Nuclear Instruments and Methods in Physics Research B 273 (2012) 186–188Contents lists available at ScienceDirect
Nuclear Instruments and Methods in Physics Research B
journal homepage: www.elsevier .com/locate /n imbA new approach to the combination of IBA techniques and wind back trajectory
data to determine source contributions to long range transport of fine particle
air pollution
David D. Cohen ⇑, Jagoda Crawford, Eduard Stelcer, Armand Atanacio
Australian Nuclear Science and Technology Organisation, Locked Bag 2001, Kirrawee DC, NSW 2232, Australia
a r t i c l e i n f o a b s t r a c tArticle history:
Available online 26 July 2011
Keywords:
IBA techniques
PIXE
Fine particles
Long range transport⇑ Corresponding author.
E-mail address: dcz@ansto.gov.au (D.D. Cohen).
0168-583X/$ - see front matter Crown Copyright  2
doi:10.1016/j.nimb.2011.07.071A new approach to link HYSPLIT back trajectories to the source of fine particle pollution as characterised
by standard IBA techniques is discussed. The example of the long range transport of desert dust from
inland Australia across the eastern coast is used to show that over a 10-year period extreme soil events
originated from major agricultural regions some 30% of the time and that dust from known deserts are
not always the problem.
Crown Copyright  2011 Published by Elsevier B.V. All rights reserved.1. Introduction
The simultaneous application of PIXE, PIGE, RBS and PESA tech-
niques for multi-elemental characterisation of fine particle air pol-
lution is now well established [1–3]. These IBA techniques provide
elemental concentrations for H to Pb with sensitivities down to
1 ng/m3 of air sampled on standard air particulate filters. This en-
ables the determination of source fingerprints containing up to 20
different elemental species as well as estimates of their contribu-
tions to the total gravimetric mass on the filters [3]. Recently this
work has been extended to include wind speed and direction data
enabling source locations to be accurately determined and long
range transport of air pollution to be followed over several years
[3,4].
This paper discusses a new approach of combining IBA data
with long term hourly wind back trajectory data from the receptor
site to identify real source regions hundreds of kilometres away.
Unlike previous approaches this method links the trajectories
using HYSPLIT [5] and the source fingerprints [4,6] to source re-
gions by counting up the number of back trajectories that originate
from the sampling site and also intersect these source regions. This
approach enables quantitative estimates, over several years, of the
contributions of these source regions to the fingerprints measured
by IBA techniques at the receptor or sampling site. To demonstrate
the application of this new method we discuss below in detail the
long range transport of dust from Australian deserts over hundreds011 Published by Elsevier B.V. Allof kilometres across the east coast of Australia between 2001 and
2010.2. Method
The four simultaneous IBA techniques of PIXE, PIGE, RBS and
PESA were used to obtain elemental concentrations (ng/cm2) for
the 24 elements: H, C, N, O, F, Na, Al, Si, P, S, Cl, K, Ca, Ti, V, Cr,
Mn, Fe, Co, Ni, Cu, Zn, Br and Pb on stretched Teflon filers. These
techniques have been described extensively in detail previously
[1,2 and references therein]. In summary, 2.6 MeV proton beams
of around 10 nA for a total charge collection of 3 lC and beam
diameters of 8 mmwere used to irradiated each Teflon filter which
was exposed for 24 h at a flow rate of 22 l/min during the study
period. PIXE provided information on elements from Al to Pb, PIGE
on F, Na and Al, RBS on C, N and O and PESA on the H concentra-
tions. Typical spectra for such IBA analyses have been published
(Ref. [1]).
In particular herewe used five of these elements, Al, Si, Ca, Ti and
Fe, todefine thefine soil content in the standardway [7]by assuming
they occurred in their oxide forms and summing the oxide contribu-
tions for thesefive elements. That is, the fine soil componentwasde-
fined as, [Soil] = 2.20[Al] + 2.49[Si] + 1.63[Ca] + 1.94[Ti] + 2.42[Fe],
where the terms in square brackets represent the elemental concen-
trations in (lg/m3) of air sampled and the constants associatedwith
each elemental reflect their common oxide forms.
Filters containing fine PM2.5 particles were collected every Sun-
day and Wednesday between 1 January 2001 and 31 December
2010 at a rural/urban site at Muswellbrook at 32.26S, 150.89E
some 250 km NW of Sydney, Australia. Fig. 1 shows this fine soilrights reserved.
D.D. Cohen et al. / Nuclear Instruments and Methods in Physics Research B 273 (2012) 186–188 187
7 Day Back Trajectories Muswellbrook 300m, 500m
-10
Darwin
-15
Broome
12 14
-20
7 Heron Is.
-25 11 10 13
6 Brisbane
-30 38 9 5 4 2
Perth
-35 1
15 Sydney
23-Nov-08 Cities Adelaide
-40 Sites >5,000
>100 >5 Hobart-45
-50
Longitude
Fig. 2. A plot of the number of 7-day back trajectory intersections per 1  1 cell
originating from the Muswellbrook sampling site on the Australian east coast for
every hour of every sampling day between January 2001 and October 2010. Greater
than 5000 intersections per cell are (), between 5000 and 100 intersections (s)
and between 100 and 5 intersections are (+). Also shown are 24 typical hourly back
trajectories for the 23 November 2008 a high soil day at Muswellbrook site.
Latitude
110
115
120
125
130
135
140
145
150
155
160concentration measured at the Muswellbrook site between 2001
and 2010. The average soil concentration was 0.37 ± 3.2 lg/m3,
being on average about 9% of the total fine mass. Sampling days
with soil concentrations above 1.3 lg/m3 were defined as extreme
events (see horizontal line in Fig. 1). There were 43 days from 961
sampling days between January 2001 and December 2010 with soil
concentrations greater than 1.3 lg/m3. The question now posed is
what was the origin of the soil associated with these extreme
events?
To answer this question we used the HYSPLIT program [5] to
plot hourly back trajectories for each of the extreme days. Fig. 2
contains such a collection of 24-hourly 7-day back trajectories
for the 23 November 2008 extreme event shown in Fig. 1. It shows
back trajectories leaving the Muswellbrook sampling site at
heights of 300 and 500 m above ground level and travelling to
the west and south and out into the southern ocean off Australia.
The 15 rectangular boxes drawn in Fig. 2 show the known desert
areas and schematically represent their extentwithin the Australian
continent. The trajectories for 23 November 2008 event intersect
with boxes 1 and 15 but not many others. As this approach demon-
strates it is a tedious and somewhat messy process to perform daily
back trajectories for all the extreme event days between 2001 and
2010. To improve the process we have developed a FORTRAN code
to plot all hourly back trajectories over 10 years and their intersec-
tions with all the 15 boxes drawn in Fig. 2. To test that these trajec-
tories cover all boxes and our region of interest we plot the number
of such intersections for every 1  1 cell in the region of interest,
which is shown in Fig. 2 for 7-day back trajectories. Clearly these
cover the region of interest well with at least more than 5 intersec-
tions per cell and cells within a 500-km radius of the sampling site
having more than 5000 intersections per cell. The longitude and
atitude range covers some 4000 individual 1  1 cells between
110–160 longitudeand50 to10 latitudewithanaveragenum-
ber of non-zero intersections per cell of 1606 and a median value of
853 intersections per cell. In most cases the desert regions depicted
in Fig. 2 aremore than ± 1 or ± 100 km across. The size of the desert
boxes and the time along back trajectories within HYSPLIT is most
important and is related to the typical wind speeds along a trajec-
tory. A detailed examination of the HYSPLIT data showed that there
were no wind speeds in excess of 100 km/h. Hence hourly points
along each 7-day back trajectory are more than sufficient and no
hourly points will miss any desert boxes with dimensions greater
than ± 100 km from the box centre or 200 km across.10
Muswellbrook Soil 2001-2010
9
8
7
6 23-Nov-08
5
4
3
2
1
0
Fig. 1. Plot of the PM2.5 soil concentrations at the Muswellbrook site between
January 2001 and December 2010.
Soil (µg/m3)
3/01/2001
3/07/2001
3/01/2002
3/07/2002
3/01/2003
3/07/2003
3/01/2004
3/07/2004
3/01/2005
3/07/2005
3/01/2006
3/07/2006
3/01/2007
3/07/2007
3/01/2008
3/07/2008
3/01/2009
3/07/2009
3/01/2010
3/07/20103. Results
Every time one of these back trajectories traverses a desert box
we place a dot in that box as shown in Fig. 3. The number of dots
per box is then a quantitative measure of the relative impact that
each of these desert regions has on the sampling site hundreds of
kilometres away.
Table 1 summarises the number of intersections per desert re-
gion (boxes 1–15) for all 43 extreme event days with
soil > 1.3 lg/m3. More than 65% of all extreme events came from
only two regions, namely, Lake Mungo (1) and the Riverina (15).
Other major desert dust sources had relatively minor impact on
the Muswellbrook site. The Riverina area is not a desert area it is
actually a major agricultural area used to grow crops to feed the
Sydney region. This is a significant finding not previously reported.-10
Darwin
-15
Broome 12 14
-20
7
11 Heron Is
-25 10 139
8 6 Brisbane
-30 35 2 MuswellbrookPerth 4 15
-35 1 Sydney
Adelaide
-40 Cape Grim
Hobart
-45
Cities Sites 300m 500m Dust
-50
Longitude
Fig. 3. Plot of 7-day back trajectories every hour with intersections per desert
region plotted as red squares for 300 m starting height and blue diamonds for
500 m starting height from the Muswellbrook sampling site (large dot) on the east
coast of Australia north of Sydney. (For interpretation of the references to colour in
this figure legend, the reader is referred to the web version of this article.).
Latitude
110
115
120
125
130
135
140
145
150
155
160
188 D.D. Cohen et al. / Nuclear Instruments and Methods in Physics Research B 273 (2012) 186–188
Table 1
The number of intersections in each of the desert boxes shown in Fig. 3 for the
extreme soil events at the Muswellbrook site.
Soil > 1.3 lg/m3 300 m 500 m
Region Soil % Soil Soil % Soil
inter. inter. inter. inter.
(1) Lake Mungo 391 33.0 382 36.5
(2) Lake Windaunka 106 8.9 103 9.8
(3) East Flinders 68 5.7 56 5.3
(4) Olympic Dam 81 6.8 60 5.7
(5) Emu Fields Salt 25 2.1 17 1.6
(6) Lake Eyre North 40 3.4 35 3.3
(7) Simpson Desert 12 1.0 11 1.1
(8) Great Victorian West 16 1.4 9 0.9
(9) Great Victorian East 30 2.5 28 2.7
(10) Gibson Desert 10 0.8 11 1.1
(11) Little Sandy Desert 4 0.3 6 0.6
(12) Great Sandy Desert 8 0.7 4 0.4
West
(13) Great Sandy Desert 8 0.7 1 0.1
East
(14) Tanami Desert 5 0.4 7 0.7
(15) Riverina Agricultural 381 32.2 317 30.3
Total with 1185 1047
soil > 1.3 lg/m3
% of all intersections 10.5 9.34. Summary
We have applied HYSPLIT back trajectories and IBA techniques
to determine fine particle soil concentrations measured over a
10-year period and to identify possible source regions. These re-
gions were previously considered to be the desert regions of
Australia up to 2000 km west of the sampling site. By linkingtrajectories to sources and receptor site we have quantitatively
determined which desert regions were most likely to contribute
to extreme soil events at the sampling site. The surprising result
was that agricultural activities more than 500 km from the sam-
pling site seemed to contribute to high fine soil events much more
than soil from desert regions as previously thought. This new ap-
proach to source identification is currently being applied to finger-
prints, obtained by IBA methods, from coal fired power stations in
Australia and in Asia to identify major polluters and their relative
contributions to ambient air pollution.
Acknowledgements
The NOAA Air Resources Laboratory (ARL) made available the
HYSPLIT transport and dispersion model and the relevant input
files for generation of back trajectories used in this paper. We
would like to acknowledge the help of the accelerator staff at
ANSTO and the financial support of Muswellbrook Council for help
with fine particle sampling and IBA support throughout this work.
References
[1] D.D. Cohen, Nucl. Instr. and Meth. B136 (138) (1998) 14.
[2] D.D. Cohen, E. Stelcer, O. Hawas, D. Garton, Nucl. Instr and Meth. B219 (220)
(2004) 145–152.
[3] D.D. Cohen, J. Crawford, E. Stelcer, V.T. Bac, Atmos. Environ. 44 (2010) 320–328.
[4] D.D. Cohen, E. Stelcer, D. Garton, J. Crawford, Aerosol pollution research 2 (2011)
182–189.
[5] R.R. Draxler, G.D Rolph, HYSPLIT (HYbrid Single-Particle Lagrangian Integrated
Trajectory) Model access via NOAA ARL READY, NOAA Air Resources Laboratory,
Silver Spring, MD, 2003. Available at: www.arl.noaa.gov/ready/hysplit4.html.
[6] P. Paatero, U. Tapper, Environmetrics 5 (1994) 111–126.
[7] W.C. Malm, J.F. Sisler, D. Huffman, R.A. Eldred, T.A. Cahill, J. Geophys. Res. 99
(1994) 1347–1370.