Influence of rain and other meteorological parameters on trace metals in size fractionated particles in a polluted urban atmosphere

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.