Duncan Fairlie ^ *

^NASA Langley Research Center*Dept. Earth and Planetary Sciences, Harvard University

A global mineral dust simulation for 2001: impact of overseas vs. domestic sources on

concentrations over North America

Duncan Fairlie^*

April 2005April 2005

Mineral Dust Module in GEOS-CHEM

• Dust mobilization (emission): •(1) Ginoux et al. (2001) for GOCART model •(2) Zender et al. (2003) for DEAD model

•Dry deposition – grav. settling, turb. transfer, impact., intercep. (Zhang et al. 2001); Vd=Vd(D,,sfc).

•Wet deposition – aerosol scavenging in convective updrafts, first-order rainout and washout from anvils and stratiform precip., plus cirrus precip, reevap. (Liu et al., 2001).

•Size bins: 0.1-1.0, 1.0-1.8, 1.8-3.0, 3.0-6.0 m radius (following Ginoux et al., 2001)

Mobilization• Generally, Fd = C S Qs

Fd – vertical dust fluxQs – horizontal saltation fluxS – source function (defines potential dust source regions,

and comprises surface factors, e.g. vegetation, snow cover, and an efficiency factor e.g. topographic anomaly)

• GOCART: Qs ~ U102 (U10 – U*t)

• DEAD: Qs ~ U*3 (1-U*t/U*) (1+U*t/U*)2

U*t – threshold friction velocity (particle size, density; air density, viscosity). U*t modulated by surface moisture.

Essential difference between Ginoux (GOCART) and Zender (DEAD) schemes

Do seasonal vegetation change and human activities crucially affect dust sources?

NO (Ginoux et al.): Source areas prescribed. Focus on topographic lows in desert and semi-desert regions: ephemeral lakes, dry river beds.

Yes (Zender et al.): Source areas not predetermined, but largely controlled by changing LAI. Focus on upstream runoff area.

[Snow cover also mitigates mobilization in DEAD.]

GOCART and DEAD source potentials October 2001

GOCARTGOCARTsourcesource

DEADDEADsourcesource

Veg.(LAI)Veg.(LAI)

Effic.Effic.

2.02.0

2.02.0

0.020.02

0.020.020.020.02

0.020.02

2.02.0

2.02.0 4.04.0

1.41.4

PM2.5 Columns (mg/m2), Apr/Oct 2001DEADDEAD GOCARTGOCART

Burdens: April OctoberBurdens: April OctoberDEAD 33Tg (fine: 12Tg); 17Tg (fine: 6Tg)DEAD 33Tg (fine: 12Tg); 17Tg (fine: 6Tg)GOCART 46Tg (fine: 17Tg); 35Tg (fine: 12Tg)GOCART 46Tg (fine: 17Tg); 35Tg (fine: 12Tg)

AprilApril

OctoberOctober

AprilApril

OctoberOctober

10001000

10001000

10001000

10001000

Key Questions

• What is the impact of overseas (in particular transpacific) transport of dust on aerosol concentrations over North America?

• What are the key factors that control dust mobilization from the Earth’s surface?

IMPROVE site locations

IMPROVE aerosol network: focuses on visibility in Fed. 1 sites data for aerosol sulfate, nitrate, ammonium, elemental carbon (EC), organic carbon (OC), fine and coarse dust, sea salt.•Fine dust (PM2.5) = 2.2*Al+2.49*Si+1.63*Ca+2.42*Fe+1.94*Ti (Malm et al., 1994)

PM2.5 seasonal-average surface dust concentrations (g/m3), 2001

DJFDJF

MAMMAM

JJAJJA

SONSON

IMPROVEIMPROVE GOCARTGOCART DEADDEAD

GOCART: dust sources in SW CONUS too strong and sustainedGOCART: dust sources in SW CONUS too strong and sustainedDEAD: spurious sources in N. plains corrupt eastern sitesDEAD: spurious sources in N. plains corrupt eastern sites

g/m3g/m30 2 40 2 40 2 40 2 4 0 2 40 2 4

PersistentPersistentHigh biasHigh bias

SpuriousSpuriousemissionsemissions

Remedies?

DEAD scheme: Restrict emissions to desert and DEAD scheme: Restrict emissions to desert and

semi-desert regions over CONUSsemi-desert regions over CONUSGOCART scheme: raise threshold velocities or GOCART scheme: raise threshold velocities or

otherwise scale back emissionsotherwise scale back emissions

Choose DEAD formulation with latest GOCART Choose DEAD formulation with latest GOCART source fn.source fn.

0 2 40 2 40 2 40 2 4 0 2 40 2 4

PM2.5 seas.-ave. (g/m3), 2001DEAD with GOC. source

DJFDJF

MAMMAM

JJAJJA

SONSON

IMPROVEIMPROVE ModelModel OverseasOverseas

Use of GOC. source eleviates issues with GOCART and DEAD schemes.Use of GOC. source eleviates issues with GOCART and DEAD schemes.But, low bias at eastern sites in JJA, SON, and high bias NW in MAM.But, low bias at eastern sites in JJA, SON, and high bias NW in MAM.

High biasHigh bias

Low biasLow bias

Model vs. IMPROVE PM2.5 by season (g/m3)

DJFDJF

MAMMAM

JJAJJA

SONSON

GOCsrceGOCsrce+ west+ westo easto east

Use of DEAD with GOC. source. improves in West, eliminates Use of DEAD with GOC. source. improves in West, eliminates spurious high values in East, but leaves East biased low.spurious high values in East, but leaves East biased low.

WestWest EastEast

IMPROVEIMPROVE

ModelModel

1010

1010

101010100.10.1pdfspdfs IMPROVEIMPROVE

ModelModel

Hells Canyon, ORHells Canyon, OR Mount Hood, ORMount Hood, OR

Mount Rainier, WAMount Rainier, WA

Model pulses in early May Model pulses in early May in NW overestimate observed in NW overestimate observed values.values.

What’s happening in April, May?What’s happening in April, May?

IMPROVEIMPROVEModelModel

1010

1010 1010

(g/m3)

Total dust conc. vs. U. Miami climatologies

Asia/Asia/N. PacificN. Pacific

E. AtlanticE. Atlantic

W. AtlanticW. Atlantic

Reproduces seasonal cycle over N. Pacific. GOC. source raises Reproduces seasonal cycle over N. Pacific. GOC. source raises ––ve bias over DEAD run. ve bias over DEAD run.

GOCsrceGOCsrce

DEADDEAD

MidwayMidwayHedoHedo

OahuOahu

SalSal

BarbadosBarbados BermudaBermudaMiamiMiami

IzanaIzana MaceMace

(g/m3)

Comparison with U. Miami climatologies

MediansMediansU. MiamiU. Miami

MediansMedians

RatioRatioDEADDEAD

RatioRatioGOCsrceGOCsrce

Some improvements with GOCsrce in SH; -ve bias reducedSome improvements with GOCsrce in SH; -ve bias reduced

(g/m3)

Summary(1) Choice of 2 mobilization schemes in GEOS-CHEM. GOCART scheme

generates about twice as much dust as DEAD with current parameters. Both have issues over North America.

(2) Use of LAI as a sole vegetative constraint on emission appears problematic (DEAD scheme). No account for senescent vegetation nor land management practices. Recommend use of DEAD with latest GOCART source potential.

(3) All solutions capture seasonal cycle over Northern Pacific in 2001, with background levels consistent with U. Miami observations.

(4) Comparisons with TRACE-P and ACE-Asia bulk aerosol show good agreement, 30% negative bias, respectively.

(5) Results suggest that transpacific transport may contribute between 0.3 and 0.6 g/m3 (seasonal mean) at surface sites in NW states for JJA and SON [c.f. EPA default estimate of fine dust nationwide of 0.5 g/m3.] However, indication that overseas contribution at NW sites are biased high. Also, other years need attention.

Extras

2001 Annual Budget

Model Emis

Tg/yr

Wet Dep. Tg/yr

Dry Dep. Tg/yr

Ave Load

Tg

Life

days

G-C (GOC.) 2568 1102 1462 35 5.0

G-C (DEAD) 1305 541 754 16.5 4.6

G-C GOCsrce 1460 601 840 18.4 4.7

GOCART (G.et al, ’01)

1814 235 1606 36 7.1

DEAD

(Z.et al, ’03)

1490 607 866 17.4 4.3

In-situ bulk aerosol measurements made by UNH (Talbot, Dibb, et al.) during TRACE-P. Measurements made by U. Hawaii (TAS) (Huebert et al.) during ACE-Asia. Use Ca2+ and Na+. Account for sea salt-Ca2+ contribution (ss-Ca2+ = 0.0439 ss-Na+ neq., Wilson, 1975). Assume Ca2+ = 6.8% dust by mass (Wang, 1999; Song and Carmichael, 2001), following Jordan et al. [2004].

Asian Rim DC8 flights for TRACE-P, March 2001 [www-gte.larc.nasa.gov]

C130 flights for ACE-Asia, April 2001 [Huebert et al.]

Comparison with bulk aerosol dust from TRACE-P (Dibb et al.) March, 2001

Use of GOCsrce raises negative bias experienced with DEADUse of GOCsrce raises negative bias experienced with DEAD

Comparison with bulk aerosol dust from ACE-Asia (TAS, Huebert) April, 2001.

GOCsrce shows a more unimodal distribution, similar to TASGOCsrce shows a more unimodal distribution, similar to TASbut is biased low.but is biased low.

PM2.5 monthly average-average (g/m3) IMPROVEIMPROVE GOCsrceGOCsrce OverseasOverseas

MarchMarch

AprilApril

MayMay

JuneJune

Springtime anomaly in NW between model and IMPROVE Springtime anomaly in NW between model and IMPROVE concentrated in May.concentrated in May.

GOCsrce emissionsJanJan AprApr

JulJul OctOct

DEAD formulation provides a consistently defined thresholdDEAD formulation provides a consistently defined threshold

GOCsrce PM2.5 columnsJanJan AprApr

JulJul OctOct

GOCsrce drydepJanJan AprApr

JulJul OctOct

GOCsrce wetdepJanJan AprApr

JulJul OctOct

Mobilization: DEADDust mobilisation directly related to horizontal

saltation flux: Fd = Am S Qs.

Qs = cs /g U*3 (1-U*t/U*) (1+U*t/U*)2

• U*t = U*t MB (particle size, density; air density, viscosity). U* computed for D=75 um.

• U* friction velocity (roughness, z0=1.0e-4 m)

• surface wetness modulates U*t (Fecan et al., 1999)

• S - efficiency factor

- Am =(1-Al-Aw)(1-As)(1-Av)

- VAI < 0.3 time-varying

- snow cover < 5 cm