The Two Signatures of Chinese Dust Storms in Beijing

The Two Signatures of Chinese Dust Storms in Beijing

Three major dust sources + Loess Plateau

Northwesterndeserts

Northern high- dust deserts

Northern low- dust deserts

Loess area

(Sun, J. 2002)

Takla Makan Gobi

Sandy desert with abundant salts (chlorides, sulfates, carbonates).

Rocky desert with less salts.

Main transport pathways

(Makra L. et al, 2002)

The 6 dust storms and their 3 types

0

2000

4000

6000

8000

10000

120003-

2-01

3-3-

01

3-9-

01

3-15

-01

3-22

-01

3-30

-01

3-19

-02

3-20

-02

3-22

-02

3-25

-02

4-1-

02

4-4-

02

4-7-

02

4-8-

02

4-10

-02

4-12

-02

4-14

-02

4-20

-02

4-23

-02

4-27

-02

TSP

DS1DS5

DS4

DS3

DS2 DS6

950 µg m-3

Dust episodes defined by meteorological reports, air pollution indices, and our TSP data.

II

IIII

III III

I

Site: BNU

Ca vs. Al

0.1

1

10

100

1000

0.1 1 10 100 1000Al (ug/m3)

Ca

(ug/

m3)

DS Iaf ter DS IDS IIaf ter DS IIDS IIINDS

Ca/Al=1.2I

Ca/Al=0.39II

Ca/Al=0.23*

S vs. Al

0.01

0.1

1

10

100

0.1 1 10 100 1000Al (ug/m3)

S (u

g/m

3)

DS Iafter DS IDS IIaf ter DS IIDS IIINDS

S/Al=0.072I

S/Al=0.030II

S/Al=0.013**

Co vs. Al

0.0001

0.001

0.01

0.1

1

0.1 1 10 100 1000Al (ug/m3)

Co

(ug/

m3)

DS Iaf ter DS IDS IIaf ter DS IIDS IIINDS

Co/Al=0.00041I

Co/Al=0.00025II

Co/Al=0.00019

Sr vs. Al

0.001

0.01

0.1

1

10

0.1 1 10 100 1000Al (ug/m3)

Sr (u

g/m

3)DS Iaf ter DS IDS IIaf ter DS IIDS IIINDS

Sr/Al=0.0030II

Sr/Al=0.0025*

Sr/Al=0.0042I

Sc vs. Al

0.0001

0.001

0.01

0.1

1

0.1 1 10 100 1000Al (ug/m3)

Sc(u

g/m

3)

DS Iaf ter DS IDS IIaf ter DS IIDS IIINDS

Sc/Al=0.00016*Sc/Al=0.00015I

Sc/Al=0.00012II

Ca vs. S

0.1

1

10

100

1000

0.1 1 10 100 1000S(ug/m3)

Ca

(ug/

m3)

DS Iaf ter DS IDS IIaf ter DS IIDS IIINDS

Ca/S=16I

Ca/S=15II

Ca/S=17*

Fig.1. Back trajectories for air parcels that arrived at Beijing during DS3 (left) and DS4 (right) caculated from NOAA (http://www.noaa.gov).

II IDS3 DS4

Back trajectories for the two types of episodes

Elements that don’t work

Fe, Ti, V, Mg, Na, Zn, Cu

Fe vs. Al

0.1

1

10

100

1000

0.1 1 10 100 1000Al (ug/m3)

Fe (u

g/m

3)DS Iafter DS IDS IIaf ter DS IIDS IIINDS

Fe/Al=0.045*Fe/Al=0.055II

Fe/Al=0.062I

Ti vs. Al

0.01

0.1

1

10

100

0.1 1 10 100 1000Al (ug/m3)

Ti (u

g/m

3)

DS Iaf ter DS IDS IIaf ter DS IIDS IIINDS

Ti/Al=0.057*Ti/Al=0.060II

Ti/Al=0.067I

V vs. Al

0.001

0.01

0.1

1

10

0.1 1 10 100 1000Al (ug/m3)

V (u

g/m

3)

DS Iaf ter DS IDS IIaf ter DS IIDS IIINDS

V/Al=0.0012*V/Al=0.0016II.I

Mg vs. Al

0.1

1

10

100

1000

0.1 1 10 100 1000Al (ug/m3)

Mg

(ug/

m3)

DS Iaf ter DS IDS IIaf ter DS IIDS IIINDS

Mg/Al=0.12*Mg/Al=0.21II

Mg/Al=0.23I

Mn vs. Al

0.001

0.01

0.1

1

10

0.1 1 10 100 1000Al (ug/m3)

Mn

(ug/

m3)

DS Iaf ter DS IDS IIaf ter DS IIDS IIINDS

Mn/Al=0.0088*Mn/Al=0.0090II.I

Na vs. Al

0.1

1

10

100

1000

0.1 1 10 100 1000Al (ug/m3)

Na

(ug/

m3)

DS Iaf ter DS IDS IIaf ter DS IIDS IIINDS

Na/Al=0.15*

Na/Al=0.21I

Na/Al=0.25II

0.001

0.01

0.1

1

10

0.1 1 10 100 1000Al (ug/m3)

Zn (u

g/m

3)DS Iaf ter DS IDS IIaf ter DS IIDS IIINDS

Zn/Al=0.0020I,II

Zn/Al=0.0012*

Zn vs. Al

Cu vs. Al

0.001

0.01

0.1

1

10

0.1 1 10 100 1000Al (ug/m3)

Cu

(ug/

m3)

DS Iaf ter DS IDS IIaf ter DS IIDS IIINDS

Cu/Al=0.00034*

Cu/Al=0.00070I

Ratio DS I/DS II

Ca/Al 3.2±0.3

S/Al 2.4±0.3

Co/Al 1.61±0.11

Sr/Al 1.37±0.07

Sc/Al 1.30±0.04

Fe/Al 1.14±0.06

Ti/Al 1.12±0.07

V/Al 1.08±0.09

Mg/Al 1.07±0.07

Mn/Al 1.01±0.05

Na/Al 0.82±0.13

Cl-/Al 0.60±0.24

Ratio DS I vs DS II DS II vs NDSCa/Al 0.000 0.000S/Al 0.000 0.000Co/Al 0.001 0.000Sr/Al 0.001 0.000Sc/Al 0.000 0.001

Fe/Al 0.066 0.032Ti/Al 0.124 0.182V/Al 0.376 0.020Mg/Al 0.355 0.000Mn/Al 0.830 0.000

T-tests (two-tailed)

Why these six elements?

Maybe just ionic substitution for Ca++ in the gypsum matrix

Principles of ionic substitution

• Free substitution when:– ionic radii are within about 15%– charges differ by zero or one unit.

• The next slide shows the ions that can freely substitute for Ca in gypsum.

• They are Na (which will go to halite instead), Sr, the REE (not measured here), and Sc.

• Except for Co, these cations match the observed cationic enrichments in the high-Ca signal.

Ionic radii main

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

0 1 2 3 4 5 6 7

Charge

Ioni

c ra

dius

, A

Na

Li

KRb

Cs

BaPbSrCaMnZn

CuNiMg

Be

La

REEScInFeGa

B

Al

U

ZrHfTi

P

V

C

GeSeMn

Si

NbTa

S

Co

Previous attempts to find chemical tracers

• Yang D. et al. (1997) In Chinese– Concentrations of 12 elements from five sites

between Beijing and source areas, but no explicit characterization of sources.

• Zhang X. et al. (1996, 1997)– Four-element tracer system (Al, Fe, Mg, Sc) that

reportedly differentiated three desert sources in China (NW, N, NE). S not measured, and Ca eliminated because of “postdepositional effects.”

Previous attempts, cont’d

• Zhang X. et al. (2003)– NW deserts have 50% higher Ca, Fe, K, and Mg, and

20% lower Si, Al, Mn, and Ti than N deserts.• Nakano et al. (2004)

– Sr/Nd isotopic data from soils used to claim that contemporary dustfall on Beijing comes more from nearby soils (<200 km) than from deserts or Loess Plateau.

– Direct dustfall not measured.• Xuan J. (2005)

– Emissions of Fe, Al, K, Mg, Mn, Na, Ca, and Ti in dust calculated from surface soils of six sources (three in Mongolia and three in China). X/Al ratios 14%–36% higher from NW deserts than from N deserts, with Mn and Ca the highest and Fe and Ti the lowest.

Summary• The main tracer system previously suggested (Fe/Al,

Mg/Al, Sc/Al) does not work after transport to Beijing.• Instead, a five-ratio system (Ca/Al, S/Al, Sr/Al, Co/Al,

Sc/Al) distinguishes the Takla Makan from the Gobi.• The enriched Ca and S of the upper line agree with

the enrichment of gypsum and other salts in the Takla Makan (Zhang et al 2003; Makra et al., 2002; Okada 2004; Chinese Soil Atlas, 1994).– Rivers from surrounding mountains drain into the Tarim

Basin and dry up there, depositing their salts (NaCl, CaCO3, CaSO4, etc.).

• The co-enrichments of Sr, Co, and Sc are consistent with ionic substitution for Ca in gypsum of the Takla Makan.

• These results also provide a way to deal with the high Ca from Beijing.– The two sources can be differentiated by

S/Al or Ca/S.– Dust storms from Xinjiang replace the high

Ca of Beijing (construction activities?) with high Ca from the Takla Makan Desert.

– Thus the aerosol really does change when a dust storm arrives.

The straightness of the 1:1 lines over an episode

• The straightness of the 1:1 lines over two orders of magnitude of concentration means that the dust storms remain a single material throughout the episode. This in turn implies at least two things:– (1) No significant SO2 is converted to sulfate on the

surface of the dust as it enters Beijing. This agrees with the findings of Zhang D. et al. for Qingdao (AE, 2003) and Song et al. for ACE-Asia (AE, 2005).

– (2) The signal from the dust overwhelms the signal of local Beijing dust, even down to very low concentrations.

• This is consistent with the “clear-out” phase in dust storms (Guo et al, 2004), which can be stronger and last longer than first thought.

• It further means that the wind speeds during dust storms in Beijing, much lower than those at the source, are too low to resuspend much dust in Beijing.

• This is particularly so when the first stage of a dust storm is falling dust, which usually has very low wind speeds because it arrives ahead of the cold front.

Some thoughts about the future

Possible areas of research• Verifying the two signatures with samples

from other times and places.• Searching for other signatures, say from the

northern low-dust area or from Kazakhstan (northwest of the Takla Makan).

• Explaining the two signatures.• Tracking dust clouds by means of their

signatures.• Using the signatures to interpret existing

sets of data.

Explaining the two signatures

• Dust storms may come from small “hot spots” in deserts rather than from evenly over the entire surface.

• In Africa (the Sahara), these hot spots are “wadis,” or ephemeral dried river channels and lake bottoms.

• Does the same thing hold for Chinese deserts?• Prospero reports that the early stages of dust

storms from the Gobi are composed of narrow plumes from hot spots that later diffuse into a broad dust cloud. (As seen from satellite photos).

Identifying the Chinese hot spots

• Examine satellite photos from early stages of dust storms in the Gobi and the Takla Makan.– See whether hot spots can be verified.– If so, see whether they occupy consistent

locations.– If so, see whether those locations can be

identified on maps.– If so, are they wadis or something else?

Determining properties of hot spots

• Go to the hot spots and sample their soil.• Compare the hot spots with each other and with

the surrounding soil (and with the average desert soil).– Average desert soil may be available from the literature.– Data on hot spots may also be available somewhere,

but I am guessing not.• May have to take multiple samples along a dried

river bed—don’t know how many yet.• Also don’t know how many hot spots have to be

sampled.

Determining properties of aerosol blown out of hot spots

• Don’t yet know how best to do it.• Sample at the very beginning of dust

storm?• Create aerosol from the soil in a wind

tunnel?• Create the aerosol right there in the desert?

(Jinghua’s idea)

The possible stages of fractionation

• Hot spots vs. deserts.• One hot spot vs. another.

– Salt in soils promotes lifting.• Aerosol vs. soil in hot spots.• Aerosol vs. aerosol-sized soil in hot spots.

– Some minerals may be lifted more effectively than other minerals.

• Depletion of coarse aerosol during transport.

The End