No dust emission Preferential source (Numbers in brackets indicate typical particle diameters) Pye (1987) 23 24

7 Transport Modes of Dust Particles Wind Friction Velocity u * = u( z ref ) ln( z κ / z ref 0 ) u: surface wind speed z ref : reference height, usually 10m κ: van Karman constant, 0.4 z 0 : surface roughness Pye (1987) Minimum u* Required for Dust Emission Dependence on soil particle radius Size Distribution of Airborne Dust Measurements of Saharan dust Continental aerosols Schuetz et al. (1981) Source Seinfeld (1997) Modeled size distribution Remote after Iversen and White (1982) Tegen et al. (2002)

8 Vertical Dust Flux from Wind Tunnel Experiments Emission Factors (Tegen et al, 2002) α=vertical flux/horizontal flux Coarse sand α=10-7 cm -1 Medium/fine sand α=10-6 cm -1 Silt α=10-5 cm -1 Clay (if < 45%) α=10-6 cm -1 Clay (if > 45%) α=10-7 cm -1 Data from Gillette (1977) Dependence of threshold wind friction velocity for on surface roughness Dust emission depending on preferential source areas ( optimum texture ) vegetation cover (explicit dependence) surface wind speed soil particle size soil moisture after Marticorena and Bergametti (1995) 31 32

9 Some dust emission schemes used in global models 2 Tegen et al., 1994: F = ( Ci ( u 6.5m / s) u ) C: Calibration i constant Marticorena et al., 1997 (Sahara) Ginoux et al, 2001 Tegen et al., 2002 Balkanski et al., ρ u = 3 u* tr + i * tr i F α u* 1 1 si g i u 2 * u * F = CS i 2 [( u u (soil moisture)) u ] tr s i i u 2 u ρ tri tri F u = 3 * * α s * i Veff g i 1 + u 1 2 u * * 2 F = Ci ( u utr (soil type)) u i α: depending on soil type S: topography factor V eff : explicit vegetation dependence for u>u tr (or u * >u *tr ) F~u 3 (or F~u *3 ) F: dust flux, u: surface wind speed, u tr : threshold wind speed, u * surface wind shear, s i : fraction of particles in size bin i, ρ= air density, g: gravitational constant 33 Usually areas with high vegetation and soil moisture are masked out Missing in Global Dust Models Crusting of soil surface (supresses dust emission) Global surface roughness (progress from remote sensing) Subgridscale variability of surface winds 34 Simulated Annual Dust Emission Fluxes Human Impact on Dust Emissions Estimates of dust fluxes from anthropogenically disturbed soils: IPCC, 2001: up to 50% (determines radiative forcing) Prospero et al, 2002: small (Natural sources dominant) Luo et al., 2003: 0-50% ( new desert source ) Tegen et al., 2004: <10% (Agricultural soils) Werner, in prep. 35 Large uncertainties! 36

10 Large Scale Transport Vertical Dust Distribution LITE data (spaceborne Lidar) To enable large-scale dust transport dust must be injected into higher atmospheric layers, e.g. East African Jet Modeling of dust transport by 3D tracer transport models or as tracer in General Circulation Models. Model results (Ginoux et al 2001) LITE data Model results (Ginoux et al 2001) Sahara China 37 Ginoux et al. (2001) 38 Deposition of Dust Aerosols Dry Deposition Sedimentation Turbulent mixing to the surface Wet deposition Resistance scheme Dry Deposition of Particles Aerodynamic resistance r a Airborne particles Quasi-laminar layer resistance r b Inverse settling velocity 1/v stokes Dry deposition: Deposition flux=v d (z) concentration(z) v d (z): deposition velocity at height z v d =1/r t (r t =total resistance, r t =r a +r b + ) 39 after Seinfeld (1997) Virtual resistance r a r b v s Deposited particles 40

11 Settling velocity v s η: kinematic viscosity r: particle radius ρ: particle density g: gravitational constant Aerodynamic resistance u * : wind stress κ: von Karman constant z: height z 0 : roughness length Friction force=gravitational force 3 6πηrv 4 s = πρgr 3 2 2ρgr vs = 9η 1 r a = ln κu* ( z) z z 0 (neutral conditions) Dry Deposition Velocity v d v 1 d = v r + r + r r v + a b a b s s Quasi-laminar resistance B: transfer coefficient 1 r b = Bu * Seinfeld (1997) (small for dust-sized particles) Wet Deposition Precipitation scavenging: in-cloud scavenging ( washout ) below-cloud scavenging ( rainout ) dm i = -fλ i PM i dt Seinfeld (1997) Mi : mass load of size fraction i f: areal fraction of precipitation P: precipitation rate Λ i : scavenging efficieny of fraction i -For dust particles, often only below-cloud scavenging is assumed -Usually using a fixed scavenging ratio based on measurements 43 Location Wet vs. dry deposition of mineral aerosols Bermuda Amsterdam Island Cape Ferrat, Mediterranean Enewetak Atoll, Samoa New Zealand North Pacific Summit Greenland Antarctica % Wet Deposition (observation) % Wet Deposition (model) Citation [Jickells et al., 1998]; Personal communication T. Church, 1999;[Kim and Church, 2001]. [Jickells and Spokes, 2001] [Guieu et al., 1997] [Arimoto et al., 1985, 1987] [Arimoto et al., 1990] [Uematsu et al., 1985] [Davidson et al., 1996] [Wolff et al., 1998] 44

12 Ratio Wet/Dry Deposition for Dust Result from a tracer transport model Possible Role of Air Pollution in Dust Wet Deposition Comparison of model results and station data: numbers: NA sites letters: Pacific sites hydrophobic hydrophilic aging dust Tegen et al., Fan et al., 2004 Deposition of Dust into Ocean Basins (Mt/year) North Pacific South Pacific North Atlantic South Atlantic Indian Ocean Global emission Duce et al., Prospero, Ginoux et al., Zender et al., Tegen et al., Luo et al., Atmospheric Lifetime of Dust Aerosol 10 µm particles: ca. 1 day 5 µm particles: ca. 3 days 1 µm particles: ca days 48

13 Seasonal Dust Aerosol Distribution: Model Results Previous Results of Dust Models Aerosol optical thickness, NH spring Tegen et al, 1996 Mahowald et al., 1999 TOMS Aerosol Index Layer 1 Dust Mixing Ratio (µg/kg) Tegen et al. (2002) 49 AOT/TOMS AI 50 Dust Deposition Flux Model (Tegen et al. (2002)) Observations (Sediment trap data) Some Estimates for Aeolian Iron Input into the Global Oceans Duce and Tindale (1991): Mineral dust input into global ocean: 900 Mt/yr Iron input (assuming 3.5% Fe content (crustal)): 30 Mt/yr Dissolved Fe (assuming 10% solubility): 3 Mt/yr Gao et al. (2001): Iron input (3.5% Fe) 14 Mt/yr Fung et al. (2000): Iron input (1.2% Fe in quartz, 5% in clay particles) 7 Mt/yr Dust Flux (g/m 2 /yr) 51 52

14 Estimate of Oceanic Iron Supply from Airborne Dust vs. Upwelled Iron Estimate of Global Mineral Distribution Can we obtain a better estimate of the iron content of dust? Global mineral distribution is not well defined, Fe content varies mole Fe/m 2 /yr Fung et al. (2000) 53 Claquin et al. (1999) 54 Iron Solubility in Dust Iron in soil dust mostly present as Fe(III) not easily soluble in natural waters. Spokes et al. (1994): ca. 1% soluble (cloud processed dust). Photoreduction during transport reduces part of Fe(III) to easily soluble Fe(II). Zhuang et al. (1992) suggest that photoreduction during long-range transport increases percentage of Fe(II) from less than 1% to more than 50% for dust samples from the North Pacific. Zhu et al. (1997): 6% (3-13%) of total iron in dust (North Atlantic samples) is soluble, controlling factor is solubility of Fe(III), rather than Fe(II) content. Measurements of Iron Solubility in Dust Models assume usually 1-10% iron solubility one of the largest uncertainty factors when estimating iron deposition into oceans 55 Mahowald et al,

15 Glacial-Interglacial Changes in Dust Deposition g/m 2 /y Loess (China) Pacific Ocean Role of dust may have been different for different climates! LGM Dust (ECHAM5) Approx. 2-fold increase of global dust emissions Increase of (i) Saharan dust emissions by wind speed changes, (ii) Asian dust emissions by source region expansion LGM mean dust deposition [g/m 2 /yr] modern mean dust deposition [g/m 2 /yr] LGM dust deposition fluxes: simulation versus observations [g/m 2 /yr] Vostok, Antarctica Figure courtesy of K. E. Kohfeld Dust Conc (ppm) Ding et al., 1994; Hovan et al., 1991; Petit et al. 1990,2001; Kohfeld and Harrison, M. Werner Projected Future Dust Emission Changes Using Global Dust Models Future changes in dust emissions can depend on: Changes in meteorology Computing dust emission using meteorological fields extracted from IPCC future scenarios. Changes in vegetation cover (as consequence of climate change) Vegetation changes computed with vegetation model (e.g. BIOME3/BIOME4). Changes in cultivation patterns Changes in emissions from cultivated regions computed using results Regional Dust Emission Changes: Projection for 2050 Tegen et al., 2004: HADCM, ECHAM Mahowald and Luo, 2003: NCAR Global: % change in emissions Future estimates in dust changes vary greatly due to uncertainties in climatology and parameterization of dust emissions in global models based on IPCC SRES scenarios, e.g. from IMAGE2.2 (RIVM)

16 Summary Aeolian dust is an important source of iron in remote ocean regions, this dust input is highly variable in space and time The lack of appropriate observational data requires the use of global models to estimate the dust deposition at global scales Solubility of iron is one of the biggest uncertainty factors in estimates of the addition of iron into the oceans 61