A mos. Chem. Phys., 25, 14449–14478, 2025
h ps://doi.o g/10.5194/acp-25-14449-2025
© Au ho (s) 2025. This wo k is dis ibu ed unde
he C ea i e Commons A ibu ion 4.0 License.
Resea ch a icle
Towa ds an imp o ed unde s anding o he impac o
clouds and p ecipi a ion on he ep esen a ion o
ae osols o e he Bo eal Fo es in GCMs
Sini Tal inen1,2, Paul Kim3, Emanuele To azzi3, Eemeli Holopainen1,4,5, Roxana C eme 2,a,
Thomas Kühn6, Ha i Kokkola1,4, Zak Kipling7, Da id Neubaue 7,8, João C. Teixei a9, Alis ai Sella 9,
Duncan Wa son-Pa is10, Yang Yang11, Jialei Zhu12, S ina h K ishnan13, Annele Vi anen1, and
Daniel G. Pa idge3
1Depa men o Technical Physics, Uni e si y o Eas e n Finland, Kuopio, 70210, Finland
2Depa men o En i onmen al Science and Bolin Cen e o Clima e Resea ch,
S ockholm Uni e si y, S ockholm, 10691, Sweden
3Depa men o Ma hema ics and S a is ics, Uni e si y o Exe e , Exe e , EX4 4QF, Uni ed Kingdom
4A mosphe ic Resea ch Cen e o Eas e n Finland, Finnish Me eo ological Ins i u e, Kuopio, 70211, Finland
5Ins i u e o Chemical Enginee ing Sciences o Founda ion o Resea ch and Technology Hellas,
Pa as, 26504, G eece
6Wea he and Clima e Change Impac Resea ch, Finnish Me eo ological Ins i u e, Helsinki, 00101, Finland
7Eu opean Cen e o Medium-Range Wea he Fo ecas s, Reading, RG2 9AX, Uni ed Kingdom
8Ins i u e o A mosphe ic and Clima e Science, ETH Zü ich, Zu ich, Swi ze land
9Me O ice Hadley Cen e, Exe e , EX1 3PB, Uni ed Kingdom
10Sc ipps Ins i u ion o Oceanog aphy and Halıcıo˘
glu Da a Science Ins i u e,
Uni e si y o Cali o nia San Diego, La Jolla, CA 92093, USA
11School o En i onmen al Sciences and Enginee ing, Nanjing Uni e si y o In o ma ion Science and
Technology, Nanjing, Jiangsu, China
12Ins i u e o Su ace-Ea h Sys em Science, School o Ea h Sys em Science,
Tianjin Uni e si y, Tianjin, 300072, China
13CICERO Cen e o In e na ional Clima e Resea ch, Oslo, 0349, No way
anow a : Leibniz Ins i u e o T oposphe ic Resea ch, Leipzig, 04318, Ge many
Co espondence: Sini Tal inen ([email p o ec ed]) and Daniel G. Pa idge (d.g.pa idge@exe e .ac.uk)
Recei ed: 15 Feb ua y 2025 – Discussion s a ed: 21 Ma ch 2025
Re ised: 24 Sep embe 2025 – Accep ed: 24 Sep embe 2025 – Published: 3 No embe 2025
Abs ac . Global clima e models (GCMs) ace unce ain ies in es ima ing Ea h’s adia i e budge due o
ae osol-cloud in e ac ions (ACI). Accu a e pa icle numbe size dis ibu ions (PNSDs) a e c ucial o imp o -
ing ACI ep esen a ion, equi ing p ecise modelling o ae osol sou ces and sinks. Using a Lag angian ajec o y
amewo k, we examine how clouds and p ecipi a ion in luence ae osols du ing anspo , and he eby in lu-
ence ae osol–cloud ela ionships in he bo eal o es . Two GCMs, he Uni ed Kingdom Ea h Sys em Model
(UKESM1) and ECHAM6.3-HAM2.3-MOZ1.0 wi h he SALSA2.0 ae osol module (ECHAM-SALSA), a e
complemen ed wi h model-de i ed ajec o ies and e alua ed agains in-si u obse a ions, which a e accompa-
nied by eanalysis ajec o ies. O e all ae osol–p ecipi a ion ends a e simila be ween GCMs and obse a-
ions. Howe e , seasonal di e ences eme ge: in summe , UKESM1 exhibi s mo e e icien ae osol emo al ia
p ecipi a ion han ECHAM-SALSA and obse a ions, whe eas in win e , he opposi e is obse ed. These di -
e ences coincide wi h key a iables con olling ae osol ac i a ion, such as sub-g id scale upd augh eloci ies
and PNSDs. Fo example, in win e , emo al o o al ae osol mass in ECHAM-SALSA was s onge han in
Published by Cope nicus Publica ions on behal o he Eu opean Geosciences Union.
14450 S. Tal inen e al.: Cloud and p ecipi a ion impac s on bo eal ae osols in GCMs
UKESM1, coinciding wi h highe ac i a ed ac ions and la ge sub-g id scale upd augh eloci ies in ECHAM-
SALSA. Fo bo h GCMs, cloud p ocessing along ajec o ies inc eased SO4mass, mainly in he accumula ion
mode, consis en wi h obse a ions and model pa ame iza ions. Disc epancies a ise mo e om di e ences in
PNSDs and upd augh eloci ies han om we emo al pa ame iza ions, an example being he unde ep esen-
a ion o small pa icles in UKESM1. While ou indings a e ep esen a i e o bo eal egion wi h p edominan ly
s a i o m p ecipi a ion, u he wo k is needed o e alua e hei applicabili y o o he egions.
1 In oduc ion
A mosphe ic ae osol pa icle concen a ions a e in luenced
by hei sou ces and sinks which a ec hei li e imes in
he a mosphe e, and also play a signi ican ole in ou cli-
ma e sys em h ough di e en mechanisms. One o he mos
impo an mechanisms a e ae osol-cloud in e ac ions (ACI),
which a e s ill causing he la ges unce ain ies on he e ec s
o ae osols on Ea h’s adia i e budge in global clima e mod-
els (GCMs, IPCC, 2013; Wa son-Pa is e al., 2019; Bellouin
e al., 2020; Fo s e e al., 2021), and he e o e pa ly mask-
ing he wa ming e ec by g eenhouse gases (Baue e al.,
2022; Quaas e al., 2022). I is c i ical, he e o e, ha he
mic ophysical p ocesses in luencing ACIs a e well unde -
s ood and accu a ely modelled. To accu a ely simula e ACI
in GCMs, he ae osol numbe size dis ibu ions need o be
co ec ly desc ibed (e.g., Mann e al., 2010). T adi ionally,
disc epancies in pa icle size dis ibu ions be ween obse -
a ions and models exceed hose be ween modal and sec-
ional app oaches, wi h sec ional me hods di iding he dis-
ibu ion in o disc e e size bins (Mann e al., 2012). How-
e e , la ge di e ences in concen a ions may eme ge when
chemis y o he ae osols is inspec ed (Laakso e al., 2022).
On he o he hand, o accu a ely ep esen he ae osol num-
be size dis ibu ions, GCMs also need o accu a ely ep e-
sen he sou ce and sink p ocesses ha ac on he ae osol
du ing i s li e ime and anspo in he a mosphe e. The im-
pac o p ecipi a ion on he e olu ion o he size dis ibu ion
is e y impo an (e.g., B owse e al., 2014; Khadi e al.,
2023), bu emains a majo unce ain y in he GCMs. O -
en, when GCM pa ame iza ions a e assessed he models a e
e alua ed agains obse a ions o o he GCMs by inspec ing
di e ences in a e ages o a iables (o ela ionships be ween
mul iple a iables) o e ce ain ime spans (e.g., Blichne e
al., 2024; Gliß e al., 2021; Labe and Ba nes, 2022; Mahe e
al., 2021; Pa hak e al., 2023) in a Eule ian pe spec i e. How-
e e , GCM e alua ions in which he e olu ion o ae osols
and o he a iables is ollowed o e bo h ime and space in
mo e de ail using GCM Lag angian ajec o y-based e alu-
a ion amewo ks ha ha e been ecen ly in oduced (e.g.,
Kim e al., 2020). Such amewo ks pa e he way o he
de elopmen o mo e igo ous obse a ional cons ain s on
unce ain physical and chemical ae osol p ocesses o GCM
e alua ion, by including empo al and spa ial in o ma ion as-
socia ed wi h he ai -mass his o y.
ACIs include sca enging o ae osol pa icles by p ecipi a-
ion, cloud d ople s and ice c ys als. We sca enging is one
o he mos e icien emo al ou es o pa icles om he a -
mosphe e (e.g., Oha a e al., 2016; Liu e al., 2020a). We
sca enging o ae osol pa icles can be u he di ided in o
in-cloud sca enging and below cloud sca enging. We sca -
enging ia in-cloud sca enging in ol es he loss o ae osol
pa icles when hey become ac i a ed in o cloud d ople s o
ice c ys als (nuclea ion sca enging) which can hen u he
collide wi h in e s i ial ae osols in-cloud (e.g., Oha a e al.,
2016; Sein eld and Pandis, 2016). Below-cloud sca enging
conce ns he emo al o ae osol by ain all om he collec-
ion o pa icles due o collisions wi h alling aind ops and
snow and ice om p ecipi a ion (e.g., Oha a e al., 2016).
Cu en unde s anding iden i ies he con ibu ion o in-cloud
sca enging, ollowed by emo al ia p ecipi a ion o be, on
a e age, he mos impo an sink globally o accumula ion
mode pa icles (pa icle diame e dp∼100–1000nm). Ul a-
ine (dp<100nm) and coa se pa icles (dp>1µm), on he
o he hand, a e mo e e icien ly emo ed by below-cloud
sca enging (e.g., And onache, 2003; Tex o e al., 2006;
C o e al., 2009; Oha a e al., 2016). In addi ion o we sca -
enging, clouds can also al e he pa icle p ope ies h ough
aqueous phase oxida ion p ocesses. Fo example, sul a e p o-
duc ion due o oxida ion o gaseous sul u dioxide inside
clouds is conside ed as one o he mos impo an mass addi-
ion p ocesses o sul a e (e.g., E ens, 2015 and e e ences
he ein). P oduc ion o o ganics h ough aqueous phase p o-
cesses has also been epo ed in some en i onmen s (e.g., E -
ens e al., 2018; Lamkaddam e al., 2021).
In es iga ion o he e ec s o p ecipi a ion and clouds has
adi ionally been Eule ian, in which local es ima es o p e-
cipi a ion a e employed (e.g., Wang e al., 2021). Lag angian
app oaches, in which ai mass ajec o ies a e exploi ed o ex-
amine he e ec s o p ecipi a ion on ae osols and hei com-
posi ion as he ai masses a el o he ecep o loca ion, ha e,
howe e , inc eased in popula i y du ing he ecen yea s
(Dadashaza e al., 2021; Heslin-Rees e al., 2024; Isokään ä
e al., 2022; Kes i e al., 2020; Khadi e al., 2023; Tun ed
e al., 2004, 2013; Tun ed and S öm, 2019). These ypes o
s udies can p o ide signi ican ly mo e de ailed insigh s by
conside ing he in e play be ween ae osols, clouds and p e-
cipi a ion du ing ai mass his o y, ha canno be achie ed us-
ing Eule ian app oaches. All hese s udies in es iga ed how
he o al accumula ed p ecipi a ion expe ienced along ai -
A mos. Chem. Phys., 25, 14449–14478, 2025 h ps://doi.o g/10.5194/acp-25-14449-2025
S. Tal inen e al.: Cloud and p ecipi a ion impac s on bo eal ae osols in GCMs 14451
mass ajec o ies de i ed om eanalysis da a a ec s a pa -
icle size dis ibu ion measu ed a a speci ic ecep o si e.
Tun ed e al. (2013), o example, in es iga ed ae osols in he
A c ic (Zeppelin s a ion, Ny-Ålesund, No way) and obse ed
s ong emo al o sub-mic on pa icula e mass up o 10 mm
o accumula ed p ecipi a ion. They sugges ed he in-cloud
sca enging ( ollowed by emo al ia p ecipi a ion) is he
dominan emo al pa hway, as la ge pa icles showed i s
a dec ease in hei concen a ion as a unc ion o accumu-
la ed p ecipi a ion du ing anspo , ollowed by he emo al
smalle sizes. Kes i e al. (2020) s udied ae osols a he hu-
mid opical monsoon clima e in he Maldi es, and obse ed
mo e e icien emo al o he accumula ion mode pa icles
wi h inc easing accumula ed p ecipi a ion, when compa ed
o he smalle pa icle sizes. Dadashaza e al. (2021) s udied
sub- opical en i onmen s in Be muda and concluded ha
PM2.5mass expe ienced he s onges sensi i i y o accumu-
la ed p ecipi a ion up o 5mm whe eas p ecipi a ion exceed-
ing his limi had no majo e ec s on he pa icula e mass.
Khadi e al. (2023) u he epo ed ha p ecipi a ion can, in
some ins ances, se e as a sou ce o ae osols.
In addi ion o he e ec s o p ecipi a ion o ae osols, a
p e ious s udy by Isokään ä e al. (2022) used ela i e hu-
midi y (>94%) as a p oxy o in-cloud exposu e in bo-
eal ai masses and ound a p onounced inc ease in sul a e
mass in ai masses ecen ly in luenced by non-p ecipi a ing
clouds, while no signi ican aqueous-phase p oduc ion o o -
ganic ae osol was obse ed – likely due o dominan gas-
phase biogenic sou ces. This is consis en wi h indings om
cen al Sweden (G aham e al., 2020). These ea lie esul s
sugges ha sul a e may be mo e s ongly a ec ed by cloud
p ocessing and we emo al han o ganic ae osol, wi h e-
mo al e iciency likely in luenced by ac o s such as p ecip-
i a ion iming, ae osol ype, and he s age o he ai mass
ajec o y. Ou s udy builds on his by explo ing hese as-
pec s ac oss mul iple models and obse a ions, employing
he GCM Lag angian e alua ion amewo k p esen ed by
Kim e al. (2020). Wi h his amewo k ai mass ajec o-
ies can be ob ained om global GCM simula ions. This
is achie ed by co-loca ing mul iple a iables ( o example,
ae osol size dis ibu ion and chemical composi ion) om he
GCMs o ai mass ajec o ies calcula ed om he GCM me-
eo ological da a (Kim e al., 2020). This me hodology al-
lows us o anspa en ly e alua e and compa e he we sca -
enging and aqueous-phase p ocessing be ween he obse a-
ions and GCMs wi hin he Lag angian ajec o y amewo k
in unp eceden ed de ail.
This s udy compa es he e ec s o we p ocessing (we e-
mo al and aqueous-phase p ocessing) on modelled ae osol
size dis ibu ions wi h long- e m obse a ions om Hyy iälä,
Finland. Obse a ional ajec o ies a e based on ERA-
In e im eanalysis, while model ajec o ies a e calcula ed us-
ing me eo ology da a om GCM AMIP-s yle simula ions in
which wind ields we e nudged o ERA-In e im. The GCMs
used in his s udy include UKESM1 (Uni ed Kingdom Ea h
Sys em Model, e.g., Sella e al., 2019) and ECHAM6.3-
HAM2.3-MOZ1.0 wi h sec ional ae osol module SALSA2.0
(he ea e ECHAM-SALSA, S e ens e al., 2013; Kokkola
e al., 2018; Tegen e al., 2019). Bo h GCMs a e pa o
he Ae osol Compa isons be ween Obse a ions and Models
(Ae oCom) Phase III GCM T ajec o y Expe imen (GCM-
T aj) in which a compa ison be ween he GCMs agains e-
analysis me eo ology was conduc ed o he yea s be ween
2009 and 2013. In his s udy he simula ions o UKESM1
and ECHAM-SALSA co e he yea s om 2005 o 2018
which a e also a ailable om he obse a ions. Compa -
ison be ween modal (UKESM1) and sec ional (ECHAM-
SALSA) app oaches o es ima ing he ae osol mic ophysics
p o ides addi ional insigh in o he model beha iou ia his
Lag angian e alua ion app oach. The Hyb id Single-Pa icle
Lag angian In eg a ed T ajec o y model (HYSPLIT; D axle
and Hess, 1998; S ein e al., 2015) is employed o ob ain he
backwa d ai mass ajec o ies. A key di e ence be ween ou
s udy and p e ious wo k, including Isokään ä e al. (2022),
is ou ocus on s a i o m p ecipi a ion a he han o al p e-
cipi a ion. S a i o m p ecipi a ion is he dominan ype in
mid- and high-la i ude egions (30–60° om he equa o and
polewa d), whe eas opical egions a e ypically in luenced
by con ec i e sys ems (e.g., Schumache and Funk, 2023).
Since ou s udy a ea is p ima ily he bo eal o es egion
o no he n Eu ope, s a i o m p ecipi a ion is mos ele an .
The di e ing impac s o p ecipi a ion ypes on ae osols ha e
also been highligh ed by Khadi e al. (2023), who showed
ha ecen opical p ecipi a ion – la gely con ec i e – can
be linked o downd a s ha anspo small pa icles om
highe al i udes o he bounda y laye (see also F anco e al.,
2022; Machado e al., 2021; McCoy e al., 2021; Williamson
e al., 2019).
The aim o ou esea ch can be summa ized in o wo main
objec i es (1–2):
1. Do he ela ionships be ween ae osols and expe ienced
p ecipi a ion du ing anspo di e be ween he mea-
su emen s and GCMs and wha a e he d i e s o he
obse ed di e ences?
2. Do he GCMs exhibi simila inc ease in sul a e mass
due o in-cloud p oduc ion as he obse a ions and
a e he obse ed e ec s easonable when compa ed o
model pa ame iza ions?
We s a ou in es iga ion in Sec . 2 by i s in oducing he
obse a ional da ase s, ollowed by summa ising he GCM
simula ions along wi h de ails on he ai mass ajec o y cal-
cula ions and da a co-loca ions employed in his wo k. The
ae osol p ope ies a he measu emen s a ion (Hyy iälä, Fin-
land) a e gi en in Sec . 3 as a necessa y backg ound o he
ollowing Lag angian analysis. The ela ionships be ween
p ecipi a ion and ae osol mass and numbe in he Lag angian
amewo k a e p esen ed i s (Sec . 4.1–4.3), ollowed by a
p ocess-chain ype e alua ion (Sec . 4.4) o unde s and he
h ps://doi.o g/10.5194/acp-25-14449-2025 A mos. Chem. Phys., 25, 14449–14478, 2025
14452 S. Tal inen e al.: Cloud and p ecipi a ion impac s on bo eal ae osols in GCMs
d i ing o ces in he ela ionships. Finally, in Sec . 5, he e -
ec s o aqueous-phase p ocessing a e p esen ed, ollowed by
o e all conclusions (Sec . 6) and ou look (Sec . 7).
2 Da a and me hods
2.1 Obse a ions a SMEAR II
Obse a ional da a used in his s udy include long- e m mea-
su emen s o ae osol numbe size dis ibu ions and pa -
icle chemis y om SMEAR II (S a ion o Measu ing
Ecosys em-A mosphe e Rela ions in; Ha i and Kulmala,
2005) and a e desc ibed in de ail in Isokään ä e al. (2022)
and he e e ences he ein. SMEAR II s a ion (Hyy iälä, Fin-
land) is classi ied as a u al en i onmen , su ounded by el-
a i ely homogenous Sco s pine (Pinus syl es e is) o es . In
his wo k pa icle numbe size measu emen s (co e ing pa i-
cle diame e s be ween 3–1000nm) ob ained wi h a di e en-
ial mobili y pa icle size (DMPS, e.g., Aal o e al., 2001) a e
u ilized. Chemical composi ion (o ganics, sul a e, and equi -
alen black ca bon) o he pa icles in he sub-mic on ange
we e de i ed om an ae halome e (e.g., D ino ec e al.,
2015) and ae osol chemical specia ion moni o (ACSM, Ng
e al., 2011). The da ase o pa icle numbe size measu e-
men s spans 2005–2018, sligh ly sho e han in Isokään ä e
al. (2022), o ma ch he GCM simula ion pe iod. The ASCM
da a ex ends om 2012 o 2018.
2.2 Summa ies o he GCMs used in his s udy
2.2.1 UKESM1
The Uni ed Kingdom Ea h Sys em Model (UKESM1) con-
igu a ion used in his s udy uses he a mosphe ic and land
componen s ollowing he p o ocol se by he A mosphe ic
Model In e compa ison P ojec (AMIP, Ey ing e al., 2016).
The a mosphe ic componen o he model is based on he
Global A mosphe e 7.1 (GA7.1) and he Global Land 7.0
(GL7.0) con igu a ions, as desc ibed by Wal e s e al. (2019).
These a e pa o he Hadley Cen e Global En i onmen
Model e sion 3 (HadGEM3; Hewi e al., 2011), which is
coupled o he e es ial ca bon/ni ogen cycles (Sella e
al., 2019). I includes in e ac i e s a osphe e– oposphe e
chemis y om he om he UK Chemis y and Ae osol
(UKCA) model (A chibald e al., 2020; Mo gens e n e al.,
2009; O’Conno e al., 2014).
Following he AMIP p o ocol, sea su ace empe a u e and
sea ice a e aken om he unmodi ied da ase o Du ack e
al. (2017) and ho izon ally in e pola ed o he model esolu-
ion. In his se up, he dynamic ege a ion model (Cox, 2001)
is u ned o . Ins ead, p esc ibed ege a ion om a his o i-
cal coupled UKESM1 simula ion is used o main ain con-
sis en land-use o cing be ween he coupled and AMIP ex-
pe imen s. In a simila ashion, seawa e concen a ions o
dime hyl sul ide (DMS) and chlo ophyll-a mon hly clima-
ologies a e aken om he coupled his o ical expe imen and
a e used by he a mosphe e model op calcula es luxes o
DMS and p ima y ma ine o ganic ae osol (Mulcahy e al.,
2020).
The simula ions we e nudged o ERA-In e im eanalysis
(Dee e al., 2011; Tel o d e al., 2008) u/ (ho izon al and
e ical), wind ields and su ace p essu e ollowing he se up
design o he Ae oCom GCMT aj phase III expe imen . The
model esolu ion o hese con igu a ions was 1.875°×1.25°
longi ude–la i ude, co esponding o a ho izon al esolu ion
o ∼135km in he midla i udes. The model has 85 e ical
le els which a e di ided such ha 50 le els a e be ween 0
and 18km and he emaining 35 le els co e heigh s be ween
18 and 85km.
A mosphe ic composi ion wi hin UKESM1 is imple-
men ed as pa o he UKCA model. Wi hin UKCA, he
Global Model o Ae osol P ocesses (GLOMAP; Mann e al.,
2010; Mulcahy e al., 2020) is used. This scheme simula es
mul icomponen global ae osols, including, o example, sul-
a e, black ca bon, and o ganic ma e . The ae osol pa icle
size dis ibu ion is ep esen ed using i e log-no mal modes,
nuclea ion soluble, Ai ken soluble, accumula ion soluble,
coa se soluble and Ai ken insoluble isualized in Fig. S1
in he Supplemen . Mo e de ails, including he size anges
o each ae osol mode, a e p esen ed in Sec . S1.1 in he
Supplemen . The GLOMAP model also includes a ious mi-
c ophysical p ocesses ha a ec he e olu ion o ae osol
p ope ies. We sca enging p ocesses in UKESM1, including
below-cloud (impac ion), in-cloud (nuclea ion) and plume
sca enging a e summa ized in Sec . S2 and e e ences
he ein. As a key di e ence o ECHAM-SALSA (Sec . 2.2.2)
conce ning he ae osol pa ame iza ions, new pa icle o ma-
ion in he bounda y laye is no implemen ed in his e sion
o UKESM1 (Mulcahy e al., 2020).
Fo his s udy he Ae oCom GCMT aj UKESM1 simula-
ions (2009–2013) we e ex ended o he pe iod om 2005
o 2018 o acili a e obus s a is ical compa ison wi h he
ae osol size dis ibu ions and composi ion measu emen s ob-
ained om SMEAR II. The model ou pu ields we e ex-
ac ed a high empo al esolu ion (3-hou ly ou pu ) o all
model le els (when a ailable, o he wise no ed as su ace).
The diagnos ics ields u ilized in his wo k (see also Ta-
ble S4) a e ae osol pa icle size dis ibu ion a iables (num-
be concen a ions and d y diame e s o each ae osol mode),
chemical componen s including mass mixing a ios o sul-
a e no ed he e as SO4(ex ac ed as sul u ic acid H2SO4
and hen con e ed, see Sec . S1.1), o ganic ma e (no ed
he e as OA) and black ca bon (BC), o al (including bo h liq-
uid ain and snow) s a i o m and con ec i e p ecipi a ion a
he su ace, d y ai densi y, sub-g id scale upd augh eloc-
i y, numbe o ac i a ed pa icles, o al p ecipi a ion a he
su ace, ela i e humidi y and cloud ac ions. Addi ionally,
om UKESM1, we sca enging coe icien s ( ep esen ing
emo al wi hin he whole a mosphe ic column) o he di -
e en emo al p ocesses (nuclea ion, impac ion and plume)
and species (OA, H2SO4and BC), SO2concen a ions, and
A mos. Chem. Phys., 25, 14449–14478, 2025 h ps://doi.o g/10.5194/acp-25-14449-2025
S. Tal inen e al.: Cloud and p ecipi a ion impac s on bo eal ae osols in GCMs 14453
bo h e ically esol ed and su ace liquid s a i o m p ecipi-
a ion a e inspec ed. These a iables and/o a iables de i ed
om hem a e co-loca ed o he UKESM1 de i ed HYSPLIT
back- ajec o ies as desc ibed in Sec . 2.3.
2.2.2 ECHAM-SALSA
ECHAM6.3-HAM2.3-MOZ1.0 is a global ae osol-
chemis y-clima e model consis ing o he a mosphe ic
gene al ci cula ion model ECHAM (S e ens e al., 2013)
coupled wi h he Hambu g Ae osol Model HAM (Tegen e
al., 2019) and chemis y model MOZ (Schul z e al., 2018).
Fo his wo k, as o UKESM1, simula ions ollow AMIP
s yle uns ollowing he Ae oCom phase III GCMT aj expe -
imen se up. The e o e, as o UKESM1, he u/ wind ields
and su ace p essu e we e nudged owa ds ERA-In e im
eanalysis da a. In addi ion, he sea su ace empe a u e
and sea ice co e we e p esc ibed based on mon hly mean
clima ologies ob ained om he AMIP p ojec (Ey ing e
al., 2016). The model sol es a mosphe ic ci cula ion wi h
e ical g idding o 47 laye s ex ending oughly up o 80 km.
Model ho izon al esolu ion o hese con igu a ions is
1.875°×1.875° longi ude–la i ude.
ECHAM6.3-HAM2.3-MOZ1.0 is pai ed wi h he
sec ional ae osol mic ophysics model SALSA2.0 (ECHAM-
SALSA) in which he size dis ibu ion is di ided in o
3 sub anges (dp1 =3–50nm, dp2 =50–700nm and
dp3 =700nm–10µm) including 10 size classes in loga-
i hmical size space. Sub anges dp2 and dp3 include pa allel
size classes o insoluble and soluble ae osol species, making
he o al numbe o size classes 17 (Kokkola e al., 2018),
isualized in Fig. S1. Mo e de ails o he sub anges and hei
composi ions a e gi en in Sec . S1.2. Addi ional de ails o
he ae osol p ocesses calcula ed in SALSA2.0 can be ound
in Kokkola e al. (2018) and Holopainen e al. (2020). We
sca enging pa ame iza ions a e summa ized in Sec . S2 o
below- and in-cloud sca enging.
As o UKESM1, simula ions co e he yea s om 2005
o 2018 o ECHAM-SALSA. Da a ou pu is also 3-hou ly
and e ically esol ed unless he a iable is no ed as su ace
a iable. The diagnos ics ex ac ed om ECHAM-SALSA
(see also Table S4) include ae osol pa icle size dis ibu ion
a iables (numbe concen a ions and d y diame e s o each
size class), chemical componen s including mass mixing a-
ios o sul a e (SO4), o ganics (no ed he e as OA) and black
ca bon (BC), o al (including bo h liquid ain and snow) s a -
i o m and con ec i e p ecipi a ion a he su ace, d y ai den-
si y, sub-g id scale upd augh eloci y, numbe o ac i a ed
pa icles, o al p ecipi a ion a he su ace, ela i e humid-
i y and cloud ac ions. Simila o UKESM1, hese a iables
and/o a iables calcula ed om hem a e co-loca ed o he
ECHAM-SALSA de i ed HYSPLIT back- ajec o ies as de-
sc ibed in Sec . 2.3.
2.3 Ai mass ajec o y calcula ions and da a co-loca ion
2.3.1 HYSPLIT
The 4d (96h) back ajec o ies a i ing a SMEAR II we e
calcula ed by e sion 5.1.0 o he HYSPLIT (S ein e al.,
2015) model o he pe iod om Janua y 2005 o Decem-
be 2018. The 4d long back ajec o ies we e used o ensu e
consis ency wi h he esul s om Isokään ä e al. (2022). In
addi ion, his is ypically a long enough pe iod o slowly
mo ing ai masses o a el o he bo eal en i onmen om
high a c ic and ma ine a eas. A i al heigh o he ajec o-
ies o he ecep o s a ion was se o 100m abo e he g ound
le el. To ob ain he GCM de i ed ajec o ies, he me eo-
ological ields om he GCMs we e i s con e ed in o a
consis en ne CDF4 o ma which was hen con e ed in o
he ARL packed HYSPLIT4 compa ible o ma (Kim e al.,
2020). The GCM and ERA-In e im (Dee e al., 2011) eanal-
ysis me eo ological da ase s equi ed o he HYSPLIT4 a-
jec o y calcula ions we e e-g idded o a consis en 1° ho -
izon al esolu ion. The e ical disc e iza ion o he GCM
a iables was p o ided on e ain- ollowing hyb id sigma-
p essu e le els. In UKESM1, howe e , he na i e ou pu is
on hyb id heigh le els, which is no suppo ed by HYSPLIT.
The e o e, UKESM1 was ou pu on ixed p essu e le els, se-
lec ed o closely ma ch he ERA-In e im p essu e le els.
T ajec o ies we e calcula ed o e e y 3 d hou o bo h
eanalysis da a and he GCMs, co esponding o GCM ou -
pu esolu ion. This led o 8 ajec o ies pe day, a o al o
40896 ai mass ajec o ies be ween 2005–2018 be o e ap-
plying any p e-p ocessing and empo al ha moniza ion o he
da a (Sec . 2.4). He ea e , when discussing obse a ional
da a coupled wi h he ERA-In e im back- ajec o ies, hose
a e e e ed as obse a ions unless men ioned o he wise. I
should be no ed ha eanalysis da a is no in e changeable
wi h obse a ions bu is used as a p oxy in his s udy.
2.3.2 Co-loca ion o GCM da a along he ai mass
ajec o ies
The a iables om he GCMs desc ibed in Sec . 2.2.1 and
2.2.2 we e empo ally ( ime), spa ially (la i ude, longi ude)
and e ically ( a iables which co e ed di e en model o
p essu e le els) co-loca ed o he GCM de i ed ai mass a-
jec o ies. In sho , a co-loca o ool (Kim e al., 2020) based
o he Communi y In e compa ison Sui e (CIS, Wa son-
Pa is e al., 2016) was used o co-loca e 4-dimensional da a
which uses hyb id al i ude coo dina es. As he de aul in e -
pola o wi hin CIS has o en di icul ies co-loca ing o he
nea -su ace ajec o y poin s (due o su ounding g id-boxes
being a he bounda ies o he da a domain), he modi ied co-
loca o p o ided mo e lexibili y o he in e pola ion o hese
nea -su ace poin s. This is ele an also in his wo k, as o
ou su ace si e he ajec o ies can also a el a low al i-
udes. In his imp o ed co-loca o , when he linea in e pola-
ion in he nea -su ace ajec o ies would esul in o a miss-
h ps://doi.o g/10.5194/acp-25-14449-2025 A mos. Chem. Phys., 25, 14449–14478, 2025
14454 S. Tal inen e al.: Cloud and p ecipi a ion impac s on bo eal ae osols in GCMs
ing alue, nea es -neighbou in e pola ion is used ins ead.
Thus, ex apola ion o alues can be a oided and in o ma ion
o ajec o y poin s ha a e wi hin he da a domain e ained.
The co-loca ed GCM da a om he ai mass ajec o y a i al
imes, i.e., imes when he ai mass is loca ed a SMEAR II,
a e used o ep esen he condi ions a SMEAR II, acili a ing
di ec compa ison o obse a ional da a ob ained a he si e.
A di e ence o Isokään ä e al. (2022) whe e he ERA-
In e im p ecipi a ion in e nally p ocessed by HYSPLIT on o
ajec o ies coo dina es was used, is ha he aw p ecipi a ion
ields om ERA-In e im a e employed in his wo k by co-
loca ing hem o he ai mass ajec o ies in a pos -p ocessing
s ep (as o he a iables ex ac ed om GCMs desc ibed
abo e). This app oach was chosen o e ain he o iginal nu-
me ical p ecision o ERA-In e im (and GCM) p ecipi a ion
da a, ensu ing accu a e alignmen wi h co-loca ed GCM a i-
ables (e.g., ae osol size dis ibu ions and chemical compo-
si ion), which HYSPLIT does no p o ide. He e, “consis-
ency” e e s o nume ical accu acy a he han ma ching da a
sou ces.
2.4 Da a ha moniza ion be ween measu emen s and
GCMs
2.4.1 Tempo al co-loca ion and da a p e-p ocessing
The da a om he measu emen s (1-hou ly a e ages) con-
duc ed a SMEAR II was empo ally co-loca ed wi h
he ERA-In e im de i ed back- ajec o y a i al imes (3-
hou ly). Addi ionally, he GCM de i ed ajec o ies (3-
hou ly) we e only co-loca ed wi h he imes when ae osol
obse a ions we e a ailable. By adop ing his app oach, only
GCM ajec o ies co esponding o exis ing da a poin s in ob-
se a ions we e e ained and u ilized in u he analysis. The
impo ance o empo al co-loca ion o model e alua ion is
discussed, o example, in Schu gens e al. (2016). Ha mon-
isa ion o he measu ed ae osol size dis ibu ion and com-
posi ion wi h he co esponding a iables a ailable om he
GCMs a e desc ibed in Sec . 2.4.2 and 2.4.3.
Fo consis ency wi h Isokään ä e al. (2022) iden ical p e-
p ocessing is applied he e o he in-si u ae osol obse a-
ions be o e he empo al co-loca ion desc ibed abo e. Thus,
da a poin s o which he measu ed wind di ec ion was be-
ween 120 and 140° we e emo ed due o possible in lu-
ence o s ong VOC ( ola ile o ganic compound) emissions
om he local sawmill (Heikkinen e al., 2020; Liao e al.,
2011). In addi ion, ajec o ies c ossing he a ea o Kola
Peninsula we e excluded as in Isokään ä e al. (2022) due
o s ong pollu ion sou ces wi hin he a ea (Heikkinen e al.,
2020; Kulmala e al., 2000; Riu anen e al., 2013). This
led o ae osol size dis ibu ion da a co e ing he yea s be-
ween 2005 and 2018 (numbe o inal da a ows/ ajec o-
ies: 30688) and ae osol chemical composi ion o he yea s
be ween 2012 and 2018 (numbe o inal da a ows/ ajec o-
ies: 6174). Dis ibu ion o he da a poin s o e he yea s a e
shown in Figs. S2 and S3.
2.4.2 Ae osol pa icle numbe size dis ibu ion
The DMPS (di e en ial mobili y pa icle size , e.g., Aal o e
al., 2001) obse a ions include 51 size bins in he obse ed
size ange (dp=3–1000nm). Fo UKESM1, comple e log-
no mal pa icle numbe size dis ibu ions (Sein eld and Pan-
dis, 2016) we e calcula ed by using he modal pa ame e s
(d y diame e s, numbe concen a ions and geome ic mean
diame e s) gi en by he model. The numbe size dis ibu ion
is disc e ised in o he same size g id as he obse a ions i.e.,
he bin midpoin s a e iden ical o he ones a ailable om
he DMPS measu emen s. This app oach was possible as in
SMEAR II he size g id DMPS applies s ays cons an o e
he whole in es iga ed pe iod. This ha moniza ion was con-
duc ed o each hou along he ai mass ajec o ies using he
co-loca ion app oach desc ibed in Sec . 2.3.2 as UKESM1
p o ided all needed modal pa ame e s o calcula ion o he
ull pa icle numbe size dis ibu ions (PNSD) along he a-
jec o ies.
Fo ECHAM-SALSA, he numbe concen a ions o sol-
uble and insoluble bins (i.e., size classes) we e added o-
ge he o each size bin. To make he loga i hmic numbe
size dis ibu ion compa able o UKESM1 da a and DMPS
measu emen s, he alues wi hin each size bin (i) we e di-
ided by he loga i hm o he maximum size di,max minus he
loga i hm o he minimum size di,min i.e., by log10(di,max)-
log10(di,min) o ha size bin (see Table S3). Simila o
UKESM1, his was conduc ed along he ajec o ies. Fo
ae osols, ECHAM-SALSA bins anging om 3.0 o 1700nm
in diame e a e s udied, as by s ic ly limi ing o sub-
mic on bins (≤700nm), he la ges sub-mic on pa icles
(700nm< dp≤1000nm) ha do con ibu e o he o al pa -
icle mass, would be los . Sensi i i y analysis was conduc ed
including only he sub-mic on bins, and none o he conclu-
sions changed.
In eg a ed a iables, such as o al numbe and mass con-
cen a ions ( o submic on pa icles) we e calcula ed om
he pa icle numbe size dis ibu ions by assuming he pa -
icles a e sphe ical and ha e a cons an densi y o ρ=
1.6gcm−3. This densi y co esponds o he a e age densi y
o pa icles obse ed a SMEAR II (e.g., Häkkinen e al.,
2012). Again, hese quan i ies we e calcula ed o each hou
(i.e., 96 da a poin s, see Sec . 2.3.1) along e e y single ai
mass ajec o y.
2.4.3 Chemical composi ion
Obse a ional da a o o ganic ae osol (he ea e OA) and
sul a e (he ea e SO4) was ob ained using obse a ions om
ACSM (ae osol chemical specia ion moni o , Ng e al., 2011)
which is mos e icien a measu ing pa icles wi h∼75–
650nm o acuum ae odynamic diame e , passing h ough
A mos. Chem. Phys., 25, 14449–14478, 2025 h ps://doi.o g/10.5194/acp-25-14449-2025
S. Tal inen e al.: Cloud and p ecipi a ion impac s on bo eal ae osols in GCMs 14455
pa icles up o 1µm (Liu e al., 2007). Fo UKESM1, Ai ken
and accumula ion mode a e used in his con ex by sum-
ming he mass mixing a ios (MMR, kg o species pe kg
o ai ) o hese modes, including bo h soluble and insoluble
modes when a ailable. Due o he de ini ion o he modes
in UKESM1, hese co espond o pa icle diame e s be ween
10–500nm (see Sec . S1.1), hus ha ing la ge o e lap wi h
he size ange mos e icien ly ep esen ed in ACSM. The
MMRs om UKESM1 and ECHAM-SALSA a e con e ed
in o mass concen a ions by mul iplying he MMRs wi h he
densi y o he ai o acili a e compa isons o chemis y ob-
se a ions gi en in he uni s o µgm−3. Equi alen black
ca bon (he ea e BC) was measu ed wi h an ae halome e
using a cu o diame e o 10µm (PM10). Due o mos o
he abso bing pa icles a SMEAR II being a sub-mic on
ange, he di e ence in he BC mass be ween PM1and
PM10 is only 10% (Luoma e al., 2019). The e o e, om
UKESM1, Ai ken and accumula ion modes a e also used
o es ima e he o al BC. In addi ion, o ob ain SO4 om
H2SO4(sul u ic acid) which is he UKESM1 na i e ou pu ,
a con e sion ac o is used (see Sec . S1.1). F om ECHAM-
SALSA, bins wi h diame e s anging om 19.6 o 700 nm
(see Sec . S1.2) a e used o es ima e he o al sub-mic on
OA, SO4and BC, including again bo h soluble and insol-
uble bins. He e, o ECHAM-SALSA, he la ges bin o
which a po ion also consis s o ae osols la ge han 1µm
(700nm< dp<1700nm) is no included o ensu e consis-
ency wi h he ACSM measu ing e iciency (which dec eases
om ∼650nm up o he maximum size o 1µm).
3 Ae osol p ope ies a SMEAR II – Eule ian
compa ison be ween obse a ions and GCMs
To se he scene and p o ide con ex o GCM de elopmen
since hese p e ious s udies (see also e.g., Redding on e
al., 2016), a sho assessmen o he di e ences and simi-
la i ies in Eule ian amewo k be ween he ae osol obse a-
ions, UKESM1 and ECHAM-SALSA a SMEAR II is gi en
he e. Ai mass anspo be ween ERA-In e im and he GCMs
is i s assessed (Sec . 3.1), ollowed by he ae osol pa icle
numbe size dis ibu ions (Sec . 3.2) and chemical compo-
si ion (Sec . 3.3). This p o ides he necessa y backg ound
in o ma ion o acili a e u he compa isons wi hin he La-
g angian e alua ion amewo k used in his wo k.
3.1 Compa ison o ai mass anspo be ween
ERA-In e im and he GCMs
To ensu e he di e ences shown in he ollowing sec ions o
he Eule ian analysis a e no d i en by di e ging anspo
pa hways be ween he GCMs and ERA-In e im, he ai mass
anspo ou es we e inspec ed. The ai mass anspo ou es
in Fig. 1 show e y simila pa e ns o ERA-In e im and
he GCMs, i.e., he di e ences a e, on a e age, e y small
– as expec ed o simula ions in which wind ields a e con-
sis en ly nudged o ERA-In e im eanalysis. Ve ical ans-
po di e ences exis (Fig. S5), which can be a ibu ed o
po en ial empe a u e no being nudged, which ollows s an-
da d p ac ices (Zhang e al., 2014). Fo his s a ion, howe e ,
hese di e ences a e ela i ely small, and he la ges di e -
ences a e in a eas wi h low equency o ajec o ies. The e-
o e, any obse ed di e ences in he analyses p esen ed in
he ollowing sec ions a e unlikely o be domina ed by di -
e ences in he ai mass anspo .
3.2 Ae osol pa icle numbe size dis ibu ions
In Fig. 2 pa icle numbe size dis ibu ions om he GCMs
a e compa ed wi h obse a ional da a a SMEAR II. The ig-
u e e eals ha UKESM1 unde es ima es he numbe con-
cen a ion o he small (dp<50nm) pa icles, especially du -
ing summe (Fig. 2b, Table S5). This is, howe e , expec ed,
as he new pa icle o ma ion om bounda y laye nucle-
a ion was no implemen ed in UKESM1 (Mulcahy e al.,
2020). ECHAM-SALSA does ha e a be e ep esen a ion
o he PNSD o he smalle ae osol pa icles du ing sp ing
and summe when compa ed o obse a ions (Fig. 2c), Du -
ing wa me seasons, also he absolu e numbe concen a ions
ag ee well be ween obse a ions and ECHAM-SALSA (see
nuclea ion mode om Table S5). This highligh s he impo -
ance o NPF om nuclea ion in he bounda y laye , espe-
cially in summe . Du ing win e , howe e , ECHAM-SALSA
does exhibi some o e es ima ion o Ai ken mode ae osols
(Fig. 2e and Ai ken mode om Table S5).
Du ing win e , UKESM1 o e es ima es la ge Ai ken and
accumula ion mode ae osols (dpup o 200nm) compa ed o
he obse a ions (Fig. 2b and g), bu du ing sp ing he num-
be concen a ion o he accumula ion mode ae osols is e y
close o obse a ions (367cm−3in UKESM1 s 352cm−3
in obse a ions as shown in Table S5). This is somewha
su p ising conside ing he missing g ow h o small pa icles
om NPF in o accumula ion mode, howe e , his could in-
dica e ha he e a e o he p ocesses ha domina e he accu-
mula ion mode. Du ing win e (Fig. 2g) he obse a ions ex-
hibi clea bimodal PNSD peaking a ound 50 and 200nm bu
nei he o he GCMs is able o cap u e his beha iou . O e -
all, bo h GCMs end o be shi ed owa ds he la ge sizes
in all seasons (Fig. 2d–g), and his e ec is sligh ly mo e
p onounced in UKESM1. O e all, ECHAM-SALSA be e
es ima es o he peak alues o he PNSD, excep in win e
(Fig. 1g), when i o e es ima es he pa icle concen a ions a
he size ange o dp=50–100nm.
3.3 Chemical composi ion o he ae osols
Pa icle chemical composi ion as a mass concen a ion o
each chemical species om he composi ion measu emen s
and he GCMs a SMEAR II ( ajec o y ecep o loca ion)
is illus a ed in Fig. 3, and he seasonal pa e ns a e ypical
o his loca ion. La ges concen a ion o o ganic ma e ial is
h ps://doi.o g/10.5194/acp-25-14449-2025 A mos. Chem. Phys., 25, 14449–14478, 2025
14456 S. Tal inen e al.: Cloud and p ecipi a ion impac s on bo eal ae osols in GCMs
Figu e 1. ERA-In e im ai mass ajec o y equencies o sp ing (MAM), summe (JJA), au umn (SON) and win e (DJF) a e shown in he
op ow. F equencies o UKESM1 (e–h) and ECHAM-SALSA (i–l) a e shown as di e ences o he ERA-In e im. Be o e calcula ing he
di e ences, he GCM hexagonal g id (150 hexagons in he x-di ec ion) we e i s eg idded o ma ch he g idding in ERA-In e im. Red c oss
shows he loca ion o SMEAR II.
p esen du ing summe (JJA) and smalles in win e (DJF).
Bo h GCMs also ha e p onounced OA concen a ion du -
ing summe compa ed o he o he seasons, and UKESM1
cap u es he p onounced OA concen a ions obse ed du ing
summe pa icula ly well (median OA 2.0 and 2.2 µg m−3in
UKESM1 and obse a ions, espec i ely, Table S6). A po -
ion o he small unde es ima ion o he OA concen a ions
o he GCMs du ing sp ing and summe could, howe e , be
in luenced by he heigh o he obse a ions as chemical com-
posi ion measu emen s a e conduc ed a he su ace whe eas
he GCM da a shown he e a e a he ajec o y a i al poin
heigh a he ecep o s a ion (100ma.g.l.). Scale di e ence
likely also plays a ole, as he poin measu emen s a e com-
pa ed wi h he GCM g id box alues in e pola ed o ai mass
ajec o ies. Mon hly da a (Fig. 3e) shows he second OA
peak o he obse a ions o be in Feb ua y, as expec ed based
on Heikkinen e al. (2020), and in ECHAM-SALSA his peak
alls on Janua y. UKESM1 peaks in Feb ua y, bu he di -
e ence in he concen a ions (compa ed o obse a ions) be-
ween Feb ua y and Janua y/Ma ch is e y small. The sea-
sonali y o he OA concen a ions p esen ed he e o bo h
obse a ions and GCMs also ag ees wi h he esul s om
Blichne e al. (2024) who p esen ed he same GCMs bu o
a di e en ime pe iod. Di e ences in he mon hly peak con-
cen a ion can be obse ed o BC oo, whe e obse a ions
A mos. Chem. Phys., 25, 14449–14478, 2025 h ps://doi.o g/10.5194/acp-25-14449-2025
S. Tal inen e al.: Cloud and p ecipi a ion impac s on bo eal ae osols in GCMs 14457
Figu e 2. Pa icle numbe size dis ibu ion a SMEAR II as medians o he day o he yea o DMPS measu emen s (g ound le el) a e
shown in (a), ollowed by he di e ences be ween he DMPS obse a ions and he GCMs in (b) and (c). Fo subplo (c), he measu ed size
dis ibu ion was i s eg idded o he ECHAM-SALSA bins by in eg a ing be ween he uppe and lowe limi o each ECHAM-SALSA size
bins be o e calcula ing he di e ence. Median PNSDs o each season a e shown in (d)–(g) wi h shaded a eas indica ing he 25 h and 75 h
pe cen iles.
and UKESM1 peak in Feb ua y, bu ECHAM-SALSA ex-
hibi s he la ges BC concen a ions in Janua y (Fig. 3g).
In gene al, e en hough a pe ec ha moniza ion o he pa -
icle chemical composi ion da a be ween obse a ions and
GCMs is no achie ed (see Sec . 2.4.3), he median con-
cen a ions be ween obse a ions and GCMs ag ee ela i ely
well when he o e all seasonali y is inspec ed (Fig. 3a–d); he
concen a ions a e domina ed by OA in all seasons, ollowed
by SO4and BC. Inspec ion o he mon hly median concen-
a ions (Fig. 3e–g), howe e , e ealed ha di e ences also
exis .
4 Lag angian analysis o o e all e ec s o in eg al
p ecipi a ion on ae osols a SMEAR II
In his sec ion we use he Lag angian amewo k o in es i-
ga e he po en ial we emo al o he ae osols. In Sec . 4.1
we i s examine he impac o using e ically esol ed liq-
uid p ecipi a ion (UKESM1 only), which has no p e iously
been done o Lag angian ajec o y analyses. Then we in-
spec he ela ionship be ween accumula ed p ecipi a ion and
ae osols o he wo GCMs used in his s udy: UKESM 1
and ECHAM-SALSA. In Sec . 4.2 we ocus on o al ae osol
mass and numbe , and in Sec . 4.3 we ocus on he OA,
BC, and SO4po ions o he o al mass o submic on-size
ae osols. Then, in Sec . 4.4, he p ocesses con olling he
p ecipi a ion-ae osol ela ionships p esen ed in he p e ious
sec ions a e in es iga ed, and he di e ences a e discussed
in de ail be ween he GCMs (Sec . 4.4.1) and wi hin each
GCM (Sec . 4.4.2). Supplemen a y analysis assesses he ep-
esen abili y o he models employed he e amongs la ge
g oup o GCMs (Sec . S4).
4.1 Assessmen o su ace s. e ically esol ed
p ecipi a ion in Lag angian we emo al
In ea lie s udies assessing ae osol-p ecipi a ion ela ion-
ships a SMEAR II using he Lag angian amewo k (e.g.,
Isokään ä e al., 2022; Khadi e al., 2023; Tun ed e al.,
h ps://doi.o g/10.5194/acp-25-14449-2025 A mos. Chem. Phys., 25, 14449–14478, 2025
14464 S. Tal inen e al.: Cloud and p ecipi a ion impac s on bo eal ae osols in GCMs
du ing win e (Fig. 7n) do occu in high ajec o y equency
a eas (Fig. 7s), hey a e mo e clea ly e lec ed in he alues
when a e aged o e all anspo di ec ions (Fig. 7o). As he
seasonal di e ences N80 in ECHAM-SALSA a e negligible,
i is unlikely ha he numbe o po en ial CCN is d i ing he
seasonal di e ences in ac i a ed ac ions and in he ae osol
mass-p ecipi a ion ela ionships in Fig. 4a. When he Nac is
inspec ed (Fig. S13 ), howe e , somewha la ge numbe o
pa icles ha e ac i a ed in win e compa ed o summe . Thus,
when conside ing he la ge di e ence in he o al numbe o
pa icles (Fig. S13e), he displayed di e ences in he ac i-
a ed ac ions (=Nac /N o ) a e easonable.
In addi ion o size, he chemical composi ion o he po en-
ial CCN also has an impac o hei ac i a ion. The composi-
ion o Ai ken and accumula ion mode ae osols in ECHAM-
SALSA (Fig. S14b) does e eal, ha he pa icles ha e el-
a i ely simila soluble accumula ion mode SO4con ibu ion
in bo h seasons. The con ibu ion o soluble OA in he ac-
cumula ion mode is sligh ly la ge in summe , bu du ing
win e , he smalle con ibu ion om OA (in accumula ion
mode) seems o be compensa ed by la ge con ibu ion om
soluble BC in he accumula ion mode. Thus, he con ibu ion
om soluble modes al oge he is ela i ely simila be ween
he seasons and unlikely causes la ge di e ences in he pa -
icle hyg oscopici y which could impac ac i a ion.
In o de o in es iga e whe he he seasonal di e ences
in he ac i a ed ac ions could also be due o sligh di e -
ences in he sensi i i y o ac i a ion o upd augh eloci ies,
we inspec ed he ela ionships be ween ac i a ed ac ions
and upd augh eloci ies simila o UKESM1. Fo ECHAM-
SALSA, he slope o summe is smalle (slope o 0.18,
Fig. 8c) compa ed o win e (slope 0.36, Fig. 8b). Thus, du -
ing win e , when he upd augh inc eases, he ac i a ed ac-
ion can inc ease wo imes as much compa ed o summe .
The e o e, despi e he simila numbe o po en ial CCN in
bo h seasons (N80, Fig. 7e), la ge po ion o hose ac i a e
du ing win e , esul ing o la ge Nac (Fig. S13 ) and ac i-
a ed ac ions (Fig. 7o). All hese indings discussed abo e
a e consis en wi h he s onge educ ion o pa icle mass
obse ed o ECHAM-SALSA in win e (compa ed o sum-
me ) in Fig. 4a. Du ing summe , e y li le o no educ ion
is obse ed o he pa icle numbe o ECHAM-SALSA in
Fig. 4b. The pa icle numbe concen a ion, howe e , is dom-
ina ed by he small ae osols which a e unlikely o ac i a e
(see also Figs. S13e and 2c). The e o e, e en wi h high ac-
cumula ed p ecipi a ion, no clea educ ion is obse ed in
Fig. 4b du ing summe .
4.4.2 Di e ences be ween GCMs and obse a ions
Compa ing he wo GCMs in Fig. 4 i is ob ious ha he
seasonali y in he ae osol-p ecipi a ion ela ionships is e-
e sed: UKESM1 exhibi s s onge educ ion du ing summe
bu ECHAM-SALSA in win e . This is unlikely a ising om
he di e ences be ween he in ensi y o he p ecipi a ion du -
ing he a el o he ai masses, as hose a e e y simila be-
ween he GCMs (Fig. S12a–e) wi hin each season. How-
e e , some o he win e di e ences may also be a ibu ed o
a ia ions in he numbe o ajec o ies wi h speci ic amoun s
o accumula ed p ecipi a ion (Fig. 4c). Obse a ions show a
highe equency o ajec o ies wi h low accumula ed p e-
cipi a ion (<2mm), whe eas he models p oduce sligh ly
mo e ajec o ies wi h la ge p ecipi a ion o als.
Du ing summe , UKESM1 has less po en ial CCN (N80,
see Fig. 7e) compa ed o ECHAM-SALSA, and also he up-
d augh eloci ies a e smalle in UKESM du ing summe ,
e en ually leading o smalle numbe o cloud d ople s oo
(Nac , Fig. S13 ). Compa ison o he con ibu ion o di e -
en chemical species in he accumula ion (as hese sizes ha e
la ge con ibu ion o he pa icle mass) mode (Fig. S14, op
ow), howe e , e eals ha UKESM1 has much la ge con-
ibu ion o he soluble pa icles. This indica es, ha du ing
summe , he pa icles in UKESM1 ha e la ge hyg oscopic-
i y, and could po en ially ac i a e mo e easily compa ed o
ECHAM-SALSA. Howe e , as he esul ing Nac (Fig. S13 )
in UKESM1 is smalle han in ECHAM-SALSA, he po-
en ially la ge hyg oscopici y in UKESM1 pa icles do no
seem o ha e signi ican impac on he d ople o ma ion.
When we conside he changes in he PNSD, howe e , whe e
UKESM1 has signi ican ly less pa icles bu wi h la ge a -
e age size compa ed o ECHAM-SALSA (which has mo e
pa icles bu smalle a e age size) as shown in Figs. 2g and
S13e, i is sensible ha la ge ac i a ed ac ions a e obse ed
o UKESM1 du ing summe as shown in Fig. 7o. The di e -
ence in he ac i a ed ac ion be ween he GCMs, howe e ,
is somewha la ge han wha could be expec ed based on he
di e ences in N o and Nac alone. Thus, also he ela ion-
ships be ween upd augh eloci ies and ac i a ed ac ions
we e inspec ed o gain u he insigh . This e eals (Fig. 8a
and c), ha indeed du ing summe , he slope be ween ac i-
a ed ac ions and upd augh eloci ies in UKESM1 is sig-
ni ican ly la ge (slope 2.12, Fig. 8a) compa ed o ECHAM-
SALSA (slope 0.18, Fig. 8c) – di e ence being o e 10- old.
This implies ha e en a small pe u ba ion in upd augh e-
loci y in UKESM1 could inc ease he ac i a ed ac ion d as-
ically, esul ing in he e y high ac i a ed ac ions obse ed
in Fig. 7o, despi e UKESM1 ha ing smalle upd augh e-
loci ies in gene al. This could indica e a shi in UKESM1
cloud d ople o ma ion om he upd augh -limi ed egime
o he ansi ional egime (e.g., Reu e e al., 2009). These
indings align wi h he s onge educ ion o pa icle mass in
UKESM1 as shown in Fig. 4a. The educ ion o he obse ed
pa icle mass in summe lies in-be ween o he wo GCMs,
ini ial educ ion (up o 5mm o accumula ed p ecipi a ion)
being mo e accu a ely ep esen ed by UKESM1.
The di e ences in he summe ime educ ion o pa icle
numbe (Fig. 4b) likely a ise om he lack o bounda y laye
nuclea ion in UKESM1, hus a ec ing he numbe concen-
a ion o he smalles ae osol pa icles (see e.g., Fig. 2g).
As al eady discussed in Sec . 4.4.1, in SMEAR II, NPF is
A mos. Chem. Phys., 25, 14449–14478, 2025 h ps://doi.o g/10.5194/acp-25-14449-2025
S. Tal inen e al.: Cloud and p ecipi a ion impac s on bo eal ae osols in GCMs 14465
Figu e 8. A e age expe ienced ac i a ed ac ion as a unc ion o a e age expe ienced upd augh eloci y along he ajec o ies. Dis ibu ion
o he alues a e shown wi h he his og ams. JJA deno es summe (June–July–Augus ) and DJF win e (Decembe –Janua y–Feb ua y).
Each colou ed poin deno es a median alue de e mined om a single ajec o y. The black lines show he eg ession line om o hogonal
eg ession applied o he da a shown and he legend show he slope, in e cep and Pea son co ela ion (R) be ween he i and he da a. No e
ha he black eg ession lines ex end o e he whole plo a ea only due o isualiza ion pu poses.
an impo an sou ce o ae osols and he equency o he
NPF e en s has signi ican seasonal a ia ion (Nieminen e
al., 2014), summe and sp ing being mos p onounced. Thus,
he educ ion o pa icle numbe in UKESM1 du ing sum-
me (Fig. 4b) is simila o he educ ion o pa icle mass
(Fig. 4a), as bo h a e domina ed by ela i ely la ge ae osols.
The summe ime educ ion o pa icle numbe in ECHAM-
SALSA coincides wi h obse a ions, which is o be expec ed
as he Ai ken and nuclea ion mode ae osol concen a ions
in ECHAM-SALSA a e much close o obse ed da a han
UKESM1 (Fig. 2g and Table S5).
Du ing win e , ECHAM-SALSA exhibi s s onge educ-
ion o pa icle mass compa ed o UKESM1 a e ∼5mm o
accumula ed p ecipi a ion (Fig. 4a). The N80 (Fig. 7a–e) is
ela i ely simila be ween he GCMs, bu upd augh eloci-
ies (Fig. 7j) ha e la ge di e ence: UKESM1 upd augh e-
loci ies ange 0.2–0.4ms−1, whe eas ECHAM-SALSA has
alues anging app oxima ely be ween 0.5–0.7ms−1. The
highe upd augh eloci ies in ECHAM-SALSA likely lead
o he la ge Nac (Fig. S14 ), hus e en ually leading o he
la ge ac i a ed ac ions o ECHAM-SALSA along mos
o he anspo (Fig. 7o) due o N o being ela i ely simila
be ween he GCMs (Fig. S13e) du ing win e . I should be
no ed, ha he di e ence in ac i a ed ac ions (Fig. 7o) a
away om SMEAR II is negligible. Howe e , his di e ence
d as ically inc eases when ai masses a el o SMEAR II:
ac i a ed ac ion in ECHAM-SALSA con inues o inc ease
while UKESM1 ac ions s ay nea ly cons an . Thus, i is
unlikely ha he simila ac i a ed ac ions a away om
SMEAR II signi ican ly impac he educ ion obse ed in
Fig. 4a.
Compa ison o he pa icle chemis y in he accumula-
ion mode in win e e eals ha he GCMs ha e (Fig. S14,
bo om ow) ela i ely simila ac ions o soluble ma e ial.
UKESM1 ends o ha e mo e SO4, bu ECHAM-SALSA
mo e soluble OA and BC. In ECHAM-SALSA, howe e , he
insoluble modes a e no s ic ly insoluble bu a he less in-
soluble compa ed o soluble modes (Sec . S2.3) and can hus
also ac i a e. This could lead o la ge Nac (Fig. S13o) and
hus la ge ac i a ed ac ion (Fig. 7o), conside ing ha he
di e ence in N o (Fig. S13e) be ween he GCMs is clea ly
smalle in win e han wha i was in summe . The di e -
ences in he ela ionships be ween ac i a ed ac ions and up-
d augh eloci ies o he GCMs (Fig. 8) a e mo e sub le in
win e (UKESM1 slope 0.62, ECHAM-SALSA slope 0.36)
compa ed o he alues in summe ime discussed ea lie . Ac-
i a ed ac ion in UKESM1 does exhibi highe “sensi i i y”
o upd augh eloci ies, howe e , due o he much la ge
h ps://doi.o g/10.5194/acp-25-14449-2025 A mos. Chem. Phys., 25, 14449–14478, 2025
14466 S. Tal inen e al.: Cloud and p ecipi a ion impac s on bo eal ae osols in GCMs
upd augh eloci ies in ECHAM-SALSA, his is likely no
enough o inc ease he ac i a ed ac ion o he same le el,
hus leading o less e icien educ ion. These assessmen s
align wi h he pa icle mass educ ions in win e shown in
Fig. 4a, whe e pa icles a ECHAM-SALSA each sligh ly
lowe end concen a ions wi h high accumula ed p ecipi a-
ion compa ed o UKESM1.
The di e ences in he win e ime educ ion o pa icle
numbe (Fig. 4b) a e less p onounced compa ed o hose in
pa icle mass (Fig. 4a). Ini ial educ ion seems o be mo e
e ec i e on UKESM1, howe e , a e ∼5 mm o accumu-
la ed p ecipi a ion, he educ ion in ECHAM-SALSA be-
comes s onge These di e ences be ween he GCMs, how-
e e , we e no s a is ically signi ican (K uskal-Wallis ank
sum es , p≥0.01). The obse a ional da a exhibi s s onge
educ ion han he GCMs du ing win e o he pa icle num-
be (Fig. 4b) up o ∼10mm o accumula ed p ecipi a ion.
A e ha , he obse a ions o e lap wi h ECHAM-SALSA.
These inconsis encies could also a ise om he ac ha bo h
GCMs ha e di icul ies ep esen ing he bimodal pa icle
numbe size dis ibu ion co ec ly du ing he win e mon hs
(Fig. 2i).
4.4.3 Addi ional easons o in e -model di e ences
Aside om di e ences d i en by ae osol ac i a ion, i is im-
po an o no e ha du ing bo h summe and win e , addi-
ional ac o s can also con ibu e o he obse ed di e ences
in he educ ions (Fig. 4). Fo example, he di e ences in he
educ ion o he pa icle mass (Fig. 4b) could be in luenced
by he plume sca enging scheme, a ea u e only p esen in
UKESM1 (see Sec . S2.4). In his p ocess, ae osol ac i a e
in o cloud d ople s wi hin he con ec i e upd augh and all
ou ia he main p ecipi a ion sha o he cumulonimbus
(Kipling e al., 2013; Mulcahy e al., 2020). No e ha e en
hough he pa icle mass is shown as a unc ion o accumu-
la ed s a i o m p ecipi a ion (Fig. 4), he ai mass ajec o-
ies ha e expe ienced con ec i e p ecipi a ion oo. Thus, e-
mo al ia nuclea ion (which is mo e e icien o la ge pa -
icles) ollowed by ainou in he con ec i e plume, could
also con ibu e. Inspec ion o he con ibu ion o he p ecip-
i a ion ypes e eals ha he con ibu ion om he con ec-
i e p ecipi a ion du ing summe is indeed sligh ly la ge in
UKESM1 compa ed o ECHAM-SALSA (Fig. S15). This
di e ence could be e lec ed in mo e e ec i e summe ime
educ ion in he pa icle mass in UKESM1. Ano he expla-
na ion o he mo e e ec i e educ ion o he ae osols du ing
summe ime in UKESM1 could be a ising om he di e -
ences in he pa ame iza ions o he e-e apo a ion o he
alling d ople s. In UKESM1, his p ocess is no consid-
e ed (see Sec . S2.3 and Mulcahy e al., 2020) whe eas in
ECHAM-SALSA e apo a ion o he d ople s can occu and
hus elease he ae osols back o he a mosphe e (e.g., S ie
e al., 2005). Du ing summe ime, his e-e apo a ion could
be enhanced due o highe empe a u es, leading o less e -
ec i e obse ed educ ion o ae osols in ECHAM-SALSA
compa ed o UKESM1. Howe e , he e can also be o he ex-
plaining ac o s, such as loca ion o he p ecipi a ion du ing
a el, emissions and d y deposi ion, which could also indi-
ec ly cause di e ences be ween he models. Quan i ying he
exac p ocesses om model pa ame iza ions causing he di -
e ences be ween he obse ed ela ionships be ween ae osol
mass and in eg al p ecipi a ion likely equi es speci ic model
sensi i i y simula ions o in es iga e his, hus being ou o
he scope o his s udy.
5 Lag angian analysis on he e ec s o aqueous
phase p ocessing on ae osol chemical
composi ion
In he analysis p esen ed in his sec ion, he ela ionship be-
ween he chemical p ocessing occu ing wi hin clouds and
ogs in he aqueous-phase is in es iga ed. A special in e es is
in aqueous-phase SO4 o ma ion due o i s high occu ence
in he a mosphe e (e.g., E ens, 2015; Huang e al., 2019;
Liu e al., 2020b). We employ a cloud p oxy based on ela-
i e humidi y (RH) along he ajec o ies simila o Isokään ä
e al. (2022). To his end, he his o y o he ai mass is in es-
iga ed, and i he RH exceeds 94 %, we assume he ai mass
is in cloud. Fu he , he ai masses we e hen sepa a ed in o
“clea sky” in which hey had no expe ience o clouds o p e-
cipi a ion du ing he las 24h, and “in-cloud” when he RH
exceeded 94% a leas a one ajec o y poin bu no p ecip-
i a ion e en s occu ed du ing he las 24h (Table S7). Only
he las 24h o he ai mass his o y we e conside ed, as wi h
longe ai mass his o ies (i.e., longe in es iga ed ime) he
numbe o s ic ly in-cloud ajec o ies dec eases due o in-
c easing possibili y o p ecipi a ion e en s. Sensi i i y es s
we e conduc ed by adjus ing bo h he RH limi ( om 90%
o 98%) and ajec o y leng h ( om 12 o 60h), bu hey did
no a ec ou conclusions. I was ound ha he ajec o y
leng h adjus men has la ge e ec on he s a is ical eliabil-
i y o he esul s, hence he in es iga ion is limi ed o he las
24h and hus also s ayed consis en wi h he p e ious in es-
iga ion in Isokään ä e al. (2022). This app oach is applied
o ERA-In e im eanalysis and o he GCM ajec o ies in
simila manne .
Reade should also no e ha UKESM1, ECHAM-SALSA
and ERA-In e im do no necessa ily ha e iden ical de ini-
ions o RH which could impac he esul s. To acknowledge
his, we also in es iga ed how well he RH along he ajec o-
ies ac ually desc ibes he in-cloud cases by compa ing his
RH-based p oxy o he co-loca ed cloud ac ion da a om
GCMs. This analysis is p esen ed in Sec . S6, and o e all,
he cloud e en s (numbe o he e en s and hei loca ions a
he ajec o ies) om bo h app oaches we e simila , leading
o simila conclusions as p esen ed in Sec . 5.1 and 5.2 be-
low. The p ecipi a ion used in he classi ica ions he e is he
o al p ecipi a ion (including bo h s a i o m and con ec i e
A mos. Chem. Phys., 25, 14449–14478, 2025 h ps://doi.o g/10.5194/acp-25-14449-2025
S. Tal inen e al.: Cloud and p ecipi a ion impac s on bo eal ae osols in GCMs 14467
p ecipi a ion), as aqueous-phase p ocesses a e aking place
no ma e he cloud ype. Rela i e humidi y da a is om he
HYSPLIT ou pu ins ead o using aw GCM/ERA-In e im
ou pu s wi h manual co-loca ion. This is because UKESM1
was ex ac ed on p essu e le els ins ead o model le els,
and he la e we e used in his wo k o he manual co-
loca ion allowing consis ency be ween o he a iables. The
seasonal di ision applied he e is based on he empe a u e,
as in Sec . 4.2. To see whe he anspo di ec ions and con-
sequen ly he p ecu so emissions ma e , da a is di ided in o
mo e clean and mo e pollu ed ai masses ( ajec o ies is-
i ing la i udes below 60° No h assigned o pollu ed sec o
as in Isokään ä e al., 2022). T ajec o y equency maps o
hese sec o s a e shown in Fig. S16.
In his sec ion, he a ia ion in he o al submic on mass
o di e en chemical species depending on he expe ienced
condi ions is i s examined and discussed o he GCMs
(Sec . 5.1) and e lec ed o obse a ions. Then, in he nex
sec ion (Sec . 5.2), a size- esol ed analysis is conduc ed o
de e mine whe he addi ional insigh in o in-cloud p ocess-
ing in GCMs could be p o ided.
5.1 E ec s o in-cloud p ocessing o o al submic on
ae osol mass
Bo h obse a ions and GCMs show highe SO4mass con-
cen a ions o cloud-p ocessed ai masses wi hin he “cold
and pollu ed” (CP) sec o (Fig. 9), consis en wi h indings
om Isokään ä e al. (2022). This pa e n holds despi e he
educed obse a ional da ase due o empo al ha moniza ion
wi h he GCMs (see Sec . 2.4). O he ai mass sec o s a e
shown in he supplemen a y ma e ial (Fig. S18).
Ac oss all ai mass sec o s, bo h GCMs ag ee well wi h
obse a ions, conside ing expec ed di e ences in he o-
al mass concen a ions. S a is ically signi ican inc eases in
SO4mass o in-cloud e sus clea -sky ai masses we e
ound in bo h obse a ions and models (p≤0.001, K uskal-
Wallis es ; Table S8), excep o he wa m and clean sec-
o (Fig. S17g– ), whe e no clea di e ence was obse ed.
As in Isokään ä e al. (2022), his may e lec limi ed SO2
a ailabili y o aqueous-phase oxida ion in cleane , wa me
ai masses. Suppo ing his, UKESM1 shows he lowes SO2
le els in clean sec o s (CC and WC; Fig. S18e), while highe
SO2in pollu ed sec o s (CP and WP) coincide wi h g ea e
SO4di e ences. Recen indings om he Holuh aun e up-
ion (Jo dan e al., 2024) also sugges aqueous-phase oxi-
da ion domina es SO2- o-SO4con e sion in GCMs. While
u u e inc eases olcanic ac i i y (Chim e al., 2023), could
enhance SO2le els and boos in-cloud SO4p oduc ion, on-
going emission con ols may educe an h opogenic SO2, po-
en ially coun e ac ing his e ec and in luencing ae osol size
and composi ion.
The obse a ions shown he e do no exhibi s a is ically
signi ican di e ences o OA be ween he clea sky and in-
cloud ai masses in any o he sec o s. The median mass o
OA in ECHAM-SALSA is la ge o he in-cloud ai masses
o he cold and pollu ed sec o (Fig. 9c and Table S8), bu
no o he sec o s exhibi s a is ically signi ican di e ences.
Howe e , his di e ence in he OA mass in he cold and
pollu ed sec o is unlikely due o o ma ion o aqSOA, as
he simula ions employed in his s udy he e did no explic-
i ly model he o ma ion o SOA. UKESM1 displays la ge
di e ences in he OA mass, in which mos a e also s a is i-
cally di e en . Howe e , he same applies as o ECHAM-
SALSA, i.e., he model simula ions do no include he o -
ma ion o SOA, and hus he di e ences mus a ise om
o he a ec ing ac o s. Bo h GCMs employ CMIP6 emis-
sion da ase s as no ed in he model se up o Ae oCom Phase
III GCM T ajec o y Expe imen , and hus he di e ences
obse ed he e unlikely a ise om a ying emissions. One
should also keep in mind ha he ep esen a ions o OA in
he GCMs migh di e , and especially hei ela ionship wi h
empe a u e, ele an d i e o SOA o ma ion in gene al,
has been shown o exhibi la ge s uc u al unce ain ies be-
ween he GCMs (Blichne e al., 2024).
Isokään ä e al. (2022) did no obse e signi ican aqueous-
phase SOA (he ea e , aqSOA) o ma ion om he obse -
a ions and his has also been no ed p e iously (G aham
e al., 2020) o simila bo eal en i onmen . Fo ma ion o
SOA om gaseous p ecu so s domina es his bo eal egion
(see e.g., Pe äjä e al., 2022), and hus dis inguishing aqSOA
om he o al o med SOA wi h ou me hodology is chal-
lenging. Fo isop ene-domina ed en i onmen s, he o ma-
ion o aqSOA is a signi ican sou ce o o al SOA bu den
(e.g., Lamkaddam e al., 2021). Also biomass bu ning emis-
sions ha e been iden i ied as a po en ial sou ce o aqSOA
(Gila doni e al., 2016; Wang e al., 2024).
I was epo ed ea lie ha he obse a ions also sugges ed
inc ease in he mass ac ion o SO4when he ai masses had
been exposed o in-cloud condi ions long enough (Isokään ä
e al., 2022). To in es iga e whe he simila beha iou could
be obse ed o he GCMs, we calcula ed he o al ime
spen unde he in luence o non-p ecipi a ion clouds om
he 96h long ajec o ies. Figu e 10 demons a es sligh in-
c eases in he mass ac ion o SO4wi h inc easing ime
spen in non-p ecipi a ing clouds o bo h GCMs. This, how-
e e , is somewha a ec ed by he da a size. I inspec ing he
GCM da a which is empo ally ha monised o he obse a-
ions (Fig. 10a–b), he conclusion is no as ob ious compa ed
o he case we e inspec ing all a ailable GCM da a (Fig. 10c–
d). This highligh s he impo ance o long enough GCM sim-
ula ions needed in his ype o Lag angian analysis u ilizing
single pa icle ai mass ajec o ies unless ensemble ajec o-
ies a e u ilised.
5.2 E ec s o in-cloud p ocessing o size- esol ed
ae osol mass
To see whe he he obse ed in-cloud o med SO4mass in
he GCMs (Fig. 9b–c) is con ibu ing o same pa icle sizes
h ps://doi.o g/10.5194/acp-25-14449-2025 A mos. Chem. Phys., 25, 14449–14478, 2025
14468 S. Tal inen e al.: Cloud and p ecipi a ion impac s on bo eal ae osols in GCMs
Figu e 9. Median (black ho izon al lines and nume ical alues) pa icle mass concen a ions a SMEAR II wi h 25 h–75 h pe cen iles (boxes)
o OA, eBC, and SO4 o he cold and pollu ed (CP) ai mass sec o . The expe ienced condi ions by he ai mass a e deno ed as clea sky
and in-cloud (non-p ecipi a ing). Subplo s include (a) SMEAR II +ERA-In e im, (b) UKESM1 and (c) ECHAM-SALSA.
Figu e 10. The mass ac ions o OA, SO4, and BC o he mo e pollu ed ai masses as a unc ion o ime spen in in non-p ecipi a ing cloud.
The op ow (a–b) shows he empo ally ha monised da a and bo om ow displays he GCM da a wi hou ha moniza ion. The igu e shows
mass ac ions de i ed om median concen a ions o each 1h bin.
as in he obse a ions epo ed in Isokään ä e al. (2022),
he analysis was epea ed he e o he GCMs. The obse -
a ions indica ed SO4mass o igina ing om aqueous-phase
p ocesses is mos ly con ibu ing o pa icles wi h diame e s
o 200–1000nm. Figu e 11 shows he pa icle mass concen-
a ions o a ious size classes de i ed om he PNSDs om
he GCMs o he clea sky and cloud p ocessed ai masses
o he cold and pollu ed sec o . The h ee o he sec o s a e
shown in Fig. S19, and Table S9 shows he esul s o he
GCMs om he s a is ical signi icance es ing be ween he
clea sky and in-cloud g oups wi hin each size class. Com-
pa ed o obse a ions, UKESM1 da a (Figs. 11a and S19)
implies he mass inc ease seems o be mos ly dis ibu ed o
bins wi h dp=100–350nm and up o 600nm in he cold
and pollu ed and cold and clean sec o s. This is likely due o
UKESM1 ha ing la ge concen a ions o pa icles in gene al
wi hin his size ange (see e.g., Fig. 2d). Like he obse a-
ions, UKESM1 does no exhibi any mass inc eases o any
o he size bins in he wa m and clean sec o (Fig. S19e), be-
ing in line wi h no obse ed inc ease in he SO4mass in he
A mos. Chem. Phys., 25, 14449–14478, 2025 h ps://doi.o g/10.5194/acp-25-14449-2025
S. Tal inen e al.: Cloud and p ecipi a ion impac s on bo eal ae osols in GCMs 14469
same sec o (WC) be ween he clea sky and cloud p ocessed
ai masses (Fig. S17h).
ECHAM-SALSA (Figs. 11b and S19), exhibi s inc eased
mass concen a ions o sizes s a ing om dp=50nm (only
in cold and pollu ed sec o ) up o 1700nm, depending on
he sec o . The la ges bin he e in ECHAM-SALSA migh
also be in luenced by dp=1–1.7µm pa icles, which a e nei-
he conside ed in UKESM1 no in he obse a ions when
inspec ing he chemical componen s (see Sec . 2.4.2). Like
UKESM1, ECHAM-SALSA also does no exhibi mass in-
c eases o any o he size bins o he wa m and clean sec o
(Fig. S21 ).
An ad an age o he GCMs used in his s udy is hei
p o ision o size- esol ed chemical composi ion, shown as
mass ac ions in Fig. S20. Fo UKESM1, inc ease in he
soluble SO4in he accumula ion mode can be obse ed
(Fig. S20a). Due o he model s uc u e, howe e , he accu-
mula ion mode i sel consis o a la ge sp ead o pa icle sizes
(dp=100–1000nm), i.e., in e nally mixed ae osols wi h ex-
e nal size modes, hus no p o iding addi ional in o ma ion
o ou PNSD based analysis. Fo ECHAM-SALSA, he o ig-
inal sec ional bins can be inspec ed (Fig. S20c) hus co e-
sponding o he PNSD bins p esen ed in Fig. 11b. All size
bins ha exhibi ed mass inc eases in Fig. 11b also exhibi
highe mass ac ion o SO4in Fig. S20c.
The obse ed changes in pa icle numbe size dis ibu-
ions (Fig. 11) e lec he ac ual model pa ame e iza ions. In
UKESM1, SO4p oduced ia aqueous-phase chemis y is al-
loca ed o he soluble accumula ion mode (dp>100nm) and
coa se mode (dp>500nm) (Mann e al., 2010), wi h he e-
sul s he e showing inc eases in he 100–600nm ange. In
ECHAM-SALSA, aqueous-phase SO4is dis ibu ed ac oss
soluble size bins spanning 50–10000 nm (2a bins; see
Table S3, Be gman e al., 2012), wi h sec o -dependen
mass inc eases obse ed be ween 50–1700 nm. In e ms o
aqueous-phase oxida ion o SO2, bo h GCMs ha e simi-
la pa ame iza ions, and o example, oxida ion o SO2by
ozone (O3) and hyd ogen pe oxide (H2O2) is conside ed in
bo h (Be gman e al., 2012; Ha dac e e al., 2021).
6 Conclusions
In his s udy we in es iga ed he e ec s o s a i o m p e-
cipi a ion (we emo al) and clouds (aqueous-phase oxida-
ion) on submic on ae osols along ai mass ajec o ies. Two
global clima e models – UKESM1 and ECHAM-SALSA –
we e analysed using a Lag angian amewo k consis en wi h
Isokään ä e al. (2022), now being seamlessly applicable o
GCMs (Kim e al., 2020). Ou geog aphical ocus was he
SMEAR II s a ion in Hyy iälä, Finland, and he su ound-
ings, ep esen a i e o he bo eal en i onmen .
Ou i s objec i e was o in es iga e whe he he
ajec o y-based ela ionships be ween ae osols mass, num-
be and p ecipi a ion a y be ween he obse a ions and he
GCMs. Fo ae osol mass, he de i ed emo al o obse a-
ions gene ally ell be ween hose simula ed by ECHAM-
SALSA and UKESM1 ac oss seasons. This indica es ha
bo h models cap u ed he obse ed mass–p ecipi a ion ela-
ionship o o al ae osol and indi idual species (OA, SO4,
BC). In con as , ae osol numbe e ealed clea model biases
ha a ied by season. In summe , UKESM1 exhibi ed a p o-
nounced loss o pa icle numbe ia p ecipi a ion compa ed
o bo h obse a ions and ECHAM-SALSA. This bias likely
s ems om he absence o bounda y laye nuclea ion, which
p oduces ewe small pa icles and lea es a la ge ac ion o
pa icles suscep ible o we emo al.
Key a iables in luencing he we emo al p ocesses, such
as numbe o po en ial cloud condensa ion nuclei (N80) and
upd augh eloci ies, we e also examined o e alua e he ob-
se ed emo als. In UKESM1, a s ong summe co ela ion
be ween ac i a ed ac ion and upd augh eloci y (Fig. 8)
may u he inc ease pa icle numbe emo al. Howe e ,
analogous s udy examining d ople numbe /CCN e sus up-
d augh (Vi anen e al., 2025) show subs an ial a iabili y
ac oss models, highligh ing ha he ela ionship. In win e ,
bo h models o e p edic ed pa icle numbe emo al ela i e
o obse a ions. This o e p edic ion may in pa e lec di -
e ences in p ecipi a ion s a is ics, wi h models simula ing
ewe low-p ecipi a ion ajec o ies (<2mm) han obse ed
(Fig. 4c). Howe e , o he ac o s such as pa icle size dis-
ibu ions, ac i a ion e iciencies, and limi a ions in he ep-
esen a ion o subg id-scale me eo ology a e also likely o
con ibu e. O e all, ou esul s emphasize he need o be e
ep esen a ion o pa icle numbe size dis ibu ions (PNSDs)
in GCMs.
Ea lie wo k has indica ed ha ae osol ac i a ion in o
cloud d ople s ollowed by ainou is he dominan we
emo al p ocess. Ou esul s suppo his, wi h UKESM1
showing nuclea ion ollowed by ainou as he la ges con-
ibu o . Supplemen a y analysis compa ing a wide ensem-
ble o GCMs indica ed ha hese wo models we e b oadly
ep esen a i e, wi h hei ae osol–p ecipi a ion ela ionships
gene ally alling nea he middle o he in e -model sp ead.
O e all, ou me hod using no malized submic on mass and
numbe as a unc ion o accumula ed p ecipi a ion p o ed o
be e ec i e in compa ing emo al ac oss models, hough i
lacks de ails on pa icle size e olu ion – an impo an opic
o u u e wo k.
Ea lie s udies (Isokään ä e al., 2022; Khadi e al., 2023)
ha e no ed ha su ace p ecipi a ion da a, commonly used in
ajec o y analyses, may no accu a ely e lec p ecipi a ion
expe ienced by ai masses a ajec o y heigh . He e, we used
e ically esol ed p ecipi a ion om UKESM1 and ound
ha su ace p ecipi a ion se es as a good p oxy in his en i-
onmen , whe e ajec o ies la gely emain wi hin he mixed
laye and s a i o m p ecipi a ion domina es. Howe e , his
analysis only conside ed liquid p ecipi a ion and may no ap-
ply o egions whe e con ec i e p ecipi a ion is mo e p e a-
len . In such en i onmen s, he e ical dis ibu ion, in ensi y,
h ps://doi.o g/10.5194/acp-25-14449-2025 A mos. Chem. Phys., 25, 14449–14478, 2025
14470 S. Tal inen e al.: Cloud and p ecipi a ion impac s on bo eal ae osols in GCMs
Figu e 11. Median (black ho izon al lines and nume ical alues) pa icle mass concen a ions wi h 25 h–75 h pe cen iles (boxes) o selec ed
size bins o (a) obse a ions wi h ERA-In e im, (b) UKESM1 and (c) ECHAM-SALSA o he cold and pollu ed (CP sec o ). Fo he la e ,
he na i e size bins a e shown (bo om ow o he legend). The expe ienced condi ions by he ai mass a e deno ed as clea sky and in-cloud
(non-p ecipi a ing).
and equency o p ecipi a ion could di e subs an ially, po-
en ially al e ing he accumula ed we emo al along ajec o-
ies. The e o e, while ou esul s a e ep esen a i e o bo eal
egions wi h s a i o m p ecipi a ion, u he wo k is needed
o assess how applicable hey a e o egions wi h di e en
p ecipi a ion egimes.
Ou second objec i e was o in es iga e whe he he GCMs
exhibi simila inc ease in sul a e mass due o in-cloud p o-
duc ion as he obse a ional da a. Bo h GCMs exhibi ed s a-
is ically signi ican di e ence in he SO4mass when ai
masses wi h only clea sky expe ience we e compa ed o in-
cloud p ocesses ai masses. The SO4mass was la ge o
he cloud p ocessed ai masses o all o he ai mass sec-
o s (based on empe a u e and di ec ion) excep he wa m
and clean ai masses, whe e GCMs showed no signi ican
di e ence be ween clea sky and in-cloud ai masses. These
esul s we e consis en wi h ea lie wo k (Isokään ä e al.,
2022). A ailabili y o he SO2 o be oxidised is likely de-
e mining whe he we see in-cloud p oduc ion o SO4, and
om UKESM1 his was u he suppo ed by heSO2con-
cen a ions and hei seasonali y. The size- esol ed analysis
e lec ed he model pa ame iza ions, he aqueous-phase SO4
being mos ly dis ibu ed in he la ge ae osol sizes.
As expec ed based on Isokään ä e al. (2022), we did no
obse e signi ican aqueous-phase SOA o ma ion. This is
likely due o he s udied en i onmen (bo eal o es ), and has
also been no ed p e iously (G aham e al., 2020) o sim-
ila bo eal o es en i onmen . Howe e , some inc eases in
OA mass we e seen in he GCMs despi e he ac ha aq-
SOA o ma ion was no explici ly modeled, possibly e lec -
ing o he p ocesses o model inconsis encies. A ecen s udy
om Blichne e al. (2024) also poin ed ou he la ge di e -
ences be ween GCMs conce ning hei OA- empe a u e e-
la ionships, which could also con ibu e o he disc epancies
obse ed he e.
O e all, bo h GCMs ep oduced he obse ed exponen-
ial dec ease in ae osol mass wi h inc easing p ecipi a ion
and showed simila cloud-p ocessing beha iou o SO4.
Ye key seasonal di e ences emain, especially in ae osol–
p ecipi a ion ela ionships and hei unde lying d i e s. A
p ima y model bias iden i ied in his s udy is he di e ence
in ae osol numbe size dis ibu ions compa ed o obse a-
ions, pa icula ly he unde ep esen a ion o small pa icles
in UKESM1. Ou esul s sugges ha disc epancies a ise
mo e om di e ences in ae osol size dis ibu ions and up-
d augh eloci ies han om he we emo al pa ame iza-
ions hemsel es. These a iables also a ec ac i a ed ac-
ions and cloud in e ac ions, and hey a e shaped by p o-
cesses beyond he 4d analysis window.
7 Ou look
While ou esul s show encou aging ag eemen be ween ob-
se a ions and GCMs in o e all ae osol–p ecipi a ion ela-
ionships, key di e ences – especially ela ed o seasonali y
and ae osol numbe – highligh he need o u he wo k. Fu-
u e s udies should in es iga e he e olu ion o ae osol size
dis ibu ions along ai mass ajec o ies in mo e de ail and
be e disen angle gas-phase and aqueous-phase sul a e o -
ma ion. Expanding analyses o egions wi h dominan con-
ec i e p ecipi a ion is also impo an , as he indings he e
a e limi ed o s a i o m, liquid-phase condi ions ypical o
bo eal en i onmen s. Including a wide ange o GCMs,
despi e he compu a ional demands, would help cla i y he
s uc u al causes behind he di e ences obse ed. Toge he ,
hese e o s a e essen ial o imp o ing he ep esen a ion o
ae osol–cloud–p ecipi a ion in e ac ions in clima e models.
A mos. Chem. Phys., 25, 14449–14478, 2025 h ps://doi.o g/10.5194/acp-25-14449-2025
S. Tal inen e al.: Cloud and p ecipi a ion impac s on bo eal ae osols in GCMs 14471
Appendix A
The lack o e ical esolu ion in he p ecipi a ion da a om
ERA-In e im eanalysis o Global Da a Assimila ion Sys em
(GDAS, (h p:// eady.a l.noaa.go /a chi es.php, las access:
3 Feb ua y 2024) in s udies using Lag angian app oaches is
now being ecognised (Dadashaza e al., 2021; Isokään ä e
al., 2022; Khadi e al., 2023). Un o una ely, e ically e-
sol ed p ecipi a ion da a om eanalysis da ase s o GCMs,
wi h high enough ime esolu ion o be use ul o ajec-
o y models, is no a commonly p o ided diagnos ic. Fo
UKESM1, his diagnos ic can be ex ac ed. He e, we con-
duc ed a compa ison be ween he e ically esol ed and su -
ace p ecipi a ion da a along he ai mass ajec o ies o in-
es iga e how well he su ace p ecipi a ion desc ibes he ac-
ual expe ienced p ecipi a ion by he ai mass. Only liquid
(s a i o m) p ecipi a ion is inspec ed, as e ically esol ed
snow all was no included in he a iable ex ac ion wi h high
enough e ical esolu ion o his model un.
Figu e A1. No malized o al (dp=3–1000 nm) pa icle mass (a) and numbe (b) a SMEAR II o summe (JJA) and win e ime (DJF) as
a unc ion o 0–25mm o accumula ed liquid s a i o m p ecipi a ion along he 96-hou long ai mass ajec o ies a he heigh o he ai
mass ( e e ed as 3D) and a he su ace ( e e ed as 2D) o UKESM1. The colou ed poin s show he median alues o each 0.5mm bin o
accumula ed p ecipi a ion when he numbe o ajec o ies in he bin was 10 o la ge . The sample size o he co esponding bins is shown
in (c).
We i s inspec ed he ela ionship be ween he no mal-
ized pa icle mass and numbe wi h he accumula ed s a i-
o m p ecipi a ion, simila o Fig. 4. This assessed whe he
ae osol–p ecipi a ion ela ionships di e be ween su ace
and e ically esol ed p ecipi a ion. Displayed in Fig. A1,
he esul s indica e he e ec s o s a i o m p ecipi a ion a
he heigh o he ai mass a e simila o he e ec s o s a i-
o m p ecipi a ion a he su ace. This is likely ela ed o he
a e age al i ude o he ai masses, as o SMEAR II hey end
o a el well below he op o bounda y laye .
To in es iga e whe he he heigh o he ai mass plays a
ole, as specula ed in Isokään ä e al. (2022), he ai mass
ajec o y al i udes we e clus e ed wi h Kmeans (e.g., Ha i-
gan and Wong, 1979) and 3 clus e s wi h dis inc heigh p o-
iles we e selec ed o u he analysis. Clus e ing each sea-
son sepa a ely p o ided simila heigh p o iles as clus e ing
o he whole da a, and hus he la e app oach is p esen ed.
h ps://doi.o g/10.5194/acp-25-14449-2025 A mos. Chem. Phys., 25, 14449–14478, 2025
14472 S. Tal inen e al.: Cloud and p ecipi a ion impac s on bo eal ae osols in GCMs
Figu e A2 shows he median al i udes o he clus e s and
he co esponding mean s a i o m ain all a es. O e all,
he mean ain all a es show simila alues despi e he p e-
cipi a ion diagnos ic. In he low-al i ude clus e (Fig. A2d),
o e all highes ain all a es (mean o e all ajec o ies and
hou s o su ace p ecipi a ion,∼0.033 mm h−1) a e ob-
se ed. In he mid-al i ude clus e , ain all a es a e smalle
(∼0.016mmh−1) compa ed o he low-al i ude clus e , and
in he high-al i ude clus e , he ain all a es a e he small-
es (∼0.010mmh−1). In he high-al i ude clus e (Fig. A2 )
mo e di e ences eme ge be ween he wo p ecipi a ion
ypes, especially a a om SMEAR II.
Each clus e was hen u he sepa a ed by season. The
median al i udes, i inspec ed sepa a ely o each season, a e
nea ly iden ical be ween he seasons wi hin each clus e , and
hus no shown he e. Figu e S21 shows he di e ences be-
ween he mean liquid ain all a es be ween su ace and e -
ically esol es s a i o m p ecipi a ion (posi i e di e ence
indica ing he ain all a es a he su ace a e highe ) o each
clus e and each season.
Figu e A2. Clus e s based on ai mass ajec o y al i udes o UKESM1. In (a)–(c) he black lines show median ajec o y al i ude as a
unc ion o ime om SMEAR II and 25 h o 75 h pe cen iles a e shown wi h he shaded a ea. The used a i al heigh a SMEAR II gi en
o HYSPLIT is indica ed wi h blue ho izon al line. The co esponding mean ain all a es a e shown in (d)–( ). Clus e s a e named based on
he maximum al i ude he ajec o y has esided du ing he las 4d. No e he di e en y-axis limi s in subplo s (a)–(c).
Du ing au umn (SON) he wo app oaches o he p ecipi-
a ion exhibi obse able di e ences only in he high-al i ude
clus e , whe e he su ace p ecipi a ion shows some o e es-
ima ion o he ac ual expe ienced p ecipi a ion by he ai
mass wi h inc easing end when mo ing a he away om
SMEAR II. This could imply ha he ai mass has spen
some ime abo e o inside he p ecipi a ing cloud, as also he
ai mass al i ude inc eases when mo ing away om he s a-
ion (Fig. A2a–c). Du ing summe (JJA), all clus e s mos ly
show p ecipi a ion a he ai mass heigh being la ge han
he su ace p ecipi a ion, expec in he high-al i ude clus e
(Fig. S21c) 72 o 96h be o e a i al o SMEAR II. As he
empe a u es du ing summe a e highe han in o he seasons,
his could be indica ion o e apo a ion as he su ace p ecip-
i a ion in UKESM1 includes only p ecipi a ion ha eaches
he su ace i.e., i is no column in eg a ed. Du ing sp ing
(MAM) and win e (DJF) he su ace p ecipi a ion shows
small o e es ima ion a some poin s along he ajec o ies,
and he di e ences a e la ges a he high-al i ude clus e –
whe e, howe e , he ain all a es a e e y small o e all (see
Fig. A2 ) o bo h p ecipi a ion ypes.
A mos. Chem. Phys., 25, 14449–14478, 2025 h ps://doi.o g/10.5194/acp-25-14449-2025
S. Tal inen e al.: Cloud and p ecipi a ion impac s on bo eal ae osols in GCMs 14473
Code a ailabili y. Da a analysis was conduc ed in R s a is ical
so wa e (R e sion 4.2.0, R Co e Team, 2022) and Py hon ( e sion
3.10.4), and colou maps o he igu es conside ing colou ision
de iciencies we e inspi ed by C ame i e al. (2020).
The sc ip s o ep oduce he main indings bo h in R
and py hon can be ound om Tal inen e al. (2025a,
h ps://doi.o g/10.5281/zenodo.16902872). Py hon sc ip s o he
da a con e sion (GCM ou pu in o ARL) and co-loca ion o he
GCM and eanalysis da a a iables o he ajec o ies can be ob-
ained om DGP.
Da a a ailabili y. Raw obse a ional da a we e collec ed by
INAR, Uni e si y o Helsinki. Field da a (pa icle numbe size dis-
ibu ions and black ca bon) a e eely a ailable om h ps://smea .
a aa.csc. i/download (las access: 20 Feb ua y 2022; Junninen e
al., 2009). The ACSM da a on ae osol composi ion a e eely a ail-
able om he EBAS da abase a h p://ebas.nilu.no/ (las access:
20 Feb ua y 2022).
The p e-p ocessed obse a ional da a, ERA-In e im and
GCMs ajec o ies along wi h he co-loca ed a iables used
in his s udy can be ound om Tal inen e al. (2025b,
h ps://doi.o g/10.5281/zenodo.15552901).
Supplemen . The supplemen ela ed o his a icle is a ailable
online a h ps://doi.o g/10.5194/acp-25-14449-2025-supplemen .
Au ho con ibu ions. DGP and AV p oposed he s udy. ST,
DGP and AV designed he esea ch ques ions. ST had he lead ole
in da a analysis wi h suppo ing con ibu ion om PK, DGP, ET
and RC. The modelling amewo k o calcula e ajec o ies om
GCM me eo ological ields was concei ed and pe o med by DGP
wi h suppo om ZK and JT. The de elopmen and applica ion
o his amewo k o he Ae oCom GCMT aj model submissions
was pe o med by PK wi h suppo om DGP. Model simula ions
and da a submissions we e pe o med by ET, DGP, TK, EH, HK,
TK, DN, DWP, YY, JZ and SK. UKESM1 model simula ion con-
igu a ion was suppo ed by AS, and ZK suppo ed he p ocessing
o ERA-In e im eanalysis da a. HYSPLIT ajec o ies we e calcu-
la ed by PK and he co-loca ion sc ip s we e de eloped by PK wi h
suppo ing con ibu ion om DGP, ET, ST and RC. Co-loca ion o
GCM da a and ERA-In e im p ecipi a ion o he ajec o ies we e
pe o med by ST. Resul s we e in e p e ed by ST, DP and AV wi h
suppo ing con ibu ion om all co-au ho s. The manusc ip was
w i en by ST wi h suppo ing con ibu ion om DP. All co-au ho s
commen ed, edi ed and ga e eedback on he manusc ip .
Compe ing in e es s. A leas one o he (co-)au ho s is a mem-
be o he edi o ial boa d o A mosphe ic Chemis y and Physics.
The pee - e iew p ocess was guided by an independen edi o , and
he au ho s also ha e no o he compe ing in e es s o decla e.
Disclaime . Publishe ’s no e: Cope nicus Publica ions emains
neu al wi h ega d o ju isdic ional claims made in he ex , pub-
lished maps, ins i u ional a ilia ions, o any o he geog aphical ep-
esen a ion in his pape . While Cope nicus Publica ions makes e -
e y e o o include app op ia e place names, he inal esponsibil-
i y lies wi h he au ho s. Views exp essed in he ex a e hose o he
au ho s and do no necessa ily e lec he iews o he publishe .
Acknowledgemen s. We acknowledge use o he Monsoon2 sys-
em, a collabo a i e acili y supplied unde he Join Wea he and
Clima e Resea ch P og amme, a s a egic pa ne ship be ween he
UK Me O ice and he Na u al En i onmen Resea ch Council.
We also hank all he people esponsible o he de elopmen o
UKESM1 and ECHAM-HAM-SALSA. The ECHAM-HAMMOZ
model is de eloped by a conso ium composed o ETH Zu ich,
Max Planck Ins i u u Me eo ologie, Fo schungszen um Julich,
Uni e si y o Ox o d, he Finnish Me eo ological Ins i u e and he
Leibniz Ins i u e o T oposphe ic Resea ch and managed by he
Cen e o Clima e Sys ems Modeling (C2SM) a ETH Zu ich.
We hank echnical and scien i ic s a om SMEAR II s a ion.
DGP would like o ex end pe sonal hanks o Ben Johnson and
Andy Jones, who p o ided suppo o he con igu a ion o he
UKESM1 simula ions pe o med as pa he Ae oCom GCM T a-
jec o y expe imen on which hese simula ions a e based. DGP also
wishes o hank Hamish S u he s who suppo ed p elimina y es -
ing o CAM simula ion ou pu du ing he de elopmen o he cod-
ing amewo k o con e GCM ields in o he equi ed o ma o
ajec o y calcula ions, and Pe e Tun ed o aluable inpu and dis-
cussions du ing he de elopmen o his amewo k.
We also wish o hank Eliza Duncan om he aluable inpu ,
echnical help and discussions du ing he de elopmen o his wo k.
Financial suppo . This wo k has been suppo ed by Eu o-
pean Union’s Ho izon 2020 esea ch and inno a ion p og amme
FORCeS (G an Ag eemen No. 821205), Ho izon Eu ope p o-
g amme ia p ojec CERTAINTY (Cloud-aERosol inTe ac ions
& hei impAc s IN The ea h sYs em, G an Ag eemen No.
101137680) and by he p ojec CleanCloud (G an ag eemen No.
101137639).
This wo k has also ecei ed suppo om he Academy o
Finland (g an No. 317373 and 317390), Academy o Finland
Flagship unding (g an No. 337550) and he Academy o Finland
compe i i e unding o s eng hen uni e si y esea ch p o iles
(PROFI) o he Uni e si y o Eas e n Finland (g an No. 325022
and 352968).
The publica ion o his a icle was unded by he
Swedish Resea ch Council, Fo e, Fo mas, and Vinno a.
Re iew s a emen . This pape was edi ed by Ann F idlind and
e iewed by h ee anonymous e e ees.
Re e ences
Aal o, P., Hame i, K., Becke , E., Webe , R., Salm, J., Makela,
J. M., Hoell, C., O’Dowd, C. D., Ka lsson, H., Hansson, H.
C., Vake a, M., Koponen, I. K., Buzo ius, G., and Kulmala,
M.: Physical cha ac e iza ion o ae osol pa icles du ing nucle-
h ps://doi.o g/10.5194/acp-25-14449-2025 A mos. Chem. Phys., 25, 14449–14478, 2025
