In luence o ain and o he me eo ological pa ame e s
on ace me als in size ac iona ed pa icles
in pollu ed u ban a mosphe e
An onio J. Fe nández Espinosa* and Miguel Te ne o Rod íguez
Depa men o Analy ical Chemis y, Facul y o Chemis y, Uni e si y o Se ille,
C. P o esso Ga cía González s/n, E-41012 Se ille, Spain; *E-mail: [email p o ec ed]s
Abs ac —Rela ionships be ween me eo ological a iables and pa icles, me als and
hei size ac ion concen a ions we e de e mined om samples collec ed in di e en
clima ic si ua ions. Special a en ion was ocused on he e ec o ain. Single co ela ion
and mul iple eg ession s a is ical me hods we e pe o med on samples om bo h ainy
and d y days. Coa se pa icle concen a ions diminished linea ly when ain all
inc eased. The me als Fe, V, Ni, Ti, and Mn diminished as well. A subs an ial
concen a ion e ec was p oduced by he empe a u e on TSP, Fe, Ti, and Mn
concen a ions in ine pa icles. Also, a dispe sion e ec was p oduced by he
a mosphe ic p essu e on TSP, Mn, Fe, Ti, Ni, and V concen a ions in hese pa icles.
Besides, he e was a dispe sion e ec by he wind speed on TSP and Cd concen a ions.
The ine pa icles and me als be ween 1.3 and 0.6 mic ome e s a e hose bes co ela ed
wi h me eo ological pa ame e s. Mul iple linea eg ession was demons a ed o be a
powe ul ool o explain he pa icle and me al le els in ela ion o size dis ibu ion and
me eo ology.
Key-wo ds: size dis ibu ion, me als, suspended pa icles, me eo ology, ain, mul iple
linea eg ession.
1. In oduc ion
Suspended a mosphe ic pa icles ha e long li e imes depending on size and
me eo ological condi ions. Size dis ibu ion depends on ae osol sou ces, bu i
is also a ec ed by p e ailing me eo ological condi ions. Apa om addi ional
ac o s, me eo ology is he mos de e minan ac o in he emo al o
dispe sion mechanisms o pa icles and me als in he ai . Many s udies ha e
in es iga ed he ela ionship be ween me eo ological condi ions and ai bo ne
pa icles (Elsom and Chandle , 1978; Wi z and Moo e, 1981; B ooks and
Salop, 1983). The la e wo used mul iple linea eg ession analyses o ela e
me eo ological pa ame e s o pa icle and se e al me al concen a ions in he
USA, speci ically in Los Angeles ci y and Sou heas e n Vi ginia S a e. We
ca ied ou a simila s udy on pa icles and lead in Se ille (Melga ejo e al.,
1986) using single and mul iple co ela ions. The ela ion be ween me eo ology
and pa icle size dis ibu ion has also been he subjec o nume ous ecen and
ea lie wo ks, such as Choula on e al. (1982) in Manches e (UK), Väke ä e
al. (2000) in Helsinki, and Despiau e al. (1996) in a Medi e anean zone o
F ance (Toulon), all o hem using single co ela ions on pa icles.
To s udy he na u e and magni ude o he me eo ology e ec on di e en
pa icle sizes, he i s ool is he s a is ical analysis o co ela ions (single
linea eg ession). Thus, he i s objec i e is o know, how me eo ology
in luences di e en sizes, ocusing specially on ine pa icles and oxic me als,
which a e he mos ha m ul.
As a second impo an objec i e, mul iple linea eg ession was
conside ed as a s a is ical echnique o he bes in o ma ion o pa icle and
me al beha io caused by he me eo ological pa ame e s.
2. Expe imen al sec ion
2.1 Measu emen si es
This s udy was done in Se ille (3812’–3651’N, 439’–632’W, 10 m a.s.l.),
he la ges ci y in sou he n Spain. Analy ical da a we e ob ained om a
ne wo k o wel e sampling s a ions al eady used in ou p e ious wo ks (Use o
e al., 1988; Fe nández e al., 1999 and 2000). The s a ions ha e di e en
a ic in ensi ies and di e en indus ial ac i i ies, as well as zones wi h clean
ai . Se ille is loca ed in he cen e o he Guadalqui i Valley, which opens
owa d he ocean a he base o he iangle ha he alley o ms.
The ci y has a wa m and d y Medi e anean clima e wi h mean annual
empe a u e o 18C, ain all 600 mm, a mosphe ic p essu e 1014 hPa, ela i e
humidi y 65%, and wind speed 2 m s–1. Se ille is cha ac e ized by high
empe a u es and low wind speeds. P e alen ai cu en s come om he SW–
NE di ec ion. All hese da a a e a e aged om he 1961–1990 pe iod (MMA,
1997). P edominan winds p oceed om he A lan ic Ocean (sou hwes ).
The e o e, he si ua ion o his sampling ne wo k in Se ille ep esen s he
me eo ological e ec s o a Medi e anean zone in luenced by he A ican
winds as in o he ci ies o sou he n Eu ope.
2.2 Pa icula e sampling
A mosphe ic pa icles we e collec ed wi h a high- olume sample (MCV,
Model CAV-A/HF) equipped wi h a i e-s age cascade impac o plus a back-
up il e (MCV, Model IC/CAV), which e ec i ely sepa a es he pa icles. I
has he ollowing equi alen cu -o diame e s a 50% e iciency (Dp): >10 m
(A pa icles), 10–4.9 m (B pa icles), 4.9–2.7 m (C pa icles), 2.7–1.3 m (D
pa icles), 1.3–0.6 m (E pa icles), and <0.6 m (backup, F pa icles). Fo A
o E s ages, i e cu il e s (14.2 cm  14.2 cm) we e used, and o he backup
il e an uncu il e (20.3 cm  25.4 cm) was used. Cu and uncu mic o- ib e
glass il e s we e pu chased om WHATMAN (GF/A).
G anulome ic ac ions a e in acco dance wi h he pa icle size ac ion
de ini ions o heal h- ela ed sampling (ISO 7708, 1995), which de ines he
ine pa icles as he ac ion below 1 m. Thus, s ages B o F can be associa ed
wi h PM10 pa icles, s ages D o F wi h PM2.5 pa icles, and s ages E plus F
wi h PM1 pa icles (Fe nández e al., 2001). The low a e should be se a he
alue o 68 m3 h–1 o ge he size sepa a ion. The low a e is calib a ed e e y
h ee mon hs a he Andalusian Re e ence Labo a o y o he Ai Quali y
(LARCA) in Se ille.
Ca e was aken in handling he ib eglass il e s in o de o a oid
con amina ion p oblems, and all il e ma e ials and samples we e handled
wi hin a e ical lamina ai low cabine , o ensu ing ai cleanliness s anda ds
o class 100 acco ding o Fede al S anda d 209E.
Fo y-one samples we e collec ed in 1996, and h ee o ou samples we e
usually aken a each sampling s a ion. The sampling ime- ame was usually
48 h (abou 3264 m3), and a weekly sample was aken on di e en days du ing
he ollowing week, so ha a possible dis o ing e ec could be a oided.
2.3 Reagen s and appa a us
Ve ical lamina ai low cabine wi h a HEPA il e was om INDELAB
(Model IDL-48V). Wa e ba h was om JULABO (Model SW-20C).
Cen i uge was om SIGMA (Model 3-15). S anda d solu ions o me als and
acids we e om MERCK. Ul a-pu e wa e was om WATERS-MILLIPORE
(Milli-Q-g ade, Model Plus).
Samples we e analyzed o ele en me als (Ca, Fe, Mg, Pb, Cu, Mn, Ti, V,
Ni, Co, Cd) by a omic emission spec ome y wi h induc i ely coupled plasma
(ICP-AES) using a Fisons-ARL 3410 sequen ial mul i-elemen ins umen .
De e mina ions in he mul i-elemen analysis we e done in iplica e o each
sample.
2.4 Me hodology o he chemical analysis
Samples and blank il e s used we e s o ed, ea ed in a da k oom and
analyzed indi idually (A-F) o me al concen a ions as in ou p e ious wo k,
Fe nández e al. (2001). The pa icle concen a ion o each s age o suspended
pa icles (FSP) was exp essed in g m–3. The o al suspended pa icles concen-
a ion (TSP) was hen calcula ed by summing he pa icle concen a ions o
he six ac ions (FSP) o each sample.
The me al concen a ion o each il e (FM) was exp essed in ng m–3. The
o al me al concen a ion (TM) was hen calcula ed by summing he concen a-
ion o he six ac ions (FM) o each sample. To di e en ia e be ween he
di e en size ac ions, hese we e numbe ed om 1 o 6, e.g., Fe1 and Fe6
co esponded o he me al Fe collec ed in s ages A and F, espec i ely.
2.5 Me eo ological da a
Me eo ological da a we e p o ided daily by ou local se ice o he Na ional
Ins i u e o Me eo ology (INM) in Se ille. These da a co espond o he same
pe iods o each pa icula e sampling. These alues o each sampling
co espond o he a e age hou ly da a p o ided h oughou he sampling
pe iods. Rain all alues we e summed wi h he hou ly da a du ing he sampling
pe iod. The me eo ological da a a e ep esen ed as ollows: p ecipi a ion (PP)
in mm, ambien empe a u e (AT) in ºC, a mosphe ic p essu e (AP) in hPa,
wind speed (WS) in m s–1, and wind di ec ion (WD) acco ding o he ca dinal
poin s.
In his s udy we conside ed he ainy and d y days sepa a ely in wo
ma ices, a e di iding he “ o al ma ix”, which showed some in e es ing
esul s. The “d y days ma ix” was o med by samples collec ed on non- ainy
days, “ ainy days ma ix” was o med he by samples collec ed on ainy days.
2.6 Mul i a ia e s a is ical analyses
On he basis o he esul s, basic and mul i a ia e s a is ical analyses we e
applied o he analy ical and me eo ological da a. Fo hese analyses he
STATISTICA (S a So , 1999) so wa e package was used.
Co ela ion s udies we e ca ied ou by applying he simple linea
eg ession (SLR) echnique. Because a high co ela ion coe icien does no
necessa ily imply linea i y, linea i y was e i ied by g aphical examina ion.
Any non-linea case was disca ded. The p ocedu e o his no only consis ed
o choosing he highes co ela ion coe icien s, bu also e i ying he linea i y
by obse ing he linea p o ile o he poin s, i.e., he pai s o da a x-y.
Some imes, high co ela ion coe icien s do no gi e a eal linea i y on a g aph,
because hei line is o med by accumula ion o poin s a he ex eme o hei
linea ange and a lone poin a he o he ex eme. In hese cases, he
co ela ion coe icien s a e due o only one sample and no all he expe imen al
da a. Ou lie s a e misleading and should he e o e be disca ded.
La e on, mul iple linea eg ession (MLR) analysis was used o ela e he
analy ical a iables, s a is ically signi ican in he SLR, o he me eo ological
pa ame e s, h ough ma hema ical mul i a ia e linea unc ions. Fo bo h SLR
and MLR, when he co ela ion coe icien ( ) was no su icien ly high, a wo-
ailed - es was applied wi h a 95% con idence le el o assess whe he  0
was signi ican . Thus, i he calcula ed o he s uden es was g ea e han he
abula ed, hen was signi ican ly g ea e han ze o.
F om he i s calcula ions o he da a, we obse ed, ha in Se ille ain is
he pa ame e wi h he g ea es in luence on pa icle and me al le els as
opposed o he o he pa ame e s. The he e ogenei y and dis o ion ac o
in oduced by he washing e ec o he ain makes he conclusions un eliable
and jus i ies he need o ea each si ua ion sepa a ely.
The s a is ical s udies ook in o conside a ion, i s ly, he e ec o ain on
o al suspended pa icles and o al me al concen a ions and hei size
dis ibu ions, and secondly, he e ec o he emaining me eo ological
pa ame e s on pa icle and me al concen a ions and hei size dis ibu ions.
3. Resul s and discussions
3.1 Me eo ological cha ac e is ics o he sampling pe iod
Du ing he whole yea in 1996, a o al o 916.8 mm o ain all was egis e ed,
which ep esen s a conside able inc ease compa ed o he p e ious yea s
(580 mm in 1993, 327 mm in 1994, and 503 mm in 1995). Me eo ological
pa ame e s eco ded only o he sampling pe iod a e p esen ed in Table 1. The
mean ain all on ainy days was 13.6 mm. Mean wind speed was low, 2.2 m
s–1, al hough i was su icien o cause he esuspension o soil pa icles in he
ai . Depending on maximum alues, he e ec can each he dispe sion o
pollu ion. I should be no ed also, ha he empe a u es we e high. Wi h ega d
o he equency o he wind di ec ion, he mos common winds we e sou h-
wes e ly o sou he ly, al hough, he s onges winds came om he wes (Fig.
1-a). This was o be expec ed, because he wind blows up he axis o he
Guadalqui i Valley (SW–NE).

Table 1. Mean alues o he analy ical and me eo ological a iables co esponding
o he sampling pe iod
Va iable Mean Range Uni s RSD (%)
Me eo ological a iables
PP*
WD
WS
AT
AP
13.6
SW
2.2
21.0
1010.5
(1.2 – 141.0)
(NW – W)
(0.6 – 4.2)
(15.3 – 29.8)
(1003.7 – 1018.4)
mm
m s–1
C
hPa
167.8
27.0
44.3
20.5
0.4
To al and ac iona ed pa icle a iables
TSP
A
B
C
D
E
F
78.7
12.1
12.4
4.9
4.0
3.6
41.7
(31.1 – 158.1)
(3.1 – 27.3)
(6.6 – 21.1)
(2.8 – 7.5)
(2.0 – 6.0)
(0.8 – 7.1)
(11.7 – 104.5)
g m–3 44.0
53.6
33.4
26.9
31.2
46.4
57.0
To al me al a iables
Ca
Fe
Mg
Pb
Cu
Mn
Ti
V
Ni
Co
Cd
2956
689
366
97
28.0
19.8
7.4
5.9
1.97
0.56
0.44
(345 – 8680)
(131 – 1974)
(200 – 669)
(15 – 1335)
(8.3 – 74.2)
(4.1 – 141.0)
(1.7 – 16.9)
(0.6 – 18.0)
(0.47 – 4.31)
(0.09 – 2.01)
(0.09 – 4.73)
ng m–3 81.6
58.3
29.2
60.0
51.5
52.6
52.0
76.5
58.9
76.7
61.4
*: Mean ain all on ainy days, RSD: Rela i e s anda d de ia ion in pe cen age, WD: wind di ec ion
The ela ionship be ween wind speed, ain all, a mosphe ic p essu e, and
empe a u e is an impo an clima ic ea u e. Thus, he co ela ion analysis is a
good ool, because i shows me eo ological beha io , o example, he well
known ela ions be ween he ain, he dec ease o empe a u e and a mosphe ic
p essu e, and he inc ease o wind speeds. O he esul s show ha mild wa m
winds come om he sou h (p oceeding om A ica) and sligh ly less wa m
winds om he Sou hwes ( he A lan ic Ocean). The coldes winds a e also he
s onges , hose coming om he wes ( he A lan ic Ocean c ossing Po ugal).
The highes ain all alues a e co ela ed wi h sou h-wes e ly winds, because
he mois u e easily pene a es in o he alley.
Fig. 1. (a) F equency o wind di ec ion, alues o median, maximum, and minimum
o wind speeds o all samples. (b) Dis ibu ion o TSP le els acco ding o he wind
di ec ions wi h alues o median, maximum, minimum, and qua iles.
3.2 Basic analy ical pa ame e s
A
(a) His og am o wind di ec ion and wind speed ( o al ma ix)
Wind di ec ion
No. o obse a ions
Wind speed (m s
-1
)
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
NE E SE S SW W NW N
Max
Min
Median
Max
Min
75%
25%
Median
(b) Box & Whiske Plo o TSP s. wind di ec ion ( o al ma ix)
Wind di ec ion
TSP (µg m
-3
)
20
30
40
50
60
70
80
90
100
110
120
130
140
150
160
170
180
N NE W SW SE S E
In o de o know he le els o pa icles and me als, hey a e also included in he
basic s a is ics o Table 1. The mean alue o 78.7 g m–3 o TSP is om 50–
150 g m–3, ha would co espond o an “accep able” ai quali y acco ding o
Eu opean di ec i e 1999/30/EC o PM10 pa icles, bu no a “good” ai quali y
(unde 50). I he mean alue o A pa icles is sub ac ed om TSP, he
esul ing mean alue, which would co espond o PM10 pa icles (sum o B o
F s ages), is 66.6 g m–3, s ill abo e 50 g m–3, he inal limi alue o 24 h.
Two samples o TSP exceeding 150 g m–3, co esponds o a “poo ” ai
quali y, al hough no PM10 alue exceeds his alue. 23 samples o 41 exceed
he limi alue o 24 h in PM10 pa icles, bu addi ionally 16, 12, and 12
samples exceed his alue in PM2.5, PM1, and F pa icles, espec i ely. The F
pa icles, wi h Dp<0.6 m, eaches a maximum alue o 104.5 g m–3. This is
ele an , because his size has a high p obabili y o deposi ing in he al eola
egion o he lungs. Remembe , ha 40% o hese 41 samples we e collec ed in
ainy days. I is known ha he sum o pa icle concen a ion o a ious
consecu i e il e s om a cascade impac o con ains e o s a ibu ed o a
possible accumula ion o mass om con e sion o gas o pa icles in o he
il e s. This way, mainly pa icles lowe han 1 m om his con e sion a e
bound o each s age, inc easing he mass o each il e bu no changing
signi ican ly hei pe cen ages ega ding he sum. Al hough his ac exis s, i is
clea ha nume ous sum alues a e much highe han he limi o PM10
pa icles, e en in he sum alues co esponding o PM2.5, PM1, and F
pa icles. Because o his e ec o accumula ion o mass in he sum, we canno
include hese PM concen a ions as analy ical a iables in he s a is ical
ea men , excep o hese compa isons.
In ela ion o he o al me al concen a ions, he lead alue is i e imes lowe
(0.097 ng m–3) han he limi es ablished by he Eu opean Communi y, howe e ,
h ee samples collec ed nea a ound y exceed his limi o 0.5 g m–3. The e o e,
al hough lowe le els o lead a e emi ed by leaded uel d i en ehicles, high
alues can be de ec ed locally in he p oximi y o ound ies, o example he
maximum alue in Table 1. These high mean alues o pa icles and low mean
alues o me als indica e he impo an in luence o he No h Wes e n A ican
pa icles om he Saha a dese .
3.2.1 To al pa icle con en , size dis ibu ion, and he e ec o ain
In o de o s udy he di e ences in oduced by he ain in pa icle and me al
le els, Fig. 2 shows he a e aged alues o concen a ions by compa ing d y
days and ainy days. The g ea in luence o ain is exp essed as pe cen age.
Fig. 2. A e aged alues o (a) o al suspended pa icles, TSP, and ac iona ed
suspended pa icles, FSP, and (b), (c), (d) o al me al concen a ion, TM. E ec o
ain all compa ing d y days and ainy days.
(b)
Concen a ion (ng/m3)
0
1000
2000
3000
4000
Ca Fe Mg
D y days
Rainy days
-19.7%
-58.1% -7.8%
(d)
Concen a ion (ng/m3)
0
1
2
3
Ni Co Cd
D y days
Rainy days
-53.8%
+29.1%
-39.3
(a)
Pa icula e a iables
Concen a ion (µg/m3)
0
20
40
60
80
100
TSP A B C D E F
D y days
Rainy days
-36.3% -14.6%
-0.1% -23.1% -41.0%
-50.9%
-39.1%
(c)
Concen a ion (ng/m3)
0
20
40
60
80
100
Pb Cu Mn Ti V
D y days
Rainy days
-54.1% -56.3%
-53.8%
-27.0%
-32.2%
Cen al Eu ope. The e o e, he new Di ec i e con empla es he possibili y o
keeping in ocus he di e en geog aphical si ua ions (A icle 5.4 o he Council
Di ec i e 1999/30/EC, 1999).
S udying he ela ionship be ween empe a u es and me al concen a ions,
one ound ha ea h c us al me als such as Mn, Fe, and Ti had high
concen a ions a high empe a u e alues ( = + 0.64, = + 0.61, = + 0.59,
espec i ely, Table 2-a). This ac suppo s he wo explana ions abo e
men ioned, because bo h pa icle sizes a e coming om ea h c us al sou ces.
When we s udied hei size dis ibu ion (Table 4-a), we obse ed ha he e ec
ebounded on Fe5 pa icles ( = + 0.66), Mn4 pa icles ( = + 0.69), and pa icles o
Ti <2.7 m, mainly Ti5 ( = + 0.66), i.e., he ine me al ac ions again, con ained
in PM2.5. Howe e , in ainy samples (Table 4-c), he size ac ions, ha ha e
high co ela ion, a e he pa icles >10 m, such as Fe1 ( = + 0.62) and V1 ( = +
0.64), i.e., he coa se pa icles ha espond o high empe a u es in ainy
si ua ions.
On he o he hand, we obse ed (Table 2-b), ha , con a ily, cobal was
nega i ely co ela ed wi h empe a u e ( = – 0.65), showing beha io opposi e o
ha o TSP, Mn, Fe, and Ti. The size ac ion ha ollowed his end was
ep esen ed by pa icles be ween 4.9–2.7 m ( = – 0.58, Table 4-b). The
explana ion could be ela ed o he s abili y o he cobal complexes in he ainy
si ua ions men ioned abo e. Since empe a u e is nega i ely co ela ed wi h
ain all and cobal is posi i ely co ela ed wi h ain all, logically cobal is
an ico ela ed wi h empe a u e.
Rega ding he a mosphe ic p essu e pa ame e , he co ela ion s udy also
e eal a dispe sion e ec on TSP (Table 2-b, = – 0.67). This e ec o he
an icyclonic condi ions was s onges o he ine E pa icles (Table 2-b,
= – 0.67), bu he an ico ela ions a e high o all size ac ions.
The e we e also se e al an ico ela ions o me al concen a ions (Table 2-
b), wi h Mn ( = – 0.69), Fe ( = – 0.61), Ti ( = – 0.57), Ni ( = – 0.57), and V ( = –
0.55). S udying hei size dis ibu ion (Table 4-b) we ound, ha p edominan
size ac ions o Mn we e pa icles >4.9 m ( = – 0.70 o Mn1 pa icles),
pa icles < 1.3 m o Fe ( = – 0.60 o Fe5 pa icles), Ti5 pa icles ( = – 0.56),
Ni6 pa icles ( = – 0.54), and V5 and V6 pa icles ( = – 0.58 o V5 pa icles).
One can obse e ha he me als ha ing co ela ions o an ico ela ions wi h
he main me eo ological pa ame e s a e always he ea h c us al elemen s Fe, Ti,
and Mn. In hese cases no only ine pa icles we e highly co ela ed, bu also
he coa se pa icles in se e al me als.
Co ela ions and an ico ela ions o pa icles me als, and hei size ac ions
we e ound, which could explain se e al ela ionships wi h me eo ological
pa ame e s. A mosphe ic p essu e has shown s onge dispe sion e ec han
wind speeds on suspended pa icles and me als, and empe a u e shows a
con a y e ec on concen a ion. Bu , in some analy ical a iables he e a e
nega i e and posi i e co ela ions a he same ime. Fo example, in he ainy

ma ix, anadium is co ela ed wi h ambien empe a u e and an ico ela ed wi h
a mosphe ic p essu e, and he same ac occu s in o al and d y ma ices wi h
TSP, Fe, Mn, Ti, and Ni concen a ions. The e o e, wha will he global
beha io o hese elemen s be, aced wi h bo h pa ame e s a he same ime?
This ques ion can no be answe ed by single co ela ions, bu wi h mul iple
eg essions.
3.3 Mul iple linea eg ession analysis o co ela e he beha io o a iables
wi h me eo ological condi ions
In he p e ious sec ion, we ound signi ican co ela ions be ween analy ical
a iables and me eo ological da a h ough simple linea eg ession (SLR)
analyses, mainly wi h ain all, empe a u e, and a mosphe ic p essu e. As
expec ed, i was also possible o ind mul i a ia e equa ions o ela e only he
co ela ed a iables in he SLR (Tables 2 and 4) wi h all he me eo ological
pa ame e s a he same ime using he mul iple linea eg ession (MLR) analysis
(Fe nández e al., 2000). These equa ions can be use ul in e i ica ion i he
concen a ions in he ai can ollow a basic model o beha io o pa icles
me als, and hei size ac ions, and i his model is mo e accu a e han he SLR.
The e o e, applying he S a is ica package we ob ained hese equa ions o he
h ee ma ices (Table 5). This able shows, also unde lined, he signi ican
coe icien s o me eo ological pa ame e s.
One can obse e in Table 5, ha he equa ions ha gi e he wo s
co ela ion coe icien s o TSP concen a ions a e hose ob ained o he ainy
days ma ix ( = 0.60). Consequen ly, he coe icien is lowe o he o al ma ix
( = 0.71), which con ains ainy samples, han o he d y days ma ix ( = 0.74),
wi hou ainy samples. This is also e idence, ha ain si ua ions dis o he s udy
o a mosphe ic pollu ion a lo by o al suspended pa icles. This di e ence can
be e i ied in Fig. 3-b,c, whe e he poo capaci y o p edic ion o he equa ion
on ainy days can be seen in ela ion o he d y days. This ac should be due o
he di e en beha io s o he di e en size ac ions in ainy si ua ions mos
in luenced by he ain. Thus, a iabili y (Table 1) in ain all a iable (168 %) is
g ea e han wind speed (44%) o example, and, he e o e, he success o
p edic ion is lowe . Only se e al ac ions o soluble elemen s such as
magnesium o anadium (Fig. 3- ) ha e a be e capaci y o p edic ion in ainy
day ma ices.
Table 5. Resul s o he mul iple linea eg ession (MLR) o he TSP, FSP, TM, and FM
a iables in unc ion o all me eo ological a iables. The able unde lines he signi ican
coe icien s o me eo ological pa ame e s
Va iable
(Y)*
a b1 b2 b3
b4 Co ela ion
coe icien
( )
calc Deg ees
o
eedom
ab
(a) To al ma ix (41 samples)
TSP
3469.328
–
1.787
–
0.672
4.0
91
–
3.426
0.7128
6.35
36
2.028
F
1334.967
–
1.015
0.208
3.036
–
1.339
0.6855
5.88
36
2.028
Mn
497.983
–
0.427
–
0.264
0.914
–
0.491
0.7212
6.50
36
2.028
Mn4
1.569
–
0.013
–
0.020
0.056
–
0.001
0.7372
6.81
36
2.028
Fe
22467.817
–
19.666
–
13.169
35.747
–
22.125
0.7001
6.12
36
2.028
Fe5
469.961
–
0.509
–
1.124
1.720
–
0.461
0.7323
6.72
36
2.028
Ti
166.131
–
0.154
–
0.230
0.311
–
0.161
0.6766
5.74
36
2.028
Ti5
3.216
–
0.004
–
0.018
0.015
–
0.003
0.7655
7.43
36
2.028
Cd
8.653
0.001
–
0.027
0.016
–
0.008
0.6323
5.10
36
2.028
Cd6
4.314
0.000
–
0.016
0.014
–
0.004
0.5874
4.53
36
2.028
Ni
90.505
–
0.055
–
0.096
0.069
–
0.088
0.6448
5.27
36
2.028
Ni1 8.579 –0.001
–0.022
0.001
–0.008
0.5891 4.55 36 2.028
(b) D y days ma ix (25 samples)
TSP
7485.992
–
1.414
2.663
–
7.367
0.7421
6.91
21
2.080
E
363.016
–
0.091
0.145
–
0.358
0.7606
7.32
21
2.080
Co
13.169
–
–
0.006
–
0.038
–
0.012
0.6682
5.61
21
2.080
Co3
–
1.808
–
0.000
–
0.006
0.002
0.5902
4.57
21
2.080
Mn
1830.889
–
–
0.017
0.525
–
1.799
0.7448
6.97
21
2.080
Mn1
468.513
–
–
0.199
0.046
–
0.459
0.7741
7.64
21
2.080
Fe
74643.533
–
6.169
17.149
–
73.283
0.6445
5.26
21
2.080
Fe6
34607.535
–
16.263
10.551
–
34.147
0.6710
5.65
21
2.080
Ti
681.318
–
–
0.020
0.137
–
0.667
0.5946
4.62
21
2.080
Ti5
25.812
–
–
0.012
0.007
–
0.025
0.6597
5.48
21
2.080
Ni
242.965
–
–
0.059
0.017
–
0.237
0.5917
4.58
21
2.080
Ni1
23.130
–
–
0.031
–
0.004
–
0.022
0.7211
6.50
21
2.080
V
1064.356
–
–
0.220
–
0.145
–
1.039
0.5769
4.41
21
2.080
V1
103.101
–
–
0.064
–
0.019
–
0.100
0.6781
5.76
21
2.080
Cd
16.048
–
–
0.032
0.014
–
0.016
0.5874
4.53
21
2.080
Cd5 1.352 – –0.006
0.002
–0.001
0.5707 4.34 21 2.080
(c) Rainy days ma ix (16 samples)
TSP
255.600
–
0.453
–
3.053
7.313
–
0.287
0.5996
4.68
11
2.201
B
715.802
–
0.182
–
0.108
0.285
–
0.701
0.7330
6.73
11
2.201
V
–
416.026
–
0.074
–
0.410
0.943
0.406
0.7308
6.69
11
2.201
V2
–
2.096
–
0.010
–
0.048
0.058
0.002
0.8293
9.27
11
2.201
Mg1
6487.662
–
1.840
5.547
–
1.105
–
6.403
0.8180
8.88
11
2.201
Fe –
16502.994
–3.521
–24.701
78.671
15.682
0.6804 5.80 11 2.201
Fe1
–
2304.284
–
1.997
–
5.962
16.904
2
.163
0.7630
7.37
11
2.201
Mn1
–
214.868
–
0.071
–
0.134
–
0.006
0.218
0.6949
6.03
11
2.201
Ti1
–
56.542
–
0.019
–
0.107
0.064
0.057
0.7333
6.74
11
2.201
Cd6 –3.824 0.000
–0.017
0.000
0.004
0.6364 5.15 11 2.201
*: Equa ion: Y = a + b1 PP + b2 WS + b3 AT + b4 AP
Fig. 3. Compa ison be ween concen a ions o he expe imen al alues o analy ical pa ame e s and he co esponding
p edic ed alues o (a) TSP in o al ma ix, (b) TSP in d y days ma ix, (c) TSP in ainy days ma ix, (d) E pa icles in d y
days ma ix, (e) B pa icles in ainy days ma ix, and ( ) V2 in ainy days ma ix.
(b)
0
20
40
60
80
100
120
140
160
180
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
TSP µg m
-3
TSP = 7485.992 + 1.414 WS + 2.663 AT - 7.367 AP = 0.742
(d)
0
1
2
3
4
5
6
7
8
9
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
E ng m
-3
E = 363.016 + 0.091 W S + 0.145 AT - 0.358 AP = 0.761
(c)
0
20
40
60
80
100
120
140
160
180
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
TSP µg m
-3
TSP = 255.600 - 0.453 PP - 3.053 WS - 7.313 AT - 0.287 AP = 0.600
(a)
0
20
40
60
80
100
120
140
160
180
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41
TSP µg m
-3
TSP = 3469.328 - 1.787 PP - 0.672 WS + 4.091 AT - 3.426 AP = 0.713
(e)
0
2
4
6
8
10
12
14
16
18
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
B ng m
-3
C alcula ed O bse ed
B = 715.802 - 0.182 PP - 0.108 W S + 0.285 AT - 0.701 AP = 0.733
( )
0,0
0,2
0,4
0,6
0,8
1,0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
V2 ng m
-3
Calcula ed Obse ed
V2 = - 2.096 - 0.010 PP - 0.048 WS + 0.058 AT + 0.002 AP = 0.829
1.0
0.8
0.6
0.4
0.2
0.0
Rega ding he size ac ions o pa icles, in he d y days ma ix, he E
pa icles a e mos highly co ela ed ( = 0.76). These pa icles a e hose, in which
mo e co ela ion and an ico ela ion we e ound, he e o e, hey a e he pa icle
sizes mos in luenced by me eo ological pa ame e s. In bo h, o al and d y
ma ices, he ine pa icles (E and F ac ions) a e he mos easily p edic ed (Fig.
3-d o he E pa icles). Howe e , in ainy si ua ions, he coa se pa icles a e he
bes co ela ed, he A and B pa icles ins ead o ine pa icles (Fig. 3-e o B
pa icles). This ac is in acco dance wi h he ac ha he coa se pa icles a e
hose ha espond posi i ely o he amoun o ain all, al hough he ine pa icles
a e hose ha a e bes emo ed by he ain independen ly o he ain all. Thus,
he ainy si ua ions in oduce a majo di icul y when p edic ing he beha io o
concen a ions o ine pa icles, which a e he mos ha m ul.
Rega ding he o al me al and hei size ac ion concen a ions, in d y and
o al ma ices he me al wi h he highes co ela ion is manganese ( = + 0.74 and
= + 0.72, Table 5), bu on ainy days anadium ( = + 0.73) is he me al ha was
ound o be a model o beha io mo e accu a e han he SLR. Meanwhile,
manganese does no appea in he ainy days ma ix because o i s low
coe icien ( is signi ican ly equal o ze o). The me al i on is he only one ha
appea s wi h a coe icien highe han 0.64 in he h ee ma ices, and is also he
only one ha has a highe co ela ion coe icien in he ainy days ma ix ( = +
0.68), al hough wi h smalle di e ences, han in he d y days ma ix ( = + 0.64).
The co esponding size ac ions o me als ha e gene ally highe
coe icien s han he o al concen a ions, excep o se e al cases, such as Cd6
and Ni1 in he o al ma ix, and Co3 o Cd5 in he d y days ma ix. Many o
hese coe icien s a e high o ex emely high, o example, Ti5, Mn4, and Fe5 in
he o al ma ix, Mn1 and Ni1 in he d y days ma ix, and Fe1, Ti1, and mainly
V2 and Mg1 in ainy days ma ix, wi h coe icien s highe han 0.82. These
me al ac ions a e he bes explained in bo h ma ices (Table 5).
These esul s con i m again he impo ance o he ela ionship o he
me eo ological pa ame e s wi h he pa icles and me als depending on hei
co esponding sizes.
4. Conclusions
In e es ing in e ela ions we e ound ha explain he beha io o pa icles and
me als o di e en sizes in di e en me eo ological condi ions: he e ec o
empe a u e on he inc ease o pa icle concen a ions, he e ec o a mosphe ic
p essu e and wind speed on he dispe sion o pa icles, and he elimina ion o
pa icles by he ain.
Mul iple linea eg ession echnique was no ably use ul o con i m he poo
capaci y o p edic ion on ainy day condi ions o pa icles and me als. The bes
p edic ion was obse ed o me als ha can be in soluble o ms in he ai , such
as magnesium and anadium. The pa icles lowe han 1.3 mic ome e s a e he
mos easily p edic ed when s udying he size dis ibu ion o pa icles. The me al
concen a ions wi h he highes co ela ion coe icien s in he MLR we e
obse ed o ine pa icles, such as he Fe, Mn, and Ti me als, o coa se
pa icles, such as he Mn, Ni, Mg, Fe, V, and Ti me als.
We conclude ha ain all is he me eo ological pa ame e ha mos a ec s
pa icle elimina ion and me al pollu ion by physical and chemical washing away
o pa icles. The o he me eo ological pa ame e s we e also obse ed
in luencing he p esence o di e en size pa icles in he ai (wind speed,
empe a u e, and p essu e). These e ec s a e di e en o di e en size
ac ions.
I would be in e es ing o ex apola e om hese me hods, i used in o he
egions and coun ies, o he possible da a, checking he di e ences and
simila i ies wi h hei esul s. Al hough, ou i s objec i e a e his s udy will
be o epea his esea ch o a longe ime pe iod wi h a la ge numbe o
samples and se e al addi ional me eo ological pa ame e s. The s udy o he size
dis ibu ion and me eo ological pa ame e s cons i u es a way o inc ease he
in o ma ion abou he li e imes o pollu an s in he u ban ai . The sa is ac o y
esul s o he mul iple linea eg ession echnique should be checked in o a
con ex o a wide esea ch wi h a wide sampling pe iod.
Acknowledgemen s—We would like o hank he “Conseje ía de Medio Ambien e de la Jun a de
Andalucía” o hei inancial assis ance in ca ying ou his esea ch p ojec .
Re e ences
B ooks, L. and Salop, J., 1983: Chemical and me eo ological cha ac e is ics o a mosphe ic
pa icula es in sou heas e n Vi ginia u ilizing mul i- a ian analysis. J. Ai Con ol Associa ion
33, 222-224.
Choula on, T W., Fulla on, G., and Gay, M.J., 1982: Some obse a ions o he in luence o
me eo ological a iables on he size dis ibu ion o na u al ae osol pa icles. A mosphe ic
En i onmen 16, 315-323.
Council Di ec i e 1999/30/EC o 22 Ap il, 1999: Rela ing o limi alues o sulphu dioxide, ni ogen
dioxide and oxides o ni ogen, pa icula e ma e and lead in ambien ai . Commission o he
Eu opean Communi ies. O icial Jou nal L163, 41-60.
Despiau, S., Cougnenc, S., and Resch, F., 1996: Concen a ions and size dis ibu ions o ae osol
pa icles in coas al zone. J. Ae osol Science 27, 403-415.
Elsom, D.M. and Chandle , T.J., 1978: Me eo ological con ols on g ound le el concen a ions o
smoke and sulphu dioxide in wo u ban a eas o he U.K. A mos. En i on. 12, 1543-1554.
Fe nández, A.J., Te ne o, M., Ba agán, F.J., and Jiménez, J.C., 1999: Sou ce cha ac e iza ion o
ai bo ne pa icles in Se ille (Spain) by mul i a ia e s a is ical analyses. Időjá ás 103, 261-273.
Fe nández, A.J., Te ne o, M., Ba agán, F.J. and Jiménez, J.C., 2000: An app oach o cha ac e iza ion
o u ban ai bo ne pa icles sou ces h ough hea y me als specia ion. Chemosphe e 2, 123-136.
Fe nández, A.J., Te ne o, M., Ba agán, F.J., and Jiménez, J.C., 2001: Size dis ibu ion o me als in
u ban ae osols in Se ille (Spain). A mosphe ic En i onmen 35, 2595-2601.
Fe nández, A.J., Te ne o, M., Ba agán, F.J., and Jiménez, J.C., 2002: A chemical specia ion o ace
me als o ine u ban pa icles. A mosphe ic En i onmen 36, 773-780.

ISO 7708, 1995: Ai Quali y - Pa icle Size F ac ion De ini ions o Heal h- ela ed Sampling. ISO
Publica ions, Fi s Edi ion 1995-04-01.
Melga ejo, P.L., Te ne o, M., and G acia, I., 1986: A s udy o he a mosphe ic lead pollu ion in
Se illa, Spain. In luence o me eo ology and a ic, and ela ionship wi h o he a ic-
gene a ed pollu an s. In . J. En . Analy ical Chemis y 24, 283-296.
Mészá os, E., 2002: A mosphe ic we deposi ion as a nu ien supply o he ege a ion. Időjá ás 106,
103-111.
MMA, Minis e io de Medio Ambien e, 1997: Guía esumida del clima en España 1961-1990.
Di ección Gene al del Ins i u o Nacional de Me eo ología. Edi a Sec e a ía Técnica de Medio
Ambien e. Mad id.
Molina oli, E., Gue zoni, S., and Rampazzo, G., 1993: Con ibu ion o Saha an dus o he Cen al
Medi e anean basin. In P oc. Con olling he Composi ion o Clas ic Sedimen s (eds.: M.J.
Johnsson and A. Basu). Geological Socie y o Ame ica. Bulle in Special Pape , 303312.
Schü z, L. and Sebe , M., 1987: Mine al ae osols and sou ce iden i ica ion. J. Ae osol Science 18, 1-
10.
S a So , Inc., 1999: STATISTICA’ 99 Edi ion o Windows (Compu e p og am), Ke nel elease 5.5.
Tulsa, OK, USA.
Use o, J., Rosa, F., Te ne o, M., and G acia, I., 1988: A de e mina ion o he sou ces in he Se ille
u ban ae osol. In . J. En . Analy ical Chemis y 33, 233-244.
Väke ä, M., Häme i, K., Puhakka, T., Nilsson, E.D., Hoh i, H., and Mäkelä, J.M., 2000: E ec s o
me eo ological p ocesses on ae osol pa icle size dis ibu ion in an u ban backg ound a ea. J.
Geophysycal Resea ch 105(D8), 9807-9821.
Wi z, S. and Moo e, A.B., 1981: E ec o me eo ology on he a mosphe ic concen a ions o a ic-
ela ed pollu an s a a Los Angeles Si e. J. Ai Pollu ion Con ol Associa ion 31, 1098-1101.