This documen is he Accep ed Manusc ip e sion o a Published Wo k ha
appea ed in inal o m in:
an de Ven, D.J.; Mi al, S.; Gambhi , A.; Lamboll, R.D.; Doukas, H.; Gia ola, S.;
Hawkes, A.; Koasidis, K.; Köbe le, A.C.; McJeon, H.; Pe dana, S.; Pe e s, G.P.;
Rogelj, J.; Sognnaes, I.; Vielle, M.; Nikas, A.2023. A mul imodel analysis o pos -
Glasgow clima e a ge s and easibili y challenges. Na u e Clima e Change. 13 DOI
(
10.1038/s41558-023-01661-0).
© 2023, The Au ho (s), unde exclusi e licence o Sp inge Na u e Limi ed.
This manusc ip e sion is made a ailable unde he CC-BY-NC-ND 4.0 license
h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/
A mul i-model analysis o pos -Glasgow clima e a ge s and easibili y challenges
Di k-Jan an de Ven*1, Shi ika Mi al2, Ajay Gambhi 2, Robin Lamboll3, Ha is Doukas4, Sa a Gia ola5,
Adam Hawkes5, Kons an inos Koasidis4, Alexand e Kobe le2, Haewon McJeon6, Sigi Pe dana7, Glen P.
Pe e s8, Joe i Rogelj2,3,9, Ida Sognnaes8, Ma c Vielle7, Alexand os Nikas4
1Basque Cen e o Clima e Change (BC3), Leioa, Spain
2G an ham Ins i u e o Clima e Change and he En i onmen , Impe ial College London, London, UK
3Cen e o En i onmen al Policy, Impe ial College London, London, UK
4Ene gy Policy Uni , School o Elec ical and Compu e Enginee ing, Na ional Technical Uni e si y o
A hens, A hens, G eece
5Depa men o Chemical Enginee ing, Impe ial College London, London, UK
6Join Global Change Resea ch Ins i u e, Paci ic No hwes Na ional Labo a o y, College Pa k, MD,
Uni ed S a es o Ame ica
7École Poly echnique Fédé ale de Lausanne (EPFL), Lausanne, Swi ze land
8CICERO Cen e o In e na ional Clima e Resea ch, Oslo, No way
9In e na ional Ins i u e o Applied Sys ems Analysis (IIASA), Laxenbu g, Aus ia
*Co esponding au ho : dj. ande en@bc3 esea ch.o g
Abs ac
The COP26 Glasgow p ocess esul ed in many coun ies s eng hening hei 2030 emissions
educ ion a ge s and announcing ne -ze o pledges o 2050-2070, bu i is no clea how his would
impac u u e wa ming. He e, we use ou di e se in eg a ed assessmen models (IAMs) o assess
CO2 emission ajec o ies in he nea - and long- e m based on na ional policies and pledges,
combined wi h a non-CO2 in illing model and a simple clima e model o assess he empe a u e
implica ions. We also conside he easibili y o na ional long- e m pledges owa ds ne -ze o. Whils
nea - e m pledges alone lead o wa ming abo e 2°C, he addi ion o long- e m pledges leads o
emissions ajec o ies compa ible wi h a well-below 2°C u u e, ac oss all ou IAMs. Howe e , whils
IAM he e ogenei y ansla es o di e se deca bonisa ion pa hways owa ds long- e m a ge s, all
modelled pa hways indica e se e al easibili y conce ns, ela ing o he cos o mi iga ion, as well as
o a es and scales o deployed echnologies and measu es.
Highligh s (p o isional):
- Na ional long- e m clima e pledges keep he global peak wa ming below-2°C (be ween 1.7
and 1.8°C)
- Emission in ensi y educ ion needs o accele a e pos -2030 o achie e long- e m ambi ions i
NDCs no s eng hened
- Globally implemen ed policies a e inconsis en wi h 2030 NDCs
- Feasibili y conce ns o modelled long- e m pledges a e ound o all models and egions, ye
on di e en angles o easibili y
- Dedica ed policy equi ed o o e come mapped easibili y bo lenecks o achie ing long-
e m a ge s consis en wi h he Pa is Ag eemen
- Focus on sho - and long- e m implemen a ion o announced pledges c ucial o a oid
exceeding 2°C o global wa ming.
In he Pa is Ag eemen , adop ed in 2015, coun ies ag eed o hold global mean empe a u e ise o
well-below-2°C while pu suing o limi i o 1.5°C. The scien i ic communi y has since ocussed i s
e o s on unde s anding wha i would ake o he wo ld o mee his a ge 1. Modelle s ha e
looked a equi emen s o mee he mos s ingen 1.5°C ambi ion2, quan i ying he necessa y
ans o ma ions3,4 and he consequences o li ing in a 2°C a he han a 1.5°C wo ld5. Ini ial clima e
pledges in he con ex o Pa is nego ia ions we e soon ound inadequa e6 bu , despi e he slowe -
han-necessa y pace7,8, coun ies ha e kep amping up hei ambi ion since. Se e al s udies ha e
quan i ied he ac ual impac o in e na ional policy e o s9–11, showcasing ha ‘business-as-usual’
scena ios ha a e loosely de ined and/o miss his impac o hi he o e o s a e no help ul12,13,
inc easing p essu e o be e , mo e ealis ic analysis o whe e he wo ld is heading14. A g owing
body o li e a u e15–17, hus, shed ligh on wha policies cu en ly in place as well as o icial 2030
Na ionally De e mined Con ibu ions (NDCs) would yield, es ablishing hei inadequacy o limi
wa ming o 2°C18,19.
Al hough he c i ical clima e alks o he 26 h Con e ence o he Pa ies (COP26)—a miles one o
a che ing ambi ion—we e delayed by a yea due o COVID-19, pa ies ollowed up on hei
commi men s; by he ime COP26 was comple ed in Glasgow in No embe 2021, mo e han 120
coun ies had upg aded hei 2030 a ge s20 and majo emi e s ep esen ing o e 70% o global CO2
emissions had announced and/o adop ed ne -ze o commi men s21. A hand ul o s udies a emp ed
o quickly assess he ou come o hese new p omises22–26, showing ha —i ully implemen ed—
global clima e ambi ion could hold global empe a u e ise o jus -below-2°C by 2100. Mo e e o s
o comp ehend he e ec o he new gene a ion o NDCs and long- e m a ge s (LTTs) ollowed27,28.
None heless, each o hese s udies was based on a single model and, despi e s agna ion o new such
bold p omises la gely due o he cu en ene gy c isis29, a mul i-model assessmen o global clima e
pledges emains c i ical. Ou s udy con ibu es o his esea ch gap, aiming o enhance obus ness
and con idence in ou knowledge o possible global wa ming ou comes, by exploi ing a di e se
ensemble o in eg a ed assessmen models (IAMs) and b eaking down he clima e ac ion gap in an
implemen a ion gap ( empe a u e di e ence be ween cu en policies and NDCs), a long- e m
a che gap ( empe a u e di e ence be ween NDCs ollowed o no by LTTs), and an ambi ion gap
(di e ence be ween empe a u e achie ed by LTTs and he 1.5ºC goal) (see Figu e 1 and ‘Clima e
ac ion gap de ini ion’ in Me hods).
Model di e si y also implies a plu ali y o modelling pa adigms, heo ies, solu ion mechanisms, and
hus pa hways ha each model yields30. Gi en he poli ical commi men o 1.5°C—which was u he
s eng hened in he Glasgow Clima e Pac 1—and he ambi ion e lec ed in announced ne -ze o
pledges, special emphasis is inc easingly placed on h esholds o bounda ies ha modelled pa hways
mus s ay wi hin o be conside ed easible31. Acco ding o he 6 h Assessmen Repo o he
In e go e nmen al Panel on Clima e Change (IPCC AR6), easibili y e e s o he po en ial o a
mi iga ion (o adap a ion) op ion o be implemen ed, based on di e se con ex -dependen
ac o s32—including ins i u ional, inancial, and poli ical33–35. Con e sely, om a modelling
pe spec i e, easibili y e e s o quan i iable geophysical, en i onmen al-ecological, echnological,
economic, and e en sociocul u al ac o s36. Feasibili y assessmen s hi he o e e ed o challenges a
he global le el37,38 and/o ha e been used o explici ly assess 1.5°C-complian pa hways39, wi hou
ouching upon egional easibili y conce ns o deca bonisa ion pa hways equi ed o achie e
announced pledges. Ou s udy builds on he mul idimensional easibili y assessmen amewo k
es ablished by B u schin e al. (2021)37, o iden i y whe e and when he la ges bo lenecks o
achie ing clima e a ge s can be ound, om a socioeconomic, echnological, and physical easibili y
pe spec i e (See Figu e 1 and 'Feasibili y assessmen ’ in Me hods). Howe e , a key no el y o ou
s udy lies in he expansion o he egional disagg ega ion o his amewo k, allowing o assess o
wha ex en and om which pe spec i e coun ies’ policy a ge s, NDCs, and LTTs a e easible.
Global ac ion gap
We use h ee scena ios, each co esponding o a di e en le el o clima e ac ion o s a ed ambi ion,
as announced by June 2022 (Figu e 1): a scena io wi h cu en emission educ ion policies un il 2030,
wi h pos -2030 ex apola ion main aining 2020-2030 emission in ensi y endencies (EI; see ‘Scena io
P o ocol’ in Me hods); a scena io wi h cu en NDC a ge s o 2030, wi h he same EI ex apola ion;
and a scena io wi h NDC a ge s un il 2030 ollowed by LTTs.
To compa e emissions be ween scena ios we ocus on global CO2 emissions om ene gy and
indus ial p ocesses o 2050 as all ou IAMs ep esen hese emissions sou ces as a minimum, while
we include non-CO2 emissions and sho -li ed clima e o ce s in ou empe a u e assessmen (see
below). When including all ele an na ional and egional ene gy and clima e policies on op o
socio- and echno-economic baseline assump ions, we ind ha CO2 emissions will s abilise o s a
declining in he cu en decade, eaching 33-38 G by 2030 (Figu e 2). I policy e o is su icien ly
s eng hened o each s a ed NDC a ge s, we ind ac oss models ha emissions a e educed
owa ds 2030, eaching 30-33 G . I all coun ies con inue hei declining end in emission in ensi y
o GDP beyond 2030, global emissions will achie e le els o a ound 24-30 G and 19-23 by 2050— o
cu en policies and NDCs, espec i ely. Howe e , i coun ies accele a e ac ion pos -2030 o mee
hei long- e m emission a ge s, we ind 2050 emissions in he ange o 10-13 G .
Model sp ead is la ges o cu en policies, since models un la gely in o ecas ing mode, simula ing
he impac o policies ela i e o a model-dependen no-policy baseline. Despi e ha monisa ion o
many inpu assump ions o educe unwan ed esponse he e ogenei y (see ‘Scena io P o ocol’ in
Me hods), no-policy baselines end o di e s ongly, d i en by inhe en model cha ac e is ics and
emaining unha monised inpu s18. The e o e, despi e he con e ging e ec o modelling a common
se o cu en ene gy and clima e policies, model a ia ion s ill ends o be la ge in such exe cises14,40
(Table 1). Model sp ead o emissions signi ican ly dec eases when emission a ge s om NDCs and
LTTs a e used as absolu e cons ain s. The emaining emission sp ead can be explained by a mix o
ac o s, such as model egions o e pe o ming hei a ge s, di e ences in egional agg ega ion, and
he CO2 s non-CO2 sha e in emission educ ions. While o al emissions ou comes be ween models
con e ge when applying cons ain s, he dis ibu ion o emissions o e he di e en sec o s di e ge
be ween he models, e lec ing he he e ogenei y o mi iga ion pa hways p e e ed by each model
(see Ex ended Da a Figu e 1).
While empe a u e ou comes depend on all CO2 emissions (ene gy, indus ial p ocesses, and land
use), he emaining Kyo o gases (CH4, N2O, and luo ina ed gases), and sho -li ed clima e o ce s
(SO2, black and o ganic ca bon, e c), mos o ou models only co e a subse o hose (Table 1).
The e o e, we es ima e he empe a u e implica ions o ou scena ios a e in illing missing
g eenhouse gases (GHGs) and clima e o ce s, ha monising he emission da a, and unning he
emission ou comes h ough he simple clima e model FaIR (see Tempe a u e Assessmen sec ion in
Me hods). We ind ha cu en ambi ion le els signalled h ough implemen ed ene gy and clima e
policies will inc ease global empe a u es o 2.1-2.4°C abo e p e-indus ial le els by 2100, depending
on he model (1.9 – 2.7°C when including clima e unce ain y a he 25-75% in e al), while ambi ion
le els s a ed in p esen NDCs sligh ly limi his inc ease o 2.0-2.2°C (1.7-2.5°C o 25-75% in e al).
In bo h cases, wa ming will con inue a e 2100, as global CO2 emissions will no ha e ye eached
ne -ze o le els. I coun ies also comply wi h hei s a ed LTTs a e mee ing hei cu en NDC
pledges in 2030, empe a u e inc ease will be u he limi ed and s abilise a ound 1.7-1.8°C (1.5-
2.0°C o 25-75% in e al), which is a guably in line wi h a “well-below-2°C” u u e41. Depending on
he model applied, his ansla es o an implemen a ion gap o 0.1-0.4°C addi ional wa ming on he
one hand, and a long- e m a che gap equi alen o ano he 0.2-0.5°C o wa ming on he o he . The
emaining global ambi ion gap compa ed o he Pa is a ge o keeping global empe a u e inc ease
o 1.5°C would be a ound 0.2-0.3°C o all models. Fo 3 ou o 4 models (GEMINI, MUSE, and TIAM),
he long- e m a che gap con ibu es mos o he en i e clima e ac ion gap. This con i ms p e ious
assessmen s showing ha mi iga ion in cu en NDCs is no aligned wi h long- e m a ge s o mos
coun ies24,27. Fo GCAM, he implemen a ion gap con ibu es mos o he en i e gap, ins ead.
Disagg ega ing he emission esul s o he six la ges emi e s (Figu e 4) shows whe e he di e en
gaps a e mo e ele an . The implemen a ion gap is measu ed o be la ges (in ela i e e ms) o he
USA and Japan, which ha e ela i ely ambi ious NDCs bu hei policies a e lagging behind. Fo
coun ies wi h ela i ely less ambi ious NDCs, like China, India, and Russia, he implemen a ion gap
is smalle o non-exis en , as exis ing policies o e achie e NDC a ge s in se e al cases. In he EU,
he implemen a ion gap is ela i ely small due o ambi ious policies. The long- e m a che gap is
signi ican o all cases, meaning ha , wi h cu en NDC a ge s, all six egions equi e a signi ican
boos in pos -2030 clima e ac ion o hei ne -ze o a ge s o be achie able. Howe e , di e ences
be ween models a e non-negligible, as GCAM shows an implemen a ion gap in all coun ies excep
Russia, while GEMINI only o he USA. D i en la gely by announced emissions a ge s, pa hways
owa ds long- e m a ge s a e ela i ely simila be ween he models, excep India and Russia, due o
di e en modelled o assumed con ibu ions om non-CO2 emissions and na u al sinks owa ds ne -
ze o a ge s. We ha e no assessed he ambi ion gap a he coun y le el, as ha would equi e an
assessmen o ai ness and equi y42.
Scena io easibili y
The easibili y o he modelled scena ios based on na ional policies and a ge s is measu ed by
compa ing speci ic scena io ou comes wi h p e-de e mined h esholds (see Supplemen a y sec ion
2). Su passing a h eshold indica es ha he easibili y o achie emen in ha dimension migh
become conce ning. Global esul s show ha easibili y conce ns a y s ongly be ween models,
be ween scena ios, and o e ime (Figu e 3a). Logically, deepe mi iga ion e o s imply la ge
easibili y conce ns, as hey d i e models u he away om hei emissions-uncons ained
baselines. Achie ing s a ed long- e m a ge s among all coun ies— he only op ion o keeping
empe a u e inc ease well-below-2°C (Figu e 2)—implies ha egional easibili y h esholds mus be
passed h ee- o-six imes (global weigh ed a e age and wi h median h esholds), depending on he
model and agg ega ed o e he di e en easibili y dimensions. The easibili y me ics a e only
based on mi iga ion and do no conside he adap a ion challenges ha a e d i en by lack o
mi iga ion. In ac , he easibili y o adequa e adap a ion o make up o lack o mi iga ion may be
signi ican ly mo e conce ning—i.e., in e ms o cos s, pace o in es men scale-up, and land and
eshwa e a ailabili y43,44—bu his is ou side he scope o his s udy. These easibili y conce ns can,
he e o e, be bes unde s ood as key aspec s equi ing a en ion o success ul implemen a ion o
he ambi ious mi iga ion policies.
Since he models di e signi ican ly in s uc u e, esul ing in a wide a ie y o mi iga ion pa hways,
his also ansla es o la ge a ia ions in he le el and iming o easibili y conce ns be ween he ou
models (Table 1), wi h GEMINI showing he lowes and MUSE he highes conce ns. Howe e , he
dis ibu ion o easibili y conce ns o e he di e en dimensions and o e ime a e c ucial o he
in e p e a ion. Fo example, he high o e all conce n o MUSE is la gely d i en by high ca bon
p ices and demand educ ion pa hways. In con as , hese dimensions a e ha dly o conce n o
GCAM and TIAM, whe e he pace o echnology deploymen and—in he case o GCAM— eliance on
bioene gy and CCS a e he main sou ces o easibili y conce n. When e alua ing easibili y conce ns
o e ime, all h ee TIAM scena ios s and ou o showing mos nea - e m conce ns, p edominan ly
ela ed o he high pace o wind and sola ene gy deploymen o deli e on 2030 a ge s. In GCAM,
easibili y conce ns a e ela i ely small in he un-up o 2030, bu a ise in la e pe iods, la gely d i en
by inc easing eliance on bioene gy wi h ca bon cap u e and s o age (BECCS). The la e is
conce ning om h ee di e en easibili y pe spec i es: pace o echnology deploymen , a ailabili y
o sus ainable bioene gy esou ces, and geological ca bon s o age capaci y.
Mi iga ion scena ios in his s udy a e en i ely d i en by coun y-speci ic clima e policies and
ambi ions, hence we also speci y an o e iew o measu ed easibili y conce ns o he six la ges
emi e s, a e aged o e 2020-2050 (Figu e 4). While again la ge di e ences be ween models exis ,
o e all, we see ela i ely low conce ns in China and Russia, and high conce ns in he USA, he EU,
and Japan. These di e ences a e likely d i en by mo e ambi ious a ge s in he la e g oup, which—
despi e being al eady in an emission educ ion phase o a leas a decade—s ill con on high
easibili y cons ain s o mee hei nea - and longe - e m a ge s. Speci ic easibili y issues ha
s and ou include ene gy demand educ ions in he USA (GEMINI), ca bon p icing in he EU (MUSE),
and bioene gy and ca bon s o age po en ials o Japan (GCAM). Feasibili y conce ns abou he pace
o echnology deploymen play a ela i ely small ole in China, he USA, and he EU, as opposed o
India, Russia, and Japan.
The in e p e a ion o he measu ed easibili y conce ns can be subjec i e. The p e-de e mined
h esholds a e no se in s one, and o en la ge anges o such h esholds exis in li e a u e37,38,
while h eshold le els s ongly a ec measu ed easibili y conce ns—as showcased in a sensi i i y
analysis illus a ing he impac o h eshold unce ain y on easibili y conce n alues (Figu e 3b).
Expe s in di e en ields may ha e e y di e en iews on wha is easible o no . A he same ime,
ou de ini ion o easibili y is de ined by 10 indica o s, as cons ained by ou modelling capaci y (see
Table 2 and Feasibili y Assessmen sec ion in Me hods); including o he indica o s in he analysis
migh ha e highligh ed di e en dimensions and a ec ed o e all (agg ega e) easibili y le els. Also,
coun y-speci ic ea u es, such as coun y size, s age o de elopmen , economic s uc u e, o access
o in e na ional inancial ma ke s, would in luence he h eshold le el. While some o hese ea u es
a e weigh ed in he easibili y assessmen (e.g., he h esholds o ca bon p icing depend on GDP pe
capi a le els), no all can be conside ed (e.g., economic s uc u e). Mo eo e , ou easibili y
assessmen only applies o he unde lying socioeconomic “s o yline” p o ided in he scena io
p o ocol (see ‘Scena io P o ocol’ in Me hods). Emissions associa ed wi h ou pa hways, la gely
cons ained by 2030 NDCs and ne -ze o a ge s, a e no subjec o wide a ia ion ac oss s o ylines,
as he la e would be wi hou emissions cons ain s45 (no ing ha , o India and China, he
emissions-in ensi y-based NDCs will a y acco ding o hei p ojec ed economic g ow h). Howe e ,
exis ing analysis shows ha key easibili y- ela ed indica o s o 1.5oC pa hways—including a e age
annual CO2 seques a ion om nega i e emissions, scale-up o low-ca bon p ima y ene gy sha e,
inal ene gy demand, and ca bon p ices—show conside able a ia ion ac oss s o ylines (in his case
he sha ed socioeconomic pa hways46), e en o a gi en model2. Such analysis also demons a es
ha in e -model a ia ion is compa ably wide in many cases, so a mul i-model assessmen (e en
wi h one s o yline) s ill use ully highligh s he po en ial ange o each easibili y indica o .
A inal conside a ion a ound easibili y is ha he h esholds a e no immu able laws—indeed, some
his o ical cases p o e ha he chosen h esholds can be o e come. An example is he su ge o gas-
i ed powe in he Ne he lands and nuclea powe in F ance, which espec i ely su ged om 5% o
80% and 25% o 75% o he powe mix in one decade, su passing he applied easibili y h eshold in
his s udy o e 10- old. Since such his o ical examples o as ansi ions a e ypically d i en by public
policy and suppo 47, he easibili y analysis can also be in e p e ed as a mapping exe cise o whe e
policy suppo is s ongly needed o o e come exis ing cons ain s, which is c ucial o achie ing
s a ed clima e a ge s as all models and scena ios in his s udy su pass se e al easibili y h esholds.
Ne e heless, as he esul s show, his mapping s ongly depends on he applied model: deep
s uc u al di e ences be ween models lead o a wide a ie y o pa hways eaching he same clima e
a ge s and, hence, di e en policy in e en ions a e necessa y om di e en modelling
pe spec i es o make hese pa hways easible.
Discussion
Ou esul s sugges ha , i announced na ional nea - (2030) and longe - e m (2050-2070) emission
educ ion ambi ions h oughou he wo ld a e achie ed, global peak empe a u e inc ease will s ay
below 2°C wi h ~75% ce ain y. Howe e , i clima e ac ion is no s eng hened pos -2030, long- e m
ambi ions will no be achie able (long- e m a che gap), and global empe a u e inc ease will be
a ound o abo e 2°C by 2100. All applied models ag ee ha —wi h he cu en pace o policy
implemen a ion— empe a u e inc ease will no exceed 2.5°C o e he cou se o he cen u y bu will
s ill ha e a ising end he ea e . These esul s clea ly show ha clima e ac ion and ambi ions ha e
no ably imp o ed since 2020, when he same se o models and scena io s uc u e p ojec ed 2030
emissions o be on a e age 3 G CO2 highe (see Ex ended Da a Figu e 2) and mean global
empe a u e 0.2-0.3°C wa me by 210018. This imp o emen , alongside ou inding ha global
emissions a e se o peak in he cu en decade wi h cu en NDCs, is in line wi h ea lie pos -COP26
assessmen s22–24,27,28. The la e assessmen s, howe e , also ound inc easing emissions o cu en
policy scena ios h oughou and a e he 2020s22,27,28, whe eas all models in ou s udy ind
educ ion o s abilisa ion o emissions wi h cu en policies wi hin his decade. Fu he mo e, using a
di e se se o di e en IAMs and hus co e ing a la ge pa o s uc u al unce ain y while educing
undesi ed model esponse he e ogenei y, ou s udy o e s a mo e obus assessmen o emissions
and wa ming implica ions o pos -Glasgow pledges and ac ion, compa ed o hese ea lie single-
model assessmen s.
Accoun ing o pos -2030 ex apola ions o cu en policies and NDCs, his s udy shows signi ican ly
lowe emissions and end-o -cen u y empe a u es han o he s udies; addi ionally, ou mul i-model
LTT p ojec ions a e sligh ly mo e op imis ic (by ~0.1°C). An impo an eason o his di e ence likely
lies in he applied ex apola ion me hod: in con as o a con inued end in emission in ensi ies
used in his s udy, o he s udies applied a ca bon p ice equi alen o 2030 ac ion and inc easing o e
ime wi h GDP le els23,27,28, a me hod leading o mo e conse a i e emission educ ions in mos
models18. The e is no s aigh o wa d answe on which policy ex apola ion me hod is be e : while
con inuing a end in emission educ ions may alsely bank on an emission educ ion end ha
migh no be equally a ainable in he u u e, elying on ex apola ed ca bon p ices may pu oo
much ai h in highly unce ain u u e model assump ions, especially conside ing ha , e.g., he
decline in cos s o low-ca bon echnologies ha e adi ionally been unde es ima ed in such
assump ions12. Ano he impo an di e ence wi h se e al o he model s udies in he li e a u e is
ha cu en policies a e modelled explici ly (i.e., no p oxied ia ca bon p ices un il 2030, see
Supplemen a y Da a 1) and on op o an up- o-da e, ha monised se o socio- and echno-economic
assump ions. This con ibu es o he ela i ely op imis ic assessmen o 2030 emissions in he s udy,
as i endogenously assumes a che ing o emission educ ion a ge s (by o e pe o ming on hese
a ge s), whe e hese a e no ambi ious.
Ou scena io s uc u e e lec s h ee le els o ambi ion, whe e o each sepa a e model egion
emissions a e equal o lowe wi h each subsequen le el o ambi ion. Simila ly, unce ain y o each
he modelled emission educ ion also inc eases wi h each subsequen le el o ambi ion. Hence,
despi e agg ega ed na ional ambi ions being in line wi h a well-below-2°C empe a u e u u e, he
likelihood o all hese na ional ambi ions being achie ed should no be o e es ima ed. Fo each
indi idual coun y ha ails o mee i s a ge s, he p obabili y o highe empe a u e le els
inc eases. Ou easibili y analysis shows ha , o mos coun ies, i is a om s aigh o wa d o
achie e hei s a ed nea - and long- e m ambi ions wi h he an icipa ed u u e demand pa e ns and
echnologies. The easibili y analysis in his s udy shows ha he e is no “ ee lunch” in e ms o
easibili y o mi iga ion pa hways: while di e en model s uc u es imply he e ogenous pa hways o
deca bonise economies in line wi h p oposed pledges, each o he pa hways compa ible wi h a well-
below-2°C empe a u e u u e aces signi ican easibili y challenges, ei he in socioeconomic,
echnology scale up, physical o sus ainabili y dimensions. Howe e , such easibili y challenges
should no be in e p e ed as ha d ba ie s o mee pledged mi iga ion a ge s, bu a he as a eas
whe e addi ional policy suppo o b eak h oughs in echnology o consume beha iou 48 may be
needed o o e come such challenges.
Ou esul s show ha agg ega ing all na ional nea - and long- e m emission educ ion a ge s is s ill
insu icien o limi global empe a u e inc ease o 1.5°C (long- e m ambi ion gap), which is he
highes ambi ion o he Pa is Ag eemen . E en i he mos ambi ious scena ios in his s udy a e
achie ed, he empe a u e inc ease may s ill cause signi ican ly no able and damaging clima e
impac s49, and be su icien ly s ong o ac i a e se e al clima e ipping elemen s, such as he
collapse o he G eenland and Wes An a c ic ice shee and he die-o o low-la i ude co al ee s50,51.
The e o e, e en i ambi ions ha e signi ican ly imp o ed in he un-up o and sho ly a e COP26,
cumula i e emissions un il 2050 mus be educed signi ican ly mo e o a oid o e shoo ing he 1.5°C
a ge , o subs an i e nega i e emissions should be achie ed a e eaching LTTs o each 1.5°C by
2100 wi h a high o e shoo (see Ex ended Da a Figu e 3). Ne e heless, he esul s o his s udy
show ha , while he ocus on u he ambi ion a che ing should no be los 8, a high ocus on he
sho - and long- un implemen a ion o he exis ing se o ambi ions is signi ican ly mo e impo an
o a oid a clima e disas e .
Acknowledgemen s
D.V., S.M., A.G., H.D., S.G., A.H., K.K., A.K., S.P., G.P., J.R., I.S., M.V. and A.N. acknowledge suppo
om he H2020 Eu opean Commission P ojec PARIS REINFORCE (g an no. 820846). R.L.
acknowledges suppo om he H2020 Eu opean Commission P ojec PROVIDE (g an no.
101003687). D.V., S.M., A.G., H.D., S.G., A.H., K.K., G.P., I.S. and A.N. also acknowledge suppo om
he Ho izon Eu ope R&I p og amme p ojec IAM COMPACT (g an no. 101056306).
Au ho con ibu ion s a emen
D.V., S.M., A.G., G.P., J.R. and A.N. coo dina ed he s udy design and scena io p o ocol; all au ho s
we e in ol ed in he model analysis, wi h no able con ibu ions om D.V., H.M. (GCAM), S.P., M.V.
(GEMINI), S.G., A.H. (MUSE), S.M., A.G., A.K. (TIAM) and R.L. (Silicone and FaIR). D.V., S.M., A.G., G.P.,
I.S. and A.N. compiled and analysed he esul s and c ea ed he igu es, wi h eedback om all o he
au ho s. D.V. coo dina ed he concep ion and w i ing o he pape wi h no able con ibu ions om
A.N., K.K., S.M. and A.G. and eedback and con ibu ions om all o he au ho s.
Compe ing in e es s s a emen
The au ho s decla e no compe ing in e es s.
Table 1: Model key cha ac e is ics and mapping o cha ac e is ics o majo easibili y conce ns
GCAM-PR
GEMINI-E3
MUSE
TIAM-G an ham
Model ype
Pa ial equilib ium
Gene al equilib ium
Agen -based
ene gy-sys em
Pa ial equilib ium
Solu ion dynamic
Recu si e-dynamic
Recu si e-dynamic
Recu si e-dynamic
In e - empo al
op imisa ion
(pe ec o esigh )
Technology choice
Logi choice
Nes ed CES
unc ion
Agen decision
goals and s a egies
Winne akes all
GHG emission
co e age ( epo ed)
Fossil CO2, CH4,
N2O, Land-use CO2,
F-gases
Fossil CO2, CH4,
N2O, F-gases
Fossil CO2
Fossil CO2
Model egions
32
11
20
15
Time ho izon
2100
2050
2100
2100
Uncons ained
baseline CO2 1
High
High
Low
Medium
Impac o cu en
policies 1
Medium
High
Low
High
Majo easibili y conce ns wi h LTTs2:
Timing and indica o
Long- e m
bioene gy and
ca bon s o age
Long- e m ene gy
demand educ ion
Nea and long- e m
ca bon p icing
Nea e m
echnology scale-up
Region(s)
India, Japan
USA
EU, Japan
EU, India, Japan
Explana ion based
on model s uc u e
Due o he
endogenous
ep esen a ion o
he land sec o , no
ha d limi s a e se
o bioene gy
supply. High ene gy
p ices s imula e
bio-ene gy ou pu
om land beyond
sus ainable limi s.
As a gene al
equilib ium model,
he en i e economy
is simula ed,
including economic
eedbacks o end-
use sec o s. High
ene gy p ices
he e o e lead o
ela i ely high
demand educ ion
Due o high ine ia
by modelled agen s
and echnology
s ickiness, high
ca bon p ices a e
equi ed o swi ch
o low ca bon
echnologies
and/o educe
demand.
As an in e -
empo al pe ec
o esigh model,
agen s ha e pe ec
o esigh owa ds
he u u e, d i ing
he nea - e m
in es men in
enewable
echnologies.
1 Fo baseline CO2, High > 40 G CO2, low < 30 G CO2, Medium = 30-40 G CO2 by 2050. Fo impac om cu en
policies, High > 6 G CO2, Low < 3 G CO2, Medium = 3-6 G CO2 educ ion o baseline in 2030.
2 Majo easibili y conce ns a e iden i ied sepa a ely pe model, and no by compa ing easibili y conce ns
be ween he di e en models.
38. Gambhi , A. e al. Assessing he Feasibili y o Global Long-Te m Mi iga ion Scena ios. Ene gies
10, 89 (2017).
39. Wa szawski, L. e al. All op ions, no sil e bulle s, needed o limi global wa ming o 1.5 C: A
scena io app aisal. En i on. Res. Le . 16, 64037 (2021).
40. Gia ola, S. e al. Challenges in he ha monisa ion o global in eg a ed assessmen models: A
comp ehensi e me hodology o educe model esponse he e ogenei y. Sci. To al En i on.
783, 146861 (2021).
41. Rogelj, J. e al. Mi iga ion pa hways compa ible wi h 1.5°C in he con ex o sus ainable
de elopmen . in Special Repo on he impac s o global wa ming o 1.5 °C
(In e go e nmen al Panel on Clima e Change, 2018).
42. Robiou du Pon , Y. e al. Equi able mi iga ion o achie e he Pa is Ag eemen goals. Na . Clim.
Chang. 7, 38 (2016).
43. Pö ne , H.-O. e al. Clima e change 2022: Impac s, adap a ion and ulne abili y. IPCC Six h
Assess. Rep. (2022).
44. Ma kandya, A., Gala aga, I. & Mu ie a, E. S. De. Rou ledge handbook o he economics o
clima e change adap a ion. (2014).
45. Ma angoni, G. e al. Sensi i i y o p ojec ed long- e m CO2 emissions ac oss he Sha ed
Socioeconomic Pa hways. Na . Clim. Chang. 7, 113–117 (2017).
46. O’Neill, B. C. e al. A new scena io amewo k o clima e change esea ch: he concep o
sha ed socioeconomic pa hways. Clim. Change 122, 387–400 (2014).
47. Fouque , R. His o ical ene gy ansi ions: Speed, p ices and sys em ans o ma ion. Ene gy
Res. Soc. Sci. 22, 7–12 (2016).
48. Pe dana, S. e al. Expe pe cep ions o game-changing inno a ions owa ds ne ze o. Ene gy
S a eg. Re . 45, 101022 (2023).
49. Hoegh-Guldbe g, O., Jacob, D., Taylo , M., Bindi, S. & Zhou, G. Impac s o 1.5°C Global
Wa ming on Na u al and Human Sys ems. in Global Wa ming o 1.5°C 175–312 (Camb idge
Uni e si y P ess, 2018). doi:10.1017/9781009157940.005.
50. A ms ong McKay, D. I. e al. Exceeding 1.5°C global wa ming could igge mul iple clima e
ipping poin s. Science (80-. ). 377, (2022).
51. Len on, T. M. e al. Clima e ipping poin s — oo isky o be agains . Na u e 575, 592–595
(2019).
52. Hoesly, R. M. e al. His o ical (1750--2014) an h opogenic emissions o eac i e gases and
ae osols om he Communi y Emissions Da a Sys em (CEDS). Geosci. Model De . 11, 369–408
(2018).
53. Bye s, E. e al. AR6 Scena ios Da abase. (2022) doi:10.5281/ZENODO.5886912.
Me hods
Models included
Fou global in eg a ed assessmen models a e included in his esea ch: GCAM-PR (also e e ed o
as GCAM), GEMINI-E3 (also e e ed o as GEMINI), MUSE, and TIAM-G an ham (also e e ed o as
TIAM). These a e selec ed o e lec he b oad di e si y o modelling heo ies, spanning a ange om
leas -cos ene gy sys em op imisa ion o pa ial and gene al equilib ium and o agen -based
modelling. Di e si y o modelling s uc u e, heo y, and solu ion is ypically sough in mul i-model
s udies, aiming o each obus es ima es by e lec ing s uc u al unce ain y— a he han
pa ame ic unce ain y, which has been minimised o educe unwan ed esponse he e ogenei y40.
GCAM54 (Global Change Analysis Model) is a pa ial equilib ium IAM, achie ing equilib ium be ween
ene gy supply and demand in each ep esen ed sec o , accoun ing o he changes in ene gy p ices
esul ing om changes in uels and echnologies used o sa is y ene gy-se ice demands in hese
sec o s. The model ope a es on a ‘ ecu si e dynamic’ cos -op imisa ion basis and sol es o he
leas -cos ene gy sys em (cons ained by obse ed echnological p e e ences) in a gi en pe iod
be o e mo ing on o he nex pe iod and pe o ming he same p ocess.
TIAM55 (Times In eg a ed Assessmen Model) is also a pa ial equilib ium IAM and achie es simila
equilib ium be ween ene gy supply and demand in each sec o . Howe e , TIAM ope a es on a
‘pe ec o esigh ’ wel a e cos -op imisa ion basis, whe eby all consequences o echnology
deploymen s, uel ex ac ion, and ene gy p ice changes o e he en i e ime ho izon a e conside ed
when minimising he cos o he ene gy sys em o p o ide ene gy-se ice demands wi hin speci ied
emissions cons ain s.
GEMINI-E356 (Gene al Equilib ium Model o In e na ional–Na ional In e ac ions be ween Economy,
Ene gy and he En i onmen ) di e s in model solu ion in ha i is a gene al equilib ium IAM,
ea u ing a mo e de ailed, mul iple-sec o ep esen a ion o he economy ha conside s how he
impac s o speci ic policies sp ead ac oss economic sec o s and egions a ec en i onmen al
pa ame e s. This means ha , despi e also being d i en by ma ke equilib ium, his equilib ium is
assumed o ake place simul aneously in each ma ke / egion. I s iche ep esen a ion o he
economy equi es calib a ion o da a on na ional and in e na ional socio-accoun ing in o ma ion and
a ec o o a ious elas ici ies o subs i u ion, bu i allows endogenous calcula ion o ma ke p ices
o inpu s and ou pu s.
Finally, MUSE57 (ModUla ene gy sys ems Simula ion En i onmen ) is an agen -based, ene gy sys em
model ha p o ides a de ailed accoun o he ene gy sec o o calcula e leas -cos GHG emissions
educ ion pa hways—o he cos s o al e na i e clima e policies. I is bo om-up, in ha assumes
sho - e m mic oeconomic equilib ium on he ene gy sys em by i e a ing ma ke clea ance ac oss all
sec o modules and in e changing p ice and quan i y o each ene gy commodi y in each egion, bu
i is also agen -based, in ha i ies o de e mine a mi iga ion pa hway by p o iding an as- ealis ic-
as-possible desc ip ion o he in es men and ope a ional decision making in each geog aphical
egion wi hin a sec o .
All ou models di e in he way echnologies a e chosen ac oss sec o s: GCAM employs a logi
echnology choice mechanism, which causes g adually dec easing e u ns as a echnology is u he
di used; TIAM uses a winne - akes-all op imisa ion mechanism, implying ha he cheapes
echnology can domina e all new deploymen ; GEMINI uses a nes ed cons an elas ici y o
subs i u ion unc ion; while MUSE ollows an agen -based app oach, whe e agen decision goals and
s a egies de e mine echnology choices in each ime s ep. De ailed model documen a ion o all
ou models is a ailable online a h ps://www.i2am-pa is.eu/de ailed_model_doc.
Scena io p o ocol
S a ing his modelling exe cise, ha monised socio- and echno-economic inpu assump ions we e
applied by all models, e lec ing he la es a ailable in o ma ion and a oid “noise” in he model
ou comes ela ed o unaligned assump ions40. Fo GDP p ojec ions, he IMF WEO o Ap il 202258 o
GDP g ow h un il 2027 he OECD EO-109 (2021)59 o pos -2027 g ow h p ojec ions, e lec ing he
impac s o he COVID-19 pandemic as well as ini ial es ima ed impac s ela ed o he Uk aine
con lic . On echno-economic assump ions, powe gene a ion echnology cos s we e upda ed o
obse ed 2020 alues (IRENA) while main aining he u u e e olu ion o cos s as e lec ed in Gia ola
e al (2021)40. Fo hyd ogen, p ojec ions o di e en p oduc ion echnologies we e upda ed
acco ding o IEA es ima es (2017)60. I should be no ed ha , despi e conside able e o s o
ha monise model inpu s, he ou IAMs do no all ep esen he same po olio o echnologies; his
hampe s he e o s o educing unwan ed he e ogenei y o esponses and o a ibu ing he
esul ing model sp ead only o s uc u al unce ain y. Howe e , ou mul i-model assessmen
emains use ul in ha i p o ides an implici assessmen o he a ie y o pa hways ha could esul
no jus om s uc u al di e ences, bu also om di e en assump ions a ound he a ailabili y o
key echnologies (e.g., di ec ai cap u e61,62).
The i s scena io (Cu en Policy ex apola ed wi h Emission In ensi y, CP_EI) is based on he cu en
po olio o ac ual emission educ ion policies as well as c edible policy a ge s un il 2030 in G20
coun ies including he en i e Eu opean Union (EU) (see Supplemen a y Da a 1). Pos -2030 ac ion is
hen modelled by measu ing he a e age a e o change in emissions in ensi y o GDP om 2020 o
2030 in each egion and assuming emissions in ensi y educ ion a es will emain he same a e
2030. This me hod is also used by Ou e al (2021)22, Sognnaes e al (2021)18 and VanDyck e al
(2016)15 o assess he long- e m implica ions o NDCs. The applied policy a ge s un il 2030 (e.g.,
enewable ene gy mx a ge s, ehicle uel s anda ds) a e main ained as minimum le els beyond
2030 o a oid back acking o achie ed policies.
The second scena io (NDCs ex apola ed wi h Emission In ensi y, NDC_EI) is based on s a ed 2030
emission a ge s cap u ed in NDCs submi ed o announced by June 2022, cap u ing all mi iga ion
ambi ion upda es ela ed o he COP26 in Glasgow (see Supplemen a y Table 1). These NDC a ge s
a e applied on op o cu en policies (CP) modelled in he p e ious scena io; in model egions whe e
cu en policies o e achie e on he mi iga ion a ge s in NDCs, no addi ional emission cons ain s
a e applied, ollowing Sognnaes e al (2021)18. Emissions educ ions in NDC scena ios a e he e o e
ne e less ambi ious han wha CP implies. The same emission in ensi y me hod is applied o pos -
2030 ac ion as in CP_EI.
The hi d and mos ambi ious scena io (NDCs wi h Long-Te m Ta ge s, NDC_LTT) is buil on he
NDC_EI un il 2030 bu , o egions ha exp essed an LTT, such as ne -ze o commi men s o o he
a ge s o 2050 o la e (ei he in law, policy documen s, o only announced) (see Supplemen a y
Table 2), emission cons ain s a e applied ha linea ly decline om 2030 emissions as in he NDC_EI
scena io owa ds said long- e m a ge . Fo egions wi hou LTTs, pos -2030 emissions ollow an
iden ical pa h as in he NDC_EI scena io.
Since nea ly all coun ies ha e submi ed a leas some NDC a ge , de ining NDC a ge s o
agg ega ed model egions is ela i ely s aigh o wa d. Howe e , a bu all coun ies ha e
submi ed LTTs, hence some assump ions a e equi ed i one o mo e coun ies in an agg ega ed
model egion ha e LTTs. In such cases, he emissions le el (E) should be calcula ed by applying he
LTT o he es ima ed emissions sha e o ha speci ic coun y (i) in he en i e model egion (j)
acco ding o ei he he 2019 emissions le els52, o , i a ailable, he coun y´s emissions sha e in he
agg ega ed NDC a ge o 2030, and applying he EI me hod o he es o he egion:
𝐸𝑗,2050 =𝐸(𝐿𝑇𝑇)𝑖,2050×(𝐸𝑖,2019 / 2030
𝐸𝑗,2019 / 2030)+𝐸(𝐸𝐼)𝑗−𝑖,2050∗((𝐸𝑗,2019 / 2030−𝐸𝑖,2019 / 2030)
𝐸𝑗,2019 / 2030 )
Tempe a u e assessmen
To assess he implica ions o he modelled scena ios o global mean empe a u e inc ease, emission
ou comes om he models a e ha monised wi h his o ical endencies, in illed o include he ull se
o g eenhouse gas emissions, and ed in o a clima e model o p obabilis ic empe a u e simula ions.
In mos ins ances, he i s s age o he empe a u e assessmen is o ha monise he global
emissions ajec o ies o known alues in 2015 (in e pola ed i no al eady p esen ) using a io-
based ha moniza ion app oach63. Since no all models epo he en i e se o GHGs and o he
pollu an s equi ed o a comple e empe a u e assessmen , un epo ed emissions om each model
pa icipa ing in his in e modal compa ison exe cise a e in illed using Silicone 1.3.064, using a
quan ile olling window wi h CO2 emissions om ene gy and indus ial p ocesses as he lead
emissions, and based on an in illing da abase comp ised o he ha monised AR6 da abase53 il e ed
o ma ch he model philosophy. The models ha a e included in he AR6 da abase a e ca ego ized
based on hei model ype ( o example gene al equilib ium/ pa ial equilib ium) and solu ion ype
( ecu si e dynamics/ in e empo al). The excep ion o his is he F-gases (SF6, HFCs and PFCs),
which a e no epo ed by enough models in each ca ego y. Fo hese cases (whe e no epo ed
o he wise), we use he F-gas o al in illed as abo e, hen b eak i down in o i s componen SF6, HFC
o al and PFC o al using he whole ha monised AR6 da abase and he Silicone echnique
DecomposeCollec ionTimeDepRa io. Fo he GEMINI-E3 model simula ing global economy dynamics
o e he ime ho izon 2015 o 2050, we ex end each o he emissions ill 2100 o each scena io
using he Silicone ool Ex endLa es TimeQuan ile, using he whole AR6 da abase. Plo ing he in illed
ajec o ies o Kyo o gases ins ead o ossil CO2 p oduces e y simila esul s, as ossil CO2 co ela es
well wi h he Kyo o gas o al in he AR6 da abase (see Supplemen a y Figu e 3), and ou in illing
echnique p ese es he co ela ion be ween he modelled gas and all he cons i uen s.
When we ha e a comple e se o equi ed emissions, hey a e un h ough he simple clima e model
FaIR e sion 1.6.2, calib a ed o ma ch he AR6 Wo king G oup 1 clima e assessmen 65,66. This ou -
box model o he wo ld eplica es he impac o emissions on a mosphe ic concen a ions, clima e
o cings and empe a u es, cons ained bo h agains obse a ions and he p obabili y dis ibu ions
o undamen al clima e cha ac e is ics like TCR assessed by he IPCC.
Supplemen a y Figu e 4 shows he median empe a u e assessmen s un il 2100 om FaIR, while
also showing he unce ain y in his empe a u e assessmen ela ed o in illing he emission
ajec o ies using Silicone.
Clima e ac ion gap de ini ion
Compa ing scena io ou comes allows us o subdi ide he clima e ac ion gap—i.e., he di e ence
be ween he emission educ ions and ela ed empe a u e ou comes ha can be expec ed wi h he
cu en se o policies in all coun ies, wi h he goal o keeping global empe a u e inc ease below
1.5°C. I is wi h hese wo ajec o ies (cu en policies as in “whe e we s and” and 1.5°C as in
“whe e we wan o go”) and he wo in e media e ajec o ies (NDCs as in “ambi ion e lec ed in
nea - e m a ge s” and LTTs as in “ambi ion e lec ed in long- e m a ge s”) ha we de ine he
di e en gaps in his s udy. The i s gap, he eby e med ‘implemen a ion gap’, e e s o he
di e ence in 2100 o peak empe a u e (depending on whe he a peak is eached in he 21s
cen u y) o cu en policies and ha o 2030 NDCs, bo h ex ended by EI ends. The second gap,
he eby e med ‘long- e m a che gap’, e e s o he empe a u e di e ence be ween he 2030
NDCs ex ended by EI ends, on he one hand, and he 2030 NDCs ollowed by LTTs (whe e s a ed),
on he o he —in o he wo ds, i e e s o he pace, in which pos -2030 ac ion mus be accele a ed
ela i e o p e-2030 ac ion o deli e on long- e m a ge s. The inal gap, he eby e med ‘ambi ion
gap’, e e s o he di e ence be ween he peak empe a u e o 2030 NDCs ollowed by a ailable
LTTs and he 1.5°C a ge . These h ee gaps, al oge he making up he clima e ac ion gap, a e
illus a ed in de ail in Figu e 1 and a e no o be con used wi h he UNEP de ini ion o ‘emissions
gap’19; he la e e e s o he emissions di e ence be ween he p omised educ ions (as in NDCs
and/o LTTs) and he needed educ ions (as in leas -cos pa hways deli e ing 1.5°C), which we do
no calcula e.
Feasibili y assessmen
This s udy looks in o he easibili y o pa hways based on coun y-speci ic policies and announced
a ge s, wi h he objec i e o iden i y “whe e” (which coun y and sec o ) and “when” (which
decade be ween 2020 and 2050) we ind he la ges bo lenecks o achie ing hem. This easibili y
analysis builds la gely on he B u schin e al (2021)37 amewo k, measu ing easibili y conce ns by
compa ing speci ic model ou comes wi h h eshold alues ound in he li e a u e. I also de ines
easibili y as in ha amewo k, i.e., as he deg ee o which a scena io lies wi hin he bounda ies o
di e se socie al capaci ies o change in a gi en pe iod. Howe e , he e lec ed dimensions a e
la gely cons ained by he capaci y o quan i y wi h all models used in his s udy, while o e lapping
dimensions a e a oided o allow a ai compa ison o easibili y conce ns be ween models.
The easibili y analysis looks a speci ic a iables in model ou comes and compa es hese wi h
se e al h esholds ound in li e a u e. A o al o 10 di e en easibili y indica o s a e measu ed,
which can be di ided in o 3 ca ego ies: (a) socioeconomic easibili y conce ns ela ed wi h he cos
bu den o mi iga ion policies, (b) echnology scale-up easibili y conce ns ela ed wi h he eloci y a
which clean echnologies eplace exis ing echnologies in place, and (c) physical easibili y
cons ain s ela ed wi h he physical po en ials o bioene gy p oduc ion and ca bon s o age.
The e o e, ou analysis does no include bo om-up socio-poli ical dimensions ha canno be
quan i ied in (all) ou models, and ou de ini ion o easibili y should no be in e p e ed as b oadly as
de ined in li e a u e67,68 bu de ined by he modelled dimensions conside ed—hence, we discuss
‘ easibili y conce ns’ a he han easibili y.
Feasibili y conce ns a e measu ed by model egion and 10-yea pe iod (2020-2030, 2030-2040 and
2040-2050) o illus a e “whe e” and “when” we ind he la ges bo lenecks o clima e change
mi iga ion. Iden ically o all indica o s, a alue o he size o 1 ep esen s ha he speci ic h eshold
o ha indica o is su passed by 100% in ha speci ic model egion and pe iod. Simila ly, sco es o
0.5 and 1.5 mean su passing he h esholds by espec i ely 50% and 150%. Tha also means ha , as
long as he h eshold is no passed, e en i i comes close, he easibili y conce n is measu ed as
ze o. Simila ly o all indica o s, he global alue is buil up as a weigh ed a e age o he independen
alues in all model egions. The weigh ing a iable, howe e , a ies be ween he di e en indica o s
(see Table 2). Fo mo e de ails on how he di e en easibili y indica o s a e app oached, on he
p ecise h eshold le els as well as he sou ces hese le els a e aken om, see Supplemen a y
Sec ion 2. Fo he p ecise easibili y conce n le els unde cen ally assumed h eshold as well as
unde h eshold unce ain y (Figu e 3b), see sou ce da a o igu es 3 and 4.
Da a a ailabili y
The da ase s gene a ed du ing, and analysed in, he cu en s udy a e a ailable om a public
eposi o y (h ps://doi.o g/10.5281/zenodo.7767193).
Code a ailabili y
The code a ailabili y o he indi idual models used in his pape a ies and con ac should be made
o indi idual modelling g oups. The GCAM model is a ailable o download om
h ps://gi hub.com/JGCRI/gcam-co e. The code o he empe a u e analysis (FaIR + Silicone) is
a ailable om a public eposi o y (h ps://gi hub.com/Rlamboll/pos -Glasgow_clima e_ a ge s).
Me hods-only e e ences
54. Cal in, K. e al. GCAM 5.1: ep esen ing he linkages be ween ene gy, wa e , land, clima e,
and economic sys ems. Geosci. Model De . 12, (2019).
55. Loulou, R. & Lab ie , M. ETSAP-TIAM: he TIMES in eg a ed assessmen model Pa I: Model
s uc u e. Compu . Manag. Sci. 5, 7–40 (2008).
56. Be na d, A. & Vielle, M. GEMINI-E3, a gene al equilib ium model o in e na ional–na ional
in e ac ions be ween economy, ene gy and he en i onmen . Compu . Manag. Sci. 5, 173–
206 (2008).
57. Gia ola, S., Sachs, J., D’A ezac, M., Kell, A. & Hawkes, A. MUSE: An open-sou ce agen -based
in eg a ed assessmen modelling amewo k. Ene gy S a eg. Re . 44, 100964 (2022).
58. IMF. Wo ld Economic Ou look Da abase Ap il 2022. (2022).
59. OECD. Economic Ou look No 109 - Oc obe 2021. (2021).
60. Collodi, G. e al. Techno-economic e alua ion o SMR based s andalone (me chan ) hyd ogen
plan wi h CCS. (2017).
61. G an , N., Hawkes, A., Mi al, S. & Gambhi , A. The policy implica ions o an unce ain ca bon
dioxide emo al po en ial. Joule 5, 2593–2605 (2021).
62. Fuss, S. e al. Nega i e emissions—Pa 2: Cos s, po en ials and side e ec s. En i on. Res. Le .
13, 063002 (2018).
63. Gidden, M. J. e al. A me hodology and implemen a ion o au oma ed emissions
ha moniza ion o use in In eg a ed Assessmen Models. En i on. Model. So w. 105, 187–
200 (2018).
64. Lamboll, R. D., Nicholls, Z. R. J., Kiks a, J. S., Meinshausen, M. & Rogelj, J. Silicone 1.0.0: an
open-sou ce Py hon package o in e ing missing emissions da a o clima e change
esea ch. Geosci. Model De . 13, 5259–5275 (2020).
65. Nicholls, Z. e al. C oss-Chap e Box 7.1: Physical emula ion o Ea h Sys em Models o
scena io classi ica ion and knowledge in eg a ion in AR6. in Clima e Change 2021: The
Physical Science Basis. Con ibu ion o Wo king G oup I o he Six h Assessmen Repo o he
In e go e nmen al Panel on Clima e Change (Camb idge Uni e si y P ess, 2021).
66. Smi h, C. J. e al. FAIR 1.3: a simple emissions-based impulse esponse and ca bon cycle
model. Geosci. Model De . 11, 2273–2297 (2018).
67. S e n, P. C., Die z, T. & Vandenbe gh, M. P. The science o mi iga ion: Closing he gap
be ween po en ial and ac ual educ ion o en i onmen al h ea s. Ene gy Res. Soc. Sci. 91,
102735 (2022).
68. S e n, P. C., Die z, T., Nielsen, K. S., Peng, W. & Vandenbe gh, M. P. Feasible clima e
mi iga ion. Na . Clim. Chang. 13, 6–8 (2023).
Ex ended Da a Fig. 1 | 2050 CO2 by sec o . Remaining global CO2 emissions om di e en sec o s
o each model, in 2050.
Ex ended Da a Fig. 2 | Compa ison o p e- and pos -Glasgow emissions. Compa ison o p e- and
pos -Glasgow cu en policy and NDC 2030 global emissions.
P e-Glasgow emissions a e aken om Sognnaes e al (2021) e e ing o CP_In ensi y and
NDC_In ensi y scena ios.
Ex ended Da a Fig. 3 | Compa ison wi h AR6 scena ios emission anges. Boxplo s o cumula i e
emission anges ( om 2020 (included) o 2050 and 2100) o all
h ee scena ios in his s udy, compa ed wi h cumula i e emissions in c1, c2, c3 and c4 mi iga ion
scena ios om IPCC AR6 da abase.
