Comparison of Methodologies for Assessing Desert Dust Contribution to Regional PM10 and PM2.5 Levels: A One-Year Study Over Portugal

a mosphe e A icle Compa ison o Me hodologies o Assessing Dese Dus Con ibu ion o Regional PM10 and PM2.5 Le els: A One-Yea S udy O e Po ugal Ca la Gama 1,* , Casimi o Pio 1, Alexand a Mon ei o 1, Michael Russo 1, Ana Pa ícia Fe nandes 1, Ca los Bo ego 1, JoséMa ía Baldasano 2and Oxana Tchepel 3 1Depa men o En i onmen and Planning, CESAM, Uni e si y o A ei o, 3810-193 A ei o, Po ugal; [email p o ec ed] (C.P.); [email p o ec ed] (A.M.); michaela [email p o ec ed] (M.R.); [email p o ec ed] (A.P.F.); [email p o ec ed] (C.B.) 2En i onmen al Modeling Labo a o y, Technical Uni e si y o Ca alonia, 08034 Ba celona, Spain; [email p o ec ed] 3 Depa men o Ci il Enginee ing, CITTA, Uni e si y o Coimb a, 3030-788 Coimb a, Po ugal; [email p o ec ed] *Co espondence: [email p o ec ed] Recei ed: 15 No embe 2019; Accep ed: 21 Janua y 2020; Published: 24 Janua y 2020   Abs ac : Dese dus ou b eaks may a ec ai quali y. This s udy es ima es he impo ance o A ican dus con ibu ion o he PM10 and PM2.5 concen a ions obse ed in u al egional backg ound si es in Po ugal. Dese dus con ibu ion is e alua ed by wo di e en app oaches: A measu emen -app oach me hodology based on he mon hly mo ing 40 h pe cen ile, and a model-app oach me hodology based on WRF-CHIMERE simula ions, whose pe o mance is also assessed wi hin his wo k. Se e al dese dus episodes a ec ed a mosphe ic ae osols in he plane a y bounda y laye o e Po ugal du ing 2016. Thei in ensi y was a iable, wi h a leas wo e en s (21–22 Feb ua y and 27–28 Oc obe ) con ibu ing o exceedances o he PM10 daily limi alue de ined in he Eu opean Ai Quali y Di ec i e. A ican dus con ibu ions ob ained o he yea 2016 wi h he measu emen -app oach me hodology a e highe han he ones simula ed by WRF-CHIMERE. Con ibu ions o PM10 and o PM2.5 concen a ions ange om 0 o 90 µ g m −3 and om 0 o 30 µ g m −3 , espec i ely, in mos o he egions and days. Cau ion mus be employed when using measu emen -app oach me hodologies o quan i y dus con ibu ions o PM le els when o es i es occu simul aneously wi h he long- ange anspo o dese dus , as happened in Augus 2016. Keywo ds: dese dus ; pa icula e ma e ; Po ugal; ai quali y; CHIMERE; mon hly mo ing 40 h pe cen ile 1. In oduc ion Pa icula e ma e (PM) concen a ions in he a mosphe e a e in luenced by se e al na u al and an h opogenic sou ces. Amongs he na u al p ocesses, long- ange anspo o dus emi ed om dese s and o he a id a eas may a ec ai quali y, se e ely (e.g., Re e ences [ 1 , 2 ]). Those con ibu ions om na u al sou ces o PM le els can be assessed, bu no con olled. Despi e a educ ion, o e he las yea s, o PM concen a ions o e Eu ope [ 3 ] and Po ugal [ 4 ], PM10 and PM2.5 (pa icles wi h ae odynamic diame e s smalle han 10 µ m and 2.5 µ m, espec i ely) con inue o be pollu an s o pa icula conce n, exceeding he limi alues de ined by he Eu opean Ai Quali y Di ec i e (Di ec i e 2008/50/EC) o he p o ec ion o human heal h, e e y yea in di e en moni o ing si es. Whe e na u al con ibu ions o pollu an s in ambien ai can be de e mined wi h su icien ce ain y and when exceedances a e due in whole o in pa o na u al con ibu ions, he Di ec i e allows hei sub ac ion when assessing compliance wi h ai quali y limi alues. This A mosphe e 2020,11, 134; doi:10.3390/a mos11020134 www.mdpi.com/jou nal/a mosphe e A mosphe e 2020,11, 134 2 o 21 sub ac ion is pa icula ly impo an when penal ies (due o exceedances o he legisla ed limi alues) can be a oided. The mos accu a e me hod o quan i y dus con ibu ion o PM concen a ions is by sampling and pe o ming a chemical cha ac e iza ion o he ae osol. Howe e , such analyses a e usually pe o med du ing speci ic campaigns and no as ou ine measu emen s o he a mosphe e. Thus, o he me hods a e equi ed o es ima e he con ibu ion o dus o PM concen a ions. Cu en ly, one o he o icial me hods [ 5 ] adop ed by he Eu opean Commission o e alua ing he occu ence o A ican dus ou b eaks and quan i ying hei con ibu ions, and he one adop ed by he Po uguese En i onmen Agency, is a s a is ical app oach based on he a iabili y a u al backg ound si es [ 6 ] (he ea e e e ed o as P40). I is based on he applica ion o a 30-day mo ing 40 h pe cen ile o he PM da a se ies in u al backg ound s a ions, a e excluding hose days impac ed by A ican dus , which mus be p e iously iden i ied based on sa elli e obse a ions, model back- ajec o ies and/o dus o ecas models. The easibili y o his me hod was demons a ed by compa ing expe imen ally measu ed concen a ions o he mine al ma e de e mined a h ee Spanish u al backg ound si es (Monag ega and Mon seny, bo h loca ed in he Eas e n pa o he Ibe ian Peninsula, and Bell e , loca ed in he Balea ic Islands) e sus he es ima ed A ican dus con ibu ions ob ained by his p ocedu e. O he measu emen -based me hodologies o de ec and quan i y A ican dus con ibu ions o PM le els a e ound in he li e a u e. The Tel A i Uni e si y, in Is ael, de eloped a me hod which uses only PM10 and PM2.5 measu emen s om au oma ic s a ions [ 7 ]. This me hod employs an au oma ic algo i hm wi h h ee h esholds and does no equi e any o he inpu s, such as sa elli e obse a ions, model back- ajec o ies, dus o ecas models, o mine alogical analyses. Viana e al. [ 8 ] compa ed his me hod wi h he one desc ibed abo e, and ound ha he P40 me hod is mo e conse a i e in he de ec ion o A ican dus episodes, as i is less a ec ed by in e e ences om local dus sou ces. Ano he p ocedu e o quan i ying he wind-blown dese con ibu ions o daily a e age PM10 concen a ions om moni o ing si es was p oposed by G ó mez-Losada e al. [ 9 ]. This measu emen - based me hodology is based on he use o Hidden Ma ko Models. In ha s udy, he annual a e age Saha an PM10 con ibu ion in he Cana y Islands, Spain, was es ima ed and compa ed o he one es ima ed by he P40 me hod. The annual con ibu ion o No h A ican episodes o he PM10 mean alue in he Cana ias A chipelago coincides ma kedly wi h he same es ima ion made wi h he P40 me hod. Recen ly, some modi ica ions o he P40 me hodology we e sugges ed and ansla ed in o an au oma ic ool by Ba naba e al. [ 10 ]. This P40- e ised me hodology was success ully implemen ed o e I aly and showed be e pe o mances in p edic ing iming and absolu e alues o he dese -dus con ibu ion o he daily PM10 concen a ions wi h espec o he o icial P40 app oach. Mo eo e , in he scope o he inDUST COST Ac ion (h ps://cos -indus .eu, CA16202), a wo kg oup is es ing and compa ing he di e en me hods a ailable o quan i y dese dus con ibu ion o PM concen a ions, iden i ying hei ad an ages and limi a ions, wi h he pu pose o ha monizing his assessmen among he Eu opean Union. In addi ion o a mosphe ic moni o ing, modelling is an impo an ool o unde s and he beha io o species and pollu an s in he a mosphe e. Se e al nume ical models a e able o es ima e dus sou ces and sinks in he a mosphe e, o simula e anspo and ul ima ely con ibu ions o PM concen a ions, such as he BSC-DREAM8b [ 11 ], SKIRON [ 12 ], CHIMERE [ 13 ] and NMMB/BSC-Dus [ 14 ] models, among many o he s. Mo eo e , dus is now included in many ope a ional o ecas sys ems, and se e al dus - ela ed p oduc s a e a ailable o No he n A ica and Eu ope, o example, due o he wo k o he Sand and Dus S o m Wa ning Ad iso y and Assessmen Sys em (SDS-WAS), es ablished by he Wo ld Me eo ological O ganiza ion (WMO). In his s udy, we aim o assess dese dus con ibu ion o PM10 and PM2.5 concen a ions o e Po ugal o he yea 2016. We will apply Escude o’s s a is ical app oach (P40) o one yea o pa icula e ma e concen a ions o e Po ugal and compa e he esul s o model es ima es pe o med using he Po uguese ope a ional ai quali y o ecas sys em (based on WRF-CHIMERE simula ions). Mo eo e , A mosphe e 2020,11, 134 3 o 21 he calcula ion o he numbe o days o exceedance o he PM10 daily limi alue, be o e and a e he sub ac ion o he A ican dus con ibu ions, will be pe o med o he di e en es ima es. 2. Me hodology In his s udy, he P40 me hod, o icially me hod adop ed by he Po uguese En i onmen Agency, is applied o ai quali y da a o e Po ugal conside ing a one-yea pe iod (2016). In addi ion o he P40 me hod, a model-based me hodology is applied, and he esul s o bo h app oaches a e compa ed. In he model-based me hodology, he in luence o A ican dus ou b eaks on PM le els is quan i ied using mine al dus concen a ions simula ed by he WRF-CHIMERE modelling sys em, which esul s a e also assessed. 2.1. The Measu emen -Based P40 Me hod The measu emen -based P40 me hod is a s a is ical me hodology applied o PM da a se ies o calcula e daily A ican dus con ibu ions o PM10 and PM2.5. I is based on he applica ion o a 30-day mo ing 40 h pe cen ile o he PM10 o PM2.5 da a se ies in u al backg ound s a ions, a e excluding hose days impac ed by A ican dus . The days impac ed by A ican dus should be iden i ied based on a ious da a and in o ma ion, including sa elli e obse a ions, model back- ajec o ies and dus o ecas models. In days a ec ed by A ican dus , he 30-day mo ing 40 h pe cen ile is assumed o be he heo e ical backg ound concen a ion o PM wi hou A ican dus inpu . The e o e, he A ican dus con ibu ion is ob ained by he di e ence be ween he expe imen al PM10 o PM2.5 concen a ion and he calcula ed 40 h pe cen ile alue. This me hodology was ini ially published conside ing he 30 h pe cen ile [ 6 ]. Fo conse a i e easons, he 40 h pe cen ile was la e adop ed ins ead o he 30 h one [ 5 ]. This calcula ion, done o u al backg ound si es, pe mi s o es ima e he daily con ibu ion o A ican dus o he whole egion. The es ima ed alue may hen be sub ac ed om he daily PM le els obse ed a o he moni o ing s a ions in he same egion. O e Po ugal, six u al backg ound si es a e usually conside ed o he assessmen o A ican dus con ibu ion o PM le els, which a e iden i ied in Table 1. Table 1. Lis o u al backg ound ai quali y moni o ing s a ions selec ed as ep esen a i e by he Po uguese En i onmen Agency (o de ed by la i ude). PM10 and PM2.5 da a comple eness a e included o he yea 2016. Code Name LON LAT Heigh (m) PM10 Da a PM2.5 Da a DRN Dou o No e −7.789 41.370 1086 19.1% 20.8% FUN Fundão−7.300 40.232 473 92.1% 69.7% MOV Mon emo -o-Velho −8.677 40.183 96 79.2% - CHA Chamusca −8.468 39.353 143 98.1% 96.7% TER Te ena −7.398 38.616 187 98.4% 87.7% CER Ce o −7.680 37.312 300 85.5% 88.5% The impac o mine al dus ou b eaks on PM10 and PM2.5 ambien concen a ions is calcula ed o he i e si es Fund ã o (FUN), Mon emo -o-Velho (MOV), Chamusca (CHA), Te ena (TER) and Ce o (CER). Al hough all hose si es a e classi ied as u al egional backg ound, hey a e loca ed in di e en egions; hus, hey may exhibi he in luence o di e en landscapes, land uses, local clima e and opog aphy. Dou o No e (DRN) da a has no been used, due o he low quan i y o a ailable obse a ions (see Table 1) du ing he yea in he analysis. 2.2. The WRF-CHIMERE Modelling Sys em In his wo k, he nume ical simula ions o a mosphe ic physical and chemical p ope ies a e based on he Wea he Resea ch and Fo ecas ing (WRF) me eo ological model [ 15 ], whose ou pu s d i e he CHIMERE chemis y- anspo model [ 16 ], de eloped in F ance a he Dynamic Me eo ology A mosphe e 2020,11, 134 4 o 21 Labo a o y om he Ins i u e Pie e Simon Laplace (IPSL/LMD), F ench Na ional Ins i u e o Indus ial En i onmen and Risks (INERIS) and In e -Uni e si y Labo a o y o A mosphe ic Sys ems (IPSL/LISA). Since he WRF-CHIMERE is he modelling sys em used in he Po uguese ope a ional ai quali y o ecas sys em, i has been ex ensi ely es ed and alida ed. Those exe cises con i med he easonably good skills o he CHIMERE model o ozone and o he gaseous pollu an s concen a ions [ 17 ]. Pa icula e ma e p esen s a highe complexi y han ozone. Many s udies ha e ecognized he di icul y o models o simula e he mass o PM o e Eu ope (e.g., Re e ences [ 18 , 19 ]). As summa ized by Basa e al. [ 20 ], he unde es ima ion o PM10 may be ela ed wi h he lack o dus and esuspended ma e emissions, a possible unde es ima ion o p ima y ca bonaceous pa icles, he inaccu acy o seconda y o ganic ae osol o ma ion, he di icul y in ep esen ing p ima y PM emission om wood bu ning and o he sou ces no conside ed in he emission in en o y, such as pollu an sou ces o e No h A ica and a mo e gene al lack o knowledge on ae osol emo al, dispe sion and anspo p ocesses. The WRF-CHIMERE modelling sys em is applied o he whole o 2016 conside ing h ee nes ed domains, depic ed in Figu e 1. High- esolu ion simula ions we e pe o med, using a coa se domain co e ing Sou he n Eu ope and Saha a Dese wi h a ho izon al esolu ion o 27 × 27 km 2 (CONT27); a second domain co e ing he Ibe ian Peninsula wi h 9 × 9 km 2 o ho izon al esolu ion (IP09); and he inne -mos domain co e ing mainland Po ugal, wi h 3 × 3 km 2 (PT03). The e ical esolu ion in CHIMERE a ies upwa d in a geome ic p og ession. The lowes a mosphe ic laye s a e mo e e ined, since hese laye s a e c i ical o he modelling o bounda y laye con amina ion, pa icula ly in u ban a eas, bu also in ma ine a eas, in o de o model he sea-sal emissions co ec ly, and in a id a eas, o mine al dus emissions [ 21 ]. Al hough he mos equen ly used e ical disc e iza ion consis s o eigh e ical le els [ 16 ], in p esen simula ions, a be e ep esen a ion o he ee oposphe e is employed, using 24 le els up o he 200hPa p essu e le el. The i s model laye , which is close o he su ace, has a hickness o 3 hPa, abou 30m. Wi h his e ical disc e iza ion, we aim o ep oduce he long- ange anspo o dus om No h A ica be e , as sugges ed by Menu e al. [16]. A mosphe e 2020, 11, 134 4 o 23 d i e he CHIMERE chemis y- anspo model [16], de eloped in F ance a he Dynamic Me eo ology Labo a o y om he Ins i u e Pie e Simon Laplace (IPSL/LMD), F ench Na ional Ins i u e o Indus ial En i onmen and Risks (INERIS) and In e -Uni e si y Labo a o y o A mosphe ic Sys ems (IPSL/LISA). Since he WRF-CHIMERE is he modelling sys em used in he Po uguese ope a ional ai quali y o ecas sys em, i has been ex ensi ely es ed and alida ed. Those exe cises con i med he easonably good skills o he CHIMERE model o ozone and o he gaseous pollu an s concen a ions [17]. Pa icula e ma e p esen s a highe complexi y han ozone. Many s udies ha e ecognized he di icul y o models o simula e he mass o PM o e Eu ope (e.g., Re e ences [18,19]). As summa ized by Basa e al. [20], he unde es ima ion o PM10 may be ela ed wi h he lack o dus and esuspended ma e emissions, a possible unde es ima ion o p ima y ca bonaceous pa icles, he inaccu acy o seconda y o ganic ae osol o ma ion, he di icul y in ep esen ing p ima y PM emission om wood bu ning and o he sou ces no conside ed in he emission in en o y, such as pollu an sou ces o e No h A ica and a mo e gene al lack o knowledge on ae osol emo al, dispe sion and anspo p ocesses. The WRF-CHIMERE modelling sys em is applied o he whole o 2016 conside ing h ee nes ed domains, depic ed in Figu e 1. High- esolu ion simula ions we e pe o med, using a coa se domain co e ing Sou he n Eu ope and Saha a Dese wi h a ho izon al esolu ion o 27 × 27 km2 (CONT27); a second domain co e ing he Ibe ian Peninsula wi h 9 × 9 km2 o ho izon al esolu ion (IP09); and he inne -mos domain co e ing mainland Po ugal, wi h 3 × 3 km2 (PT03). The e ical esolu ion in CHIMERE a ies upwa d in a geome ic p og ession. The lowes a mosphe ic laye s a e mo e e ined, since hese laye s a e c i ical o he modelling o bounda y laye con amina ion, pa icula ly in u ban a eas, bu also in ma ine a eas, in o de o model he sea-sal emissions co ec ly, and in a id a eas, o mine al dus emissions [21]. Al hough he mos equen ly used e ical disc e iza ion consis s o eigh e ical le els [16], in p esen simula ions, a be e ep esen a ion o he ee oposphe e is employed, using 24 le els up o he 200hPa p essu e le el. The i s model laye , which is close o he su ace, has a hickness o 3 hPa, abou 30m. Wi h his e ical disc e iza ion, we aim o ep oduce he long- ange anspo o dus om No h A ica be e , as sugges ed by Menu e al. [16]. Figu e 1. WRF-CHIMERE domains. Figu e 1. WRF-CHIMERE domains. A mosphe e 2020,11, 134 5 o 21 A he bounda ies o he ou e mos domain, clima ologies om global model simula ions a e used. The ou pu s om LMDz-INCA (IN e ac ion wi h Chemis y and Ae osols model coupled o he Labo a oi e de M é é o ologie Dynamique Gene al Ci cula ion Model) [ 22 ] a e used o all gaseous and ae osol species, excep o mine al dus . Fo his species, simula ions om he GOCART (Godda d Chemis y Ae osol Radia ion and T anspo ) model a e used [ 23 ]. Fo he nes ed domains, he bounda y condi ions a e upda ed e e y hou using he p e iously pe o med coa se CHIMERE simula ion. Fo he yea 2016, 366 daily uns ha e been pe o med. In each un, he ini ial condi ions a e de ined by he las hou om he p e ious-day model un. In addi ion o he me eo ological ields ( om WRF) and o he chemical bounda y condi ions, he ai quali y modelling sys em mus be ed wi h p ima y pollu an emissions. The main human ac i i ies emissions ( a ic, indus ies and ag icul u e, among o he s) a e de i ed based on da a om he annual UNECE/EMEP emission da abase (Eu opean Moni o ing and E alua ion P og am, di ec ed by he Uni ed Na ions Economic Commission o Eu ope) [ 24 ], ollowing a p ocedu e o spa ial and empo al downscaling. The spa ial alloca ion o su ace emissions on he CHIMERE g id is based on a classi ica ion wi h ou landuse ypes: U ban, c op, wa e o o es . Then, o each g id cell, hou ly es ima es a e de i ed, assuming seasonal ac o s and 24-h p o iles o each coun y and SNAP (Selec ed Nomencla u e o sou ces o Ai Pollu ion) sec o . In addi ion, EMEP pollu an s a e spli in o model species, acco ding o he chemical mechanism selec ed in hese simula ions: The se en in en o y pollu an s (CO, NOx, SOx, NMVOC, NH 3 , PM10 and PM2.5) a e spli in o he 42 educed Melchio species and eigh chemical ope a o s. Na u al emissions es ima es a e also aken in o accoun by he model, namely, biogenic emissions, sea sal and dus . Fo es i e emissions a e no included in he simula ions. The emissions ela ed o he ege a ion a e compu ed online using he MEGAN model (Model o Emissions o Gases and Ae osols om Na u e) [ 25 ]. Sea sal luxes a e calcula ed, acco ding o Monahan e al. [ 26 ]. Dus emission luxes a e calcula ed using he pa ame iza ion o Ma ico ena and Be game i [ 27 ] o sal a ion and he dus p oduc ion model p oposed by Al a o and Gomes [ 28 ] o sandblas ing. This calcula ion equi es land-use da a, which is used o p o ide a dese mask speci ying wha su ace is po en ially e odible. In hese simula ions, he USGS (Uni ed S a es Geological Su ey) da abase, STATSGO-FAO, is employed, and he e odible land-use ype applied as a dese mask o dus luxes calcula ion is ba en soil (whi e sand). 3. Resul s 3.1. How is he Modelling Sys em Pe o ming? The abili y o he WRF-CHIMERE sys em o simula e g ound-le el PM10 and PM2.5 concen a ions is assessed h ough compa ison o model ou pu s wi h concen a ions ou inely obse ed wi hin he Po uguese ai quali y moni o ing ne wo k. This compa ison is pe o med o he i e u al egional backg ound si es p e iously selec ed. The compa ison be ween model esul s and obse a ions, in e ms o PM10 and PM2.5 annual means, is p esen ed in Figu e 2. The backg ound le els a e easonably cap u ed by he model in all moni o ing s a ions, bu TER. Mean di e ences (o bias) be ween modelled and obse ed concen a ions a e app oxima ely 5 µ g · m −3 o PM10 and 2 µ g · m −3 o PM2.5, in CER, CHA, FUN and MOV. TER exhibi s di e ences o 13 µ g · m −3 o PM10 and 8 µ g · m −3 o PM2.5. Among he i e u al backg ound moni o ing s a ions included in his s udy, TER has he highes obse ed PM10 and PM2.5 mean concen a ions. I s PM10 and PM2.5 mean concen a ions a e mo e han 5 µ g · m −3 highe han he ones measu ed in FUN, CHA o CER. The TER moni o ing s a ion is loca ed in Alen ejo In e io , a p edominan ly a id egion, wi h land use and land co e which may con ibu e o pa icula e esuspension and inc ease PM concen a ions in he egion. A mosphe e 2020,11, 134 6 o 21 A mosphe e 2020, 11, 134 6 o 23 (a) (b) Figu e 2. Maps o (a) PM10 and (b) PM2.5 annual mean concen a ions du ing 2016, as modelled by WRF-CHIMERE. Obse ed mean concen a ions a e included wi hin ci cles. To assess model ag eemen wi h obse a ions on a daily basis, Taylo diag ams [29] ela i e o he mean backg ound concen a ions (e.g., daily a e age concen a ions among he i e moni o ing s a ions) a e p esen ed in Figu e 3. The Taylo diag am is one o he mo e use ul me hods o e alua ing model pe o mance, since i p o ides a way o showing simul aneously how h ee complemen a y model pe o mance s a is ics a y: The co ela ion coe icien ( ), he s anda d de ia ion (sigma) and he oo -mean-squa e (RMS) e o . Co ela ion be ween modelled concen a ions and obse a ions is 0.6 o PM10 and 0.4 o PM2.5. This means ha on a yea ly basis, CHIMERE ep oduces be e he empo al a iabili y ac oss Po ugal o PM10 han PM2.5. S anda d de ia ion and oo mean squa e e o a e, in absolu e alues, highe o PM10 han PM2.5, which is ela ed o he highe PM10 daily alues in compa ison o PM2.5. (a) (b) Figu e 3. Taylo diag ams ela i e o he WRF-CHIMERE pe o mance simula ing g ound-le el (a) PM10 and (b) PM2.5 concen a ions. Concen ic dashed lines emana ing om he ‘obse ed’ poin show he alue o he RMS e o . Va iabili y in he obse a ions and model esul s is ep esen ed by Figu e 2. Maps o ( a ) PM10 and ( b ) PM2.5 annual mean concen a ions du ing 2016, as modelled by WRF-CHIMERE. Obse ed mean concen a ions a e included wi hin ci cles. To assess model ag eemen wi h obse a ions on a daily basis, Taylo diag ams [ 29 ] ela i e o he mean backg ound concen a ions (e.g., daily a e age concen a ions among he i e moni o ing s a ions) a e p esen ed in Figu e 3. The Taylo diag am is one o he mo e use ul me hods o e alua ing model pe o mance, since i p o ides a way o showing simul aneously how h ee complemen a y model pe o mance s a is ics a y: The co ela ion coe icien ( ), he s anda d de ia ion (sigma) and he oo -mean-squa e (RMS) e o . Co ela ion be ween modelled concen a ions and obse a ions is 0.6 o PM10 and 0.4 o PM2.5. This means ha on a yea ly basis, CHIMERE ep oduces be e he empo al a iabili y ac oss Po ugal o PM10 han PM2.5. S anda d de ia ion and oo mean squa e e o a e, in absolu e alues, highe o PM10 han PM2.5, which is ela ed o he highe PM10 daily alues in compa ison o PM2.5. A mosphe e 2020, 11, 134 6 o 23 (a) (b) Figu e 2. Maps o (a) PM10 and (b) PM2.5 annual mean concen a ions du ing 2016, as modelled by WRF-CHIMERE. Obse ed mean concen a ions a e included wi hin ci cles. To assess model ag eemen wi h obse a ions on a daily basis, Taylo diag ams [29] ela i e o he mean backg ound concen a ions (e.g., daily a e age concen a ions among he i e moni o ing s a ions) a e p esen ed in Figu e 3. The Taylo diag am is one o he mo e use ul me hods o e alua ing model pe o mance, since i p o ides a way o showing simul aneously how h ee complemen a y model pe o mance s a is ics a y: The co ela ion coe icien ( ), he s anda d de ia ion (sigma) and he oo -mean-squa e (RMS) e o . Co ela ion be ween modelled concen a ions and obse a ions is 0.6 o PM10 and 0.4 o PM2.5. This means ha on a yea ly basis, CHIMERE ep oduces be e he empo al a iabili y ac oss Po ugal o PM10 han PM2.5. S anda d de ia ion and oo mean squa e e o a e, in absolu e alues, highe o PM10 han PM2.5, which is ela ed o he highe PM10 daily alues in compa ison o PM2.5. (a) (b) Figu e 3. Taylo diag ams ela i e o he WRF-CHIMERE pe o mance simula ing g ound-le el (a) PM10 and (b) PM2.5 concen a ions. Concen ic dashed lines emana ing om he ‘obse ed’ poin show he alue o he RMS e o . Va iabili y in he obse a ions and model esul s is ep esen ed by Figu e 3. Taylo diag ams ela i e o he WRF-CHIMERE pe o mance simula ing g ound-le el ( a ) PM10 and ( b ) PM2.5 concen a ions. Concen ic dashed lines emana ing om he ‘obse ed’ poin show he alue o he RMS e o . Va iabili y in he obse a ions and model esul s is ep esen ed by he s anda d de ia ion, which is measu ed as he adial dis ance om he o igin o he plo . The co ela ion coe icien is shown on he a c and poin s ha lie closes o he x-axis ha e he highes co ela ion. A mosphe e 2020,11, 134 7 o 21 The esul s ob ained o he model pe o mance ega ding PM10 concen a ions in his s udy a e mode a ely be e han wha was p e iously achie ed by his modelling se up. A p e ious s udy by Russo e al. [30] shows a co ela ion coe icien o 0.4 o u al si es. 3.2. Assessmen o Dus Con ibu ion o PM10 and PM2.5 in Po ugal, Du ing 2016 3.2.1. Using he Measu emen -Based Me hodology Du ing 2016, he Po uguese En i onmen Agency iden i ied in mainland Po ugal 93 days (mo e de ails in Figu e A1) as days impac ed by A ican dus [ 31 ]. This iden i ica ion was based on he dus o ecas models BSC-DREAM8b [ 11 ] and SKIRON [ 12 ], and on he HYSPLIT (Hyb id Single-Pa icle Lag angian In eg a ed T ajec o y) model [ 32 ] Fo each day, we applied he P40 me hodology in o de o calcula e he egional A ican dus con ibu ions o PM10 and PM2.5 concen a ions (AD10 and AD2.5, espec i ely) o e Po ugal. Resul s a e p esen ed in Figu es 4and 5, o PM10 and PM2.5, espec i ely. A ican dus con ibu ion o PM le els is e iden du ing a speci ic episode in Feb ua y, du ing se e al episodes om mid-June o mid-Augus , and also du ing Oc obe and No embe . Acco ding o he P40 me hod, he highes daily con ibu ion o PM10 occu du ing he Feb ua y episode (on 22 Feb ua y in TER (AD10 =161 µ g · m −3 ) and CHA (AD10 =76 µ g · m −3 ); on 21 Feb ua y in CER, which has no measu emen s a ailable du ing he 22nd) o on 13 Augus in he Cen e egion (FUN (AD10 =82 µ g · m −3 ) and MOV (AD10 =77 µ g · m −3 ) moni o ing s a ions). No e ha du ing hese days, he con ibu ion o A ican dus (AD10), only, is al eady su icien o su pass he PM10 daily limi alue de ined by he Eu opean Ai Quali y Di ec i e o he p o ec ion o human heal h (50 µg·m−3). Fo each s a ion, he days when he A ican dus con ibu ion is highe han a ce ain h eshold (2 µ g · m −3 o PM10 and 1 µ g · m −3 o PM2.5) we e il e ed, and a daily mean con ibu ion was es ima ed, aking in o accoun hose days only. Resul s a e p esen ed in Table 2 o PM10 and in Table 3 o PM2.5. Table 2. The numbe o days when he A ican dus con ibu ion o pa icles wi h an ae odynamic diame e smalle han 10 µ m (AD10) is highe han 2 µ g · m −3 and hei mean A ican dus concen a ions (calcula ed acco ding o he P40 me hod). The day o he yea 2016 wi h he maximum A ican dus con ibu ion is also iden i ied, as well as he con ibu ion i sel . Moni o ing S a ion No. Days AD10 ≥2µg·m−3 Daily Mean Con ibu ion (µg·m−3) Max Daily Con ibu ion (µg·m−3) Day o Occu ence FUN 71 15.6 81.8 13 Augus MOV 48 15.5 77.0 13 Augus CHA 70 16.3 76.1 22 Feb ua y TER 79 18.9 160.9 22 Feb ua y CER 68 12.2 54.0 21 Feb ua y Table 3. The numbe o days when he A ican dus con ibu ion o pa icles wi h an ae odynamic diame e smalle han 2.5 µ m (AD2.5) is highe han 1 µ g · m −3 and hei mean A ican dus concen a ions (calcula ed acco ding o he P40 me hod). The day o he yea 2016 wi h he maximum A ican dus con ibu ion is also iden i ied, as well as he con ibu ion i sel . Moni o ing S a ion No. Days AD2.5 ≥1µg·m−3 Daily Mean Con ibu ion (µg·m−3) Max Daily Con ibu ion (µg·m−3) Day o Occu ence FUN 48 10.7 56.2 13 Augus CHA 72 8.8 44.4 13 Augus TER 76 8.5 21.7 23 June CER 63 8.3 99.7 22 Feb ua y A mosphe e 2020,11, 134 8 o 21 A mosphe e 2020, 11, 134 8 o 23 . Figu e 4. PM10 daily concen a ions (black line) and na u al con ibu ion as es ima ed by he o icial me hod adop ed by he Po uguese En i onmen Agency (g een ba s). Figu e 4. PM10 daily concen a ions (black line) and na u al con ibu ion as es ima ed by he o icial me hod adop ed by he Po uguese En i onmen Agency (g een ba s). The highes daily mean con ibu ion o A ican dus o PM10 concen a ions occu s in TER, which also has he highes numbe o days classi ied as impac ed by A ican dus wi h a con ibu ion highe han 2 µ g · m −3 . In his s a ion, loca ed in inland Alen ejo, acco ding o he P40 me hodology, A ican dus con ibu es, on a e age, wi h nea ly 19 µ g · m −3 o pa icles wi h an ae odynamic diame e smalle han 10 µ m, du ing he mos a ec ed 79 days by dese dus . The ac ha he daily mean con ibu ion is lowe in CER (loca ed in Alga e) han in TER should be in e p e ed ca e ully, since CER has no alid measu emen s du ing 22 Feb ua y. A mosphe e 2020,11, 134 9 o 21 A mosphe e 2020, 11, 134 9 o 23 Figu e 5. PM2.5 daily concen a ions (black line) and na u al con ibu ion as es ima ed by he o icial me hod adop ed by he Po uguese En i onmen Agency (g een ba s). Fo each s a ion, he days when he A ican dus con ibu ion is highe han a ce ain h eshold (2 µg.m−3 o PM10 and 1 µg.m−3 o PM2.5) we e il e ed, and a daily mean con ibu ion was es ima ed, aking in o accoun hose days only. Resul s a e p esen ed in Table 2 o PM10 and in Table 3 o PM2.5. Table 2. The numbe o days when he A ican dus con ibu ion o pa icles wi h an ae odynamic diame e smalle han 10 µm (AD10) is highe han 2 µg.m−3 and hei mean A ican dus concen a ions (calcula ed acco ding o he P40 me hod). The day o he yea 2016 wi h he maximum A ican dus con ibu ion is also iden i ied, as well as he con ibu ion i sel . Moni o ing S a ion No. Days AD10 ≥ 2 µg.m−3 Daily Mean Con ibu ion (µg.m−3) Max Daily Con ibu ion (µg.m−3) Day o Occu ence FUN 71 15.6 81.8 13 Augus MOV 48 15.5 77.0 13 Augus CHA 70 16.3 76.1 22 Feb ua y TER 79 18.9 160.9 22 Feb ua y CER 68 12.2 54.0 21 Feb ua y Figu e 5. PM2.5 daily concen a ions (black line) and na u al con ibu ion as es ima ed by he o icial me hod adop ed by he Po uguese En i onmen Agency (g een ba s). TER moni o ing s a ion also has he highes numbe o days classi ied as days impac ed by A ican dus when he ac ion below 2.5 µ m is conside ed and a con ibu ion highe han 1 µ g · m −3 is imposed. Howe e , i egis e s he lowes maximum daily con ibu ion (abou 22 µ g · m −3 ), which occu s on 23 June. Du ing 22 Feb ua y, al hough a con ibu ion o 161 µ g · m −3 is es ima ed a his s a ion o he pa icles wi h an ae odynamic diame e smalle han 10 µ m, a con ibu ion o 16 µ g · m −3 only is es ima ed o he ac ion below 2.5 µ m. Fo his ac ion (AD2.5), CER moni o ing s a ion has he highes maximum daily con ibu ion (AD2.5 =100 µ g · m −3 ), which is egis e ed du ing 22 Feb ua y. The e is no PM10 a ailable da a du ing his day a CER. Howe e , AD2.5 es ima ed o his day is highe han he maximum daily con ibu ion o PM10 es ima ed a his s a ion du ing he whole yea (AD10 =54 µg·m−3, es ima ed o 21 Feb ua y). In e ms o mean alues o A ican dus con ibu ion o PM2.5 ( aking in o accoun days wi h a con ibu ion highe han 1 µ g · m −3 only), FUN has he highes mean alue (abou 11 µ g · m −3 ). Howe e , his ac mus be aken ca e ully, since FUN has he lowes numbe o days wi h AD2.5 ≥ 1 µ g · m −3 , and mean alues a e calcula ed o hose days only. This happens because, as illus a ed in Figu e 5, FUN only has PM2.5 da a un il Sep embe (al hough he e a e alid obse a ions o PM10 du ing he es o he yea ). Howe e , acco ding o Re e ence [ 31 ], and as seen in Figu e A1, he e a e s ill 31 po en ial days om Oc obe o Decembe ha could add o he 48 days iden i ied o AD2.5. A mosphe e 2020,11, 134 16 o 21 A mosphe e 2020, 11, 134 17 o 23 . Figu e 10. Compa ison be ween he A ican dus con ibu ions o PM10 ob ained by he wo di e en me hodologies du ing 2016. Red poin s co espond o he pe iod 8–15 Augus . Pea son co ela ion alues a e p esen ed, and we e calcula ed excluding he ed poin s. Figu e 11. Compa ison be ween he A ican dus con ibu ions o PM2.5 ob ained by he wo di e en me hodologies du ing 2016. Red poin s co espond o he pe iod 8–15 Augus . Pea son co ela ion alues a e p esen ed, and we e calcula ed excluding he ed poin s. The main di e ences in he esul s ob ained wi h he wo me hodologies happen in he i s hal o Augus . Du ing his pe iod, a high con ibu ion o A ican dus o he pa icula e ma e o e Po ugal is es ima ed by he P40 me hodology. In he moni o ing s a ions FUN and MOV, in he Cen e o Po ugal, he maximum daily con ibu ion o A ican dus o PM10 le els o e he whole yea 2016, is es ima ed du ing his pe iod (AD10 = 82 and 77 µg.m −3 du ing 13 Augus , see Table 2). Figu e 11. Compa ison be ween he A ican dus con ibu ions o PM2.5 ob ained by he wo di e en me hodologies du ing 2016. Red poin s co espond o he pe iod 8–15 Augus . Pea son co ela ion alues a e p esen ed, and we e calcula ed excluding he ed poin s. PM ob ained om model esul s co esponds o he PM concen a ion o he i s model laye , which is close o he su ace and has a hickness o abou 30 m. Al hough his heigh is much highe han wha is ypically employed in measu emen s, his is how he model se up is ecommended when pe o ming hese simula ions. Addi ionally, al hough he model e ical esolu ion (and he i s model le el heigh ) can ha e a s ong impac on su ace concen a ions, ou es s show he a iabili y and subsequen unce ain y o he esul s in oduced wi h his disc epancy be ween heigh le els is no he main cause o he di e ences be ween he measu ed and modelled esul s. The main di e ences in he esul s ob ained wi h he wo me hodologies happen in he i s hal o Augus . Du ing his pe iod, a high con ibu ion o A ican dus o he pa icula e ma e o e Po ugal is es ima ed by he P40 me hodology. In he moni o ing s a ions FUN and MOV, in he Cen e o Po ugal, he maximum daily con ibu ion o A ican dus o PM10 le els o e he whole yea 2016, is es ima ed du ing his pe iod (AD10 =82 and 77 µ g · m −3 du ing 13 Augus , see Table 2). The same happens ega ding he con ibu ion o dese dus o PM2.5 concen a ions, wi h a con ibu ion o 56 and 44 µ g · m −3 es ima ed by he P40 me hod (see Table 3). Du ing hese days, model esul s indica e, howe e , a small in luence o long- ange anspo om No h A ican dese s o he PM10 le els in he sou h o Po ugal only (CER moni o ing s a ion, whe e a con ibu ion o abou 10 µ g · m −3 o dus was calcula ed by he model on 15 Augus ). In he emaining egions and days, he con ibu ion o A ican dus o PM le els o e Po ugal is almos negligible. Du ing he i s hal o Augus , Po ugal ba led nume ous wild i es, mos ly in he Cen e and No h o he coun y. Sa elli es began o de ec hose la ge numbe s o i es on 6 Augus . In he ollowing days, he i es g ew mo e nume ous, and he amoun o smoke inc eased d ama ically (Figu e 12), a ec ing se e ely ai quali y. Obse a ions om he AE osol RObo ic NETwo k (AERONET) (no included in his publica ion) show ha on 13 Augus he ae osol loading was domina ed by pa icles less han 1 mic on—which suppo s he claim ha a signi ican p opo ion o he ae osols on ha day we e om o es i es. Con e sely, he da a om he 21s and 22nd Feb ua y show he opposi e. When ex eme e en s, such as o es i es, occu simul aneously wi h he long- ange anspo o dese dus , he P40 me hodology may lead o an o e es ima ion o he con ibu ion o A ican dus o A mosphe e 2020,11, 134 17 o 21 PM le els. Mo eo e , he combina ion o dese dus episodes and o es i es is qui e common du ing he summe in Po ugal. Du ing dese dus episodes, wa m and d y ai masses a e anspo ed o e Po ugal, which a o he occu ence and g ow h o wild i es. Tha si ua ion occu ed du ing Augus 2016 and in Oc obe 2017 du ing he se e e o es i es in he Cen e and No h o Po ugal linked wi h he Hu icane Ophelia [38]. Acco ding o he Eu opean guidelines, wild i es should also be conside ed a na u al cause whose con ibu ion o PM10 and PM2.5 a mosphe ic le els may be sub ac ed when assessing compliance wi h ai quali y limi alues. In his sense, he usage o he P40 me hodology is no p oblema ic. Howe e , o assessmen pu poses, i is impo an o iden i y he cause o he e en . Fo ha , he cu en me hod does no allow o dis inguish be ween na u al causes (o ex ao dina y an h opogenic e en s). The au ho s ecommend complemen ing he cu en me hod wi h o he sou ces o da a, including he bes a ailable model simula ions. The inclusion o addi ional ou inely eco ded obse a ions, including AERONET and lida da a, could also assis wi h he iden i ica ion o na u al sou ces con ibu ing he mos o he ae osol load. A mosphe e 2020, 11, 134 18 o 23 The same happens ega ding he con ibu ion o dese dus o PM2.5 concen a ions, wi h a con ibu ion o 56 and 44 µg.m−3 es ima ed by he P40 me hod (see Table 3). Du ing hese days, model esul s indica e, howe e , a small in luence o long- ange anspo om No h A ican dese s o he PM10 le els in he sou h o Po ugal only (CER moni o ing s a ion, whe e a con ibu ion o abou 10 µg.m−3 o dus was calcula ed by he model on 15 Augus ). In he emaining egions and days, he con ibu ion o A ican dus o PM le els o e Po ugal is almos negligible. Du ing he i s hal o Augus , Po ugal ba led nume ous wild i es, mos ly in he Cen e and No h o he coun y. Sa elli es began o de ec hose la ge numbe s o i es on 6 Augus . In he ollowing days, he i es g ew mo e nume ous, and he amoun o smoke inc eased d ama ically (Figu e 12), a ec ing se e ely ai quali y. Obse a ions om he AE osol RObo ic NETwo k (AERONET) (no included in his publica ion) show ha on 13 Augus he ae osol loading was domina ed by pa icles less han 1 mic on—which suppo s he claim ha a signi ican p opo ion o he ae osols on ha day we e om o es i es. Con e sely, he da a om he 21s and 22nd Feb ua y show he opposi e. (a) 7 Aug, 2016 13:10 UTC (b) 9 Aug, 2016 11:20 UTC (c) 9 Aug, 2016 12:55 UTC (d) 10 Aug , 2016 12:00 UTC Figu e 12. Con . A mosphe e 2020,11, 134 18 o 21 A mosphe e 2020, 11, 134 19 o 23 (e) 12 Aug, 2016 11:50 UTC ( ) 13 Aug, 2016 13:25 UTC Figu e 12. Sa elli e images cap u ed om 6 o 13 Augus 2016 by Te a/MODIS and SNPP/VIIRS, showing ac i ely bu ning a eas (ou lined in ed) and associa ed smoke plumes. When ex eme e en s, such as o es i es, occu simul aneously wi h he long- ange anspo o dese dus , he P40 me hodology may lead o an o e es ima ion o he con ibu ion o A ican dus o PM le els. Mo eo e , he combina ion o dese dus episodes and o es i es is qui e common du ing he summe in Po ugal. Du ing dese dus episodes, wa m and d y ai masses a e anspo ed o e Po ugal, which a o he occu ence and g ow h o wild i es. Tha si ua ion occu ed du ing Augus 2016 and in Oc obe 2017 du ing he se e e o es i es in he Cen e and No h o Po ugal linked wi h he Hu icane Ophelia [38]. Acco ding o he Eu opean guidelines, wild i es should also be conside ed a na u al cause whose con ibu ion o PM10 and PM2.5 a mosphe ic le els may be sub ac ed when assessing compliance wi h ai quali y limi alues. In his sense, he usage o he P40 me hodology is no p oblema ic. Howe e , o assessmen pu poses, i is impo an o iden i y he cause o he e en . Fo ha , he cu en me hod does no allow o dis inguish be ween na u al causes (o ex ao dina y an h opogenic e en s). The au ho s ecommend complemen ing he cu en me hod wi h o he sou ces o da a, including he bes a ailable model simula ions. The inclusion o addi ional ou inely eco ded obse a ions, including AERONET and lida da a, could also assis wi h he iden i ica ion o na u al sou ces con ibu ing he mos o he ae osol load. 4. Summa y and Conclusions The impo ance o mine al dus con ibu ion o he o al ae osol mass obse ed in Po ugal du ing 2016 was es ima ed by wo di e en app oaches: The o icial me hod adop ed by he Po uguese En i onmen Agency ( he P40 me hod) and a model-based me hodology based on WRF- CHIMERE simula ions. The esul s ob ained by bo h me hods o emo e a eas o Po ugal we e compa ed. Al hough he P40 me hod is based on mo ing a e age o measu emen s, i elies on dus models o he selec ion o days iden i ied as impac ed by A ican dus . In Po ugal, he Po uguese En i onmen Agency bases his selec ion in wo dus models: The BSC-DREAM8b and he SKIRON. Then, he quan i ica ion o he dese dus con ibu ion in e ms o pa icles wi h ae odynamic diame e s smalle han 10 and 2.5 µm is based on measu ed da a only. In gene al, A ican dus con ibu ions ob ained wi h he P40 me hodology a e highe han he ones simula ed by WRF-CHIMERE. Con ibu ions o PM10 concen a ions ange om 0 o 90 µg.m−3 in mos o he egions and days (TER is an excep ion, as a con ibu ion highe han 150 µg.m−3 is es ima ed o 22 Feb ua y. Du ing his day, he la ges di e ence be ween es ima es occu , o mo e han 100 µg.m−3—bo h implying, howe e , he su pass o he legisla ion. CER, loca ed in he Figu e 12. Sa elli e images cap u ed om 6 o 13 Augus 2016 by Te a/MODIS and SNPP/VIIRS, showing ac i ely bu ning a eas (ou lined in ed) and associa ed smoke plumes. 4. Summa y and Conclusions The impo ance o mine al dus con ibu ion o he o al ae osol mass obse ed in Po ugal du ing 2016 was es ima ed by wo di e en app oaches: The o icial me hod adop ed by he Po uguese En i onmen Agency ( he P40 me hod) and a model-based me hodology based on WRF-CHIMERE simula ions. The esul s ob ained by bo h me hods o emo e a eas o Po ugal we e compa ed. Al hough he P40 me hod is based on mo ing a e age o measu emen s, i elies on dus models o he selec ion o days iden i ied as impac ed by A ican dus . In Po ugal, he Po uguese En i onmen Agency bases his selec ion in wo dus models: The BSC-DREAM8b and he SKIRON. Then, he quan i ica ion o he dese dus con ibu ion in e ms o pa icles wi h ae odynamic diame e s smalle han 10 and 2.5 µm is based on measu ed da a only. In gene al, A ican dus con ibu ions ob ained wi h he P40 me hodology a e highe han he ones simula ed by WRF-CHIMERE. Con ibu ions o PM10 concen a ions ange om 0 o 90 µ g · m −3 in mos o he egions and days (TER is an excep ion, as a con ibu ion highe han 150 µ g · m −3 is es ima ed o 22 Feb ua y. Du ing his day, he la ges di e ence be ween es ima es occu , o mo e han 100 µ g · m −3 —bo h implying, howe e , he su pass o he legisla ion. CER, loca ed in he sou he nmos egion o Po ugal, has no measu emen s a ailable du ing his day). Excluding 22 Feb ua y (in CER), A ican dus con ibu ions up o 30 µg·m−3a e ound o PM2.5 le els. When o he ex eme e en s, such as o es i es, occu simul aneously wi h long- ange anspo o dese dus , his me hodology may classi y he e en as a dese dus e en when ha is no he main ac o a ec ing PM concen a ions (e.g., he con ibu ion o A ican dus o PM le els is o e es ima ed). Mo eo e , he combina ion o dese dus episodes and o es i es is qui e common du ing he summe in Po ugal, as happened in Augus 2016. The e o e, we ecommend complemen ing he P40 me hodology wi h o he sou ces o da a, including he bes a ailable model simula ions, in o de o allow he iden i ica ion o he cause o he e en . The esul s p esen ed in his s udy con i m ha dus is an abundan ype o na u al a mosphe ic ae osol in he plane a y bounda y laye o e Po ugal. Acco ding o ou esul s, dese dus episodes we e qui e equen du ing 2016. Thei in ensi y was a iable, wi h a leas wo e en s con ibu ing o exceedances o he PM10 daily limi alue de ined in he Ai Quali y Di ec i e. Those se e e episodes ook place du ing 21–22 Feb ua y and 27–28 Oc obe 2016, and ha e a ec ed he ai quali y in all he egions s udied in his wo k. A mosphe e 2020,11, 134 19 o 21 Au ho Con ibu ions: Concep ualiza ion, C.G. and O.T.; Da a cu a ion, C.G.; Fo mal analysis, C.G.; Funding acquisi ion, C.G. and O.T.; Me hodology, C.G. and A.M.; P ojec adminis a ion, O.T.; Resou ces, C.B.; So wa e, C.G. and A.M.; Supe ision, C.P., J.M.B. and O.T.; Visualiza ion, C.G.; W i ing—o iginal d a , C.G.; W i ing— e iew and edi ing, C.P., A.M., M.R., A.P.F. and O.T. All au ho s ha e ead and ag eed o he published e sion o he manusc ip . Funding: This esea ch was unded by FCT – Po uguese Founda ion o Science and Technology and FEDER (wi hin he PT2020 Pa ne ship Ag eemen and Compe e 2020), g an numbe s UID/AMB/50017 - POCI- 01-0145-FEDER-007638 (CESAM associa ed labo a o y), MIT-EXPL/IRA/0023/2017 (ISY-AIR esea ch p ojec ) and POCI-01-0145-FEDER-029374 (ARTUR esea ch p ojec ). Acknowledgmen s: The au ho s acknowledge he Po uguese En i onmen Agency (APA) and he Regional Coo dina ion and De elopmen Commissions (CCDRs) o hei e o in es ablishing and main aining he ai quali y moni o ing si es used in his in es iga ion. Con lic s o In e es : The au ho s decla e no con lic o in e es . Appendix A A mosphe e 2020, 11, 134 20 o 23 sou he nmos egion o Po ugal, has no measu emen s a ailable du ing his day). Excluding 22 Feb ua y (in CER), A ican dus con ibu ions up o 30 µg.m−3 a e ound o PM2.5 le els. When o he ex eme e en s, such as o es i es, occu simul aneously wi h long- ange anspo o dese dus , his me hodology may classi y he e en as a dese dus e en when ha is no he main ac o a ec ing PM concen a ions (e.g., he con ibu ion o A ican dus o PM le els is o e es ima ed). Mo eo e , he combina ion o dese dus episodes and o es i es is qui e common du ing he summe in Po ugal, as happened in Augus 2016. The e o e, we ecommend complemen ing he P40 me hodology wi h o he sou ces o da a, including he bes a ailable model simula ions, in o de o allow he iden i ica ion o he cause o he e en . The esul s p esen ed in his s udy con i m ha dus is an abundan ype o na u al a mosphe ic ae osol in he plane a y bounda y laye o e Po ugal. Acco ding o ou esul s, dese dus episodes we e qui e equen du ing 2016. Thei in ensi y was a iable, wi h a leas wo e en s con ibu ing o exceedances o he PM10 daily limi alue de ined in he Ai Quali y Di ec i e. Those se e e episodes ook place du ing 21–22 Feb ua y and 27–28 Oc obe 2016, and ha e a ec ed he ai quali y in all he egions s udied in his wo k. Au ho Con ibu ions: Concep ualiza ion, C.G. and O.T.; Da a cu a ion, C.G.; Fo mal analysis, C.G.; Funding acquisi ion, C.G. and O.T.; Me hodology, C.G. and A.M.; P ojec adminis a ion, O.T.; Resou ces, C.B.; So wa e, C.G. and A.M.; Supe ision, C.P., J.M.B. and O.T.; Visualiza ion, C.G.; W i ing—o iginal d a , C.G.; W i ing— e iew and edi ing, C.P., A.M., M.R., A.P.F. and O.T. All au ho s ha e ead and ag eed o he published e sion o he manusc ip . Funding: This esea ch was unded by FCT – Po uguese Founda ion o Science and Technology and FEDER (wi hin he PT2020 Pa ne ship Ag eemen and Compe e 2020), g an numbe s UID/AMB/50017 - POCI- 01- 0145-FEDER-007638 (CESAM associa ed labo a o y), MIT-EXPL/IRA/0023/2017 (ISY-AIR esea ch p ojec ) and POCI-01-0145-FEDER-029374 (ARTUR esea ch p ojec ). Acknowledgmen s: The au ho s acknowledge he Po uguese En i onmen Agency (APA) and he Regional Coo dina ion and De elopmen Commissions (CCDRs) o hei e o in es ablishing and main aining he ai quali y moni o ing si es used in his in es iga ion. Con lic s o In e es : The au ho s decla e no con lic o in e es . Appendix A Figu e A1. Days iden i ied as impac ed by A ican dus by he Po uguese En i onmen Agency, du ing 2016. Figu e A1. Days iden i ied as impac ed by A ican dus by he Po uguese En i onmen Agency, du ing 2016. Re e ences 1. Cha ou sidou, S.E.; Kopanakis, I.; Lagou a dos, K.; Mihalopoulos, N.; Tø se h, K.; Laza idis, M. PM10 le els a u ban, subu ban, and backg ound loca ions in he eas e n Medi e anean: Local e sus egional sou ces wi h emphasis on A ican dus . Ai Qual. A mos. Heal h 2019,12, 1359–1371. [C ossRe ] 2. Gama, C.; Tchepel, O.; Baldasano, J.M.; Basa , S.; Fe ei a, J.; Pio, C.; Ca doso, J.; Bo ego, C. Seasonal pa e ns o Saha an dus o e Cape Ve de-a combined app oach using obse a ions and modelling. Tellus B 2015,67. [C ossRe ] 3. Ba mpadimos, I.; Kelle , J.; Ode bolz, D.; Hueglin, C.; P é ô , A.S.H. One decade o pa allel ine (PM2.5) and coa se (PM10–PM2.5) pa icula e ma e measu emen s in Eu ope: T ends and a iabili y. A mos. Chem. Phys. 2012,12, 3189–3203. [C ossRe ] 4. Gama, C.; Mon ei o, A.; Pio, C.; Mi anda, A.I.; Baldasano, J.M.; Tchepel, O. Tempo al pa e ns and ends o pa icula e ma e o e Po ugal: A long- e m analysis o backg ound concen a ions. Ai Qual. A mos. Heal h 2018,11, 397–407. [C ossRe ] A mosphe e 2020,11, 134 20 o 21 5. Eu opean Union. Es ablishing Guidelines o Demons a ion and Subs ac ion o Exceedances A ibu able o Na u al Sou ces Unde he Di ec i e 2008/50/EC on Ambien Ai Quali y and Cleane Ai o Eu ope. In Eu opean Comission Comission S a Wo king Pape ; Eu opean Union: B ussels, Belgium, 2011. 6. Escude o, M.; Que ol, X.; Pey, J.; Alas uey, A.; P é ez, N.; Fe ei a, F.; Alonso, S.; Rod í guez, S.; Cue as, E. A me hodology o he quan i ica ion o he ne A ican dus load in ai quali y moni o ing ne wo ks. A mos. En i on. 2007,41, 5516–5524. [C ossRe ] 7. Gano , E.; S upp, A.; Alpe , P. A me hod o de e mine he e ec o mine al dus ae osols on ai quali y. A mos. En i on. 2009,43, 5463–5468. [C ossRe ] 8. Viana, M.; Sal ado , P.; A í ñano, B.; Que ol, X.; Alas uey, A.; Pey, J.; La z, A.J.; Cabañas, M.; Mo eno, T.; Ga c í a Dos San os, S.; e al. Assessing he pe o mance o me hods o de ec and quan i y a ican dus in ai bo ne pa icula es. En i on. Sci. Technol. 2010,44, 8814–8820. [C ossRe ] 9. G ó mez-Losada, A.; Pi es, J.C.M.; Pino-Mej í as, R. Time se ies clus e ing o es ima ing pa icula e ma e con ibu ions and i s use in quan i ying impac s om dese s. A mos. En i on. 2015 ,117, 271–281. [C ossRe ] 10. Ba naba, F.; Bolignano, A.; Di Libe o, L.; Mo elli, M.; Luca elli, F.; Na a, S.; Pe ino, C.; Canepa i, S.; Basa , S.; Cos abile, F.; e al. Dese dus con ibu ion o PM10 loads in I aly: Me hods and ecommenda ions add essing he ele an Eu opean Commission Guidelines in suppo o he Ai Quali y Di ec i e 2008/50. A mos. En i on. 2017,161, 288–305. [C ossRe ] 11. Basa , S.; P é ez, C.; Nicko ic, S.; Cue as, E.; Baldasano, J.M. De elopmen and e alua ion o he BSC- DREAM8b dus egional model o e No he n A ica, he Medi e anean and he Middle Eas . Tellus B 2012 , 64. [C ossRe ] 12. Kallos, G.; Nicko ic, S.; Papadopoulos, A.; Jo ic, D.; Kakaliagou, O.; Misi lis, N.; Boukas, L.; Mimikou, N.; Sakella idis, G.; Papageo giou, J.; e al. The egional wea he o ecas ing sys em SKIRON: An o e iew. In P oceedings o he In e na ional Symposium on Regional Wea he P edic ion on Pa allel Compu e En i onmen s, A hens, G eece, 15–17 Oc obe 1997; pp. 109–122. 13. Schmech ig, C.; Ma ico ena, B.; Cha ene , B.; Be game i, G.; Rajo , J.L.; Coman, A. Simula ion o he mine al dus con en o e Wes e n A ica om he e en o he annual scale wi h he CHIMERE-DUST model. A mos. Chem. Phys. 2011,11, 7185–7207. [C ossRe ] 14. P é ez, C.; Haus ein, K.; Janjic, Z.; Jo ba, O.; Huneeus, N.; Baldasano, J.M.; Black, T.; Basa , S.; Nicko ic, S.; Mille , R.L.; e al. An online mine al dus ae osol model o meso o global scales: Model desc ip ion, annual simula ions and e alua ion. A mos. Chem. Phys. 2011,11, 13001–13027. [C ossRe ] 15. Skama ock, W.C.; Klemp, J.B.; Dudhia, J.; Gill, D.O.; Ba ke , M.; Duda, K.G.; Huang, X.Y.; Wang, W.; Powe s, J.G. A Desc ip ion o he Ad anced Resea ch WRF Ve sion 3; Technical Repo ; Na ional Cen e o A mosphe ic Resea ch: Boulde , CO, USA, 2008. 16. Menu , L.; Bessagne , B.; Kh o os yano , D.; Beekmann, M.; Blond, N.; Cole e, A.; Coll, I.; Cu ci, G.; Fo e , G.; Hodzic, A.; e al. CHIMERE 2013: A model o egional a mosphe ic composi ion modelling. Geosci. Model. De . 2013,6, 981–1028. [C ossRe ] 17. Mon ei o, A.; Mi anda, A.I.; Bo ego, C.; Vau a d, R. Ai quali y assessmen o Po ugal. Sci. To al En i on. 2007,373, 22–31. [C ossRe ] [PubMed] 18. Ma hias, V. The ae osol dis ibu ion in Eu ope de i ed wi h he Communi y Mul iscale Ai Quali y (CMAQ) model: Compa ison o nea su ace in si u and sunpho ome e measu emen s. A mos. Chem. Phys. 2008 ,8, 5077–5097. [C ossRe ] 19. Pay, M.T.; Pio , M.; Jo ba, O.; Gass ó , S.; Gonçal es, M.; Basa , S.; Dabdub, D.; Jim é nez-Gue e o, P.; Baldasano, J.M. A ull yea e alua ion o he CALIOPE-EU ai quali y modeling sys em o e Eu ope o 2004. A mos. En i on. 2010,44, 3322–3342. [C ossRe ] 20. Basa , S.; Pay, M.T.; Jo ba, O.; P é ez, C.; Jim é nez-Gue e o, P.; Schulz, M.; Baldasano, J.M. Ae osols in he CALIOPE ai quali y modelling sys em: E alua ion and analysis o PM le els, op ical dep hs and chemical composi ion o e Eu ope. A mos. Chem. Phys. 2012,12, 3363–3392. [C ossRe ] 21. Maille , S.; Menu , L.; Kh o os yano , D.; Vala i, M.; Cou ida , F.; Siou , G.; Tu que y, S.; B ian , R.; Tuccella, P.; Bessagne , B.; e al. Chime e-2017: F om u ban o hemisphe ic chemis y- anspo modeling. Geosci. Model. De . 2017,10, 2397–2423. [C ossRe ] 22. Szopa, S.; Fo e , G.; Menu , L.; Cozic, A. Impac o la ge scale ci cula ion on eu opean summe su ace ozone and consequences o modelling o ecas . A mos. En i on. 2009,43, 1189–1195. [C ossRe ] A mosphe e 2020,11, 134 21 o 21 23. Ginoux, P.; Chin, M.; Tegen, I.; P ospe o, J.; Holben, B.; Dubo ik, O.; Lin, S. Sou ces and dis ibu ions o dus ae osols simula ed wi h he GOCART model. J. Geophys. Res. A mos. 2001,106, 20255–20273. [C ossRe ] 24. EMEP/CEIP 2019. Emissions as Used in EMEP Models. A ailable online: h ps://www.ceip.a /ms/ceip_ home1/ceip_home/webdab_emepda abase/emissions_emepmodels/(accessed on 15 No embe 2019). 25. Guen he , A.; Ka l, T.; Ha ley, P.; Wiedinmye , C.; Palme , P.I.; Ge on, C. Es ima es o global e es ial isop ene emissions using MEGAN (Model o Emissions o Gases and Ae osols om Na u e). A mos. Chem. Phys. 2006,6, 3181–3210. [C ossRe ] 26. Monahan, E.C.; Spiel, D.E.; Da idson, K.L. A Model o Ma ine Ae osol Gene a ion Via Whi ecaps and Wa e Dis up ion. In Oceanic Whi ecaps; Monahan, E.C., Niocaill, G.M., Eds.; Sp inge : Do d ech , The Ne he lands, 1986. 27. Ma ico ena, B.; Be game i, G. Modeling he a mosphe ic dus cycle. 1. Design o a soil-de i ed dus emission scheme. J. Geophys. Res. A mos. 1995,100, 16415–16430. [C ossRe ] 28. Al a o, S.C.; Gomes, L. Modeling mine al ae osol p oduc ion by wind e osion: Emission in ensi ies and ae osol size dis ibu ions in sou ce a eas. J. Geophys. Res. A mos. 2001,106, 18075–18084. [C ossRe ] 29. Taylo , K. Summa izing mul iple aspec s o model pe o mance in a single diag am. J. Geophys. Res. A mos. 2001,106, 7183–7192. [C ossRe ] 30. Russo, M.A.; Gama, C.; Mon ei o, A. How does upg ading an emissions in en o y a ec ai quali y simula ions? Ai Qual. A mos. Heal h 2019,12, 731–741. [C ossRe ] 31. Ag ê ncia Po uguesa do Ambien e. Iden i icaç ã o e A aliaç ã o da Oco ê ncia de E en os de  mbi o Na u al em Po ugal em 2016, Rela ó io Anual; Po uguese En i onmen Agency: Lisboa, Po ugal, 2017; A ailable online: h ps://apambien e.p /_zda a/DAR/A /Rela o io_E en os%20Na u ais_2016_ 20170914.pd (accessed on 15 No embe 2019). 32. D axle , R.; Hess, G. An o e iew o he HYSPLIT_4 modelling sys em o ajec o ies, dispe sion and deposi ion. Aus . Me eo ol. Mag. 1998,47, 295–308. 33. Rod í guez, S.; Que ol, X.; Alas uey, A.; Kallos, G.; Kakaliagou, O. Saha an dus con ibu ion o PM10 and TSP le els in Sou he n and Eas e n Spain. A mos. En i on. 2001,35, 2433–2447. [C ossRe ] 34. Gama, C.; Ribei o, I.; Lange, A.C.; Vogel, A.; Ascenso, A.; Seixas, V.; Elbe n, H.; Bo ego, C.; F iese, E.; Mon ei o, A. Pe o mance assessmen o CHIMERE and EURAD-IM’ dus modules. A mos. Pol. Res. 2019 , 10, 1336–1346. [C ossRe ] 35. Mon ei o, A.; Fe nandes, A.P.; Gama, C.; Bo ego, C.; Tchepel, O. Assessing he mine al dus om No h A ica o e Po ugal egion using BSC-DREAM8b model. A mos. Pol. Res. 2015,6, 70–81. [C ossRe ] 36. Escude o, M.; Cas illo, S.; Que ol, X.; A ila, A.; Ala c ó n, M.; Viana, M.M.; Alas uey, A.; Cue as, E.; Rod í guez, S. We and d y A ican dus episodes o e eas e n Spain. J. Geophys. Res. 2005 ,110, D18S08. [C ossRe ] 37. Rod í guez, S.; Alas uey, A.; Alonso-P é ez, S.; Que ol, X.; Cue as, E.; Ab eu-A onso, J.; Viana, M.; P é ez, N.; Pandol i, M.; de la Rosa, J. T anspo o dese dus mixed wi h No h A ican indus ial pollu an s in he sub opical Saha an Ai Laye . A mos. Chem. Phys. 2011,13, 6663–6685. [C ossRe ] 38. Osbo ne, M.; Mala elle, F.F.; Adam, M.; Buxmann, J.; Sugie , J.; Ma enco, F.; Haywood, J. Saha an dus and biomass bu ning ae osols du ing ex-hu icane Ophelia: Obse a ions om he new UK lida and sun-pho ome e ne wo k. A mos. Chem. Phys. 2019,19, 3557–3578. [C ossRe ] © 2020 by he au ho s. Licensee MDPI, Basel, Swi ze land. This a icle is an open access a icle dis ibu ed unde he e ms and condi ions o he C ea i e Commons A ibu ion (CC BY) license (h p://c ea i ecommons.o g/licenses/by/4.0/).