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

Duncan Fairlie^{^}*

^NASA Langley Research Center

*Dept. Earth and Planetary Sciences, Harvard University

April 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,ρ,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 μm radius

(following Ginoux et al., 2001)

​

Mobilization

• Generally, Fd = C S α Qs

Fd – vertical dust flux

Qs – horizontal saltation flux

S – 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

GOCART

source

DEAD

source

Veg.(LAI)

Effic.

2.0

2.0

0.02

0.02

0.02

0.02

2.0

2.0

4.0

1.4

PM2.5 Columns (mg/m²), Apr/Oct 2001

DEAD

GOCART

Burdens: April October

DEAD 33Tg (fine: 12Tg); 17Tg (fine: 6Tg)

GOCART 46Tg (fine: 17Tg); 35Tg (fine: 12Tg)

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 (μg/m³), 2001

DJF

MAM

JJA

SON

IMPROVE

GOCART

DEAD

GOCART: dust sources in SW CONUS too strong and sustained

DEAD: spurious sources in N. plains corrupt eastern sites

μg/m3

0 2 4

0 2 4

0 2 4

Persistent

High bias

Spurious

emissions

Remedies?

DEAD scheme: Restrict emissions to desert and semi-desert regions over CONUS

GOCART scheme: raise threshold velocities or otherwise scale back emissions

​

Choose DEAD formulation with latest GOCART source fn.

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

DJF

MAM

JJA

SON

IMPROVE

Model

Overseas

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.

High bias

Low bias

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

DJF

MAM

JJA

SON

GOCsrce

+ west

o east

Use of DEAD with GOC. source. improves in West, eliminates

spurious high values in East, but leaves East biased low.

West

East

IMPROVE

Model

10

10

10

10

0.1

pdfs

IMPROVE

Model

Hells Canyon, OR

Mount Hood, OR

Mount Rainier, WA

Model pulses in early May

in NW overestimate observed

values.

What’s happening in April, May?

IMPROVE

Model

10

10

10

(μg/m3)

Total dust conc. vs. U. Miami climatologies

Asia/

N. Pacific

E. Atlantic

W. Atlantic

Reproduces seasonal cycle over N. Pacific. GOC. source raises

–ve bias over DEAD run.

GOCsrce

DEAD

Midway

Hedo

Oahu

Sal

Barbados

Bermuda

Miami

Izana

Mace

(μg/m3)

Comparison with U. Miami climatologies

Medians

U. Miami

Medians

Ratio

DEAD

Ratio

GOCsrce

Some improvements with GOCsrce in SH; -ve bias reduced

(μg/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.

​

• 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.

​

• 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 μg/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 μg/m³.] However, indication that overseas contribution at NW sites are biased high. Also, other years need attention.

Extras

2001 Annual Budget

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 DEAD

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

GOCsrce shows a more unimodal distribution, similar to TAS

but is biased low.

PM2.5 monthly average-average (μg/m³)

IMPROVE

GOCsrce

Overseas

March

April

May

June

Springtime anomaly in NW between model and IMPROVE

concentrated in May.

GOCsrce emissions

Jan

Apr

Jul

Oct

DEAD formulation provides a consistently defined threshold

GOCsrce PM2.5 columns

Jan

Apr

Jul

Oct

GOCsrce drydep

Jan

Apr

Jul

Oct

GOCsrce wetdep

Jan

Apr

Jul

Oct

Mobilization: DEAD

Dust mobilisation directly related to horizontal saltation flux: F_d = A_m S α Q_s.

Q_s = c_s ρ/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

- A_m =(1-A_l-A_w)(1-A_s)(1-A_v)

- VAI < 0.3 time-varying

- snow cover < 5 cm