The potential land requirements and related land use change emissions of solar energy

1
Vol.:(0123456789)
Scien i ic Repo s | (2021) 11:2907 | h ps://doi.o g/10.1038/s41598-021-82042-5
www.na u e.com/scien i ic epo s
The po en ial land equi emen s
and ela ed land use change
emissions o sola ene gy
Di k‑Jan an de Ven1*, Iñigo Capellan‑Pe éz2, Iñaki A o1, Ignacio Cazca o1,3,
Ca los de Cas o2, P ali Pa el4 & Mikel Gonzalez‑Eguino1,5
Al hough he ansi ion o enewable ene gies will in ensi y he global compe i ion o land, he
po en ial impac s d i en by sola ene gy emain unexplo ed. In his wo k, he po en ial sola land
equi emen s and ela ed land use change emissions a e compu ed o he EU, India, Japan and
Sou h Ko ea. A no el me hod is de eloped wi hin an in eg a ed assessmen model which links
socioeconomic, ene gy, land and clima e sys ems. A 25–80% pene a ion in he elec ici y mix o
hose egions by 2050, we ind ha sola ene gy may occupy 0.5–5% o o al land. The esul ing land
co e changes, including indi ec e ec s, will likely cause a ne elease o ca bon anging om 0 o
50 gCO2/kWh, depending on he egion, scale o expansion, sola echnology e iciency and land
managemen p ac ices in sola pa ks. Hence, a coo dina ed planning and egula ion o new sola
ene gy in as uc u es should be en o ced o a oid a signi ican inc ease in hei li e cycle emissions
h ough e es ial ca bon losses.
The echnologies ha nessing enewable ene gy sou ces a e cha ac e ized by a powe densi y se e al o de s o
magni ude lowe han ossil uels1. As a consequence, he ansi ion o hese sou ces o ene gy is expec ed o
in ensi y he global compe i ion o land2–4. Fo example, he sp awl o bioene gy has been al eady iden i ied as
he majo d i e o ecen land use change (LUC) in de eloped egions5,6. Inc easing land compe i ion can cause
a ious en i onmen al impac s in ensi ying biodi e si y loss, wa e use o indi ec land use change (iLUC) emis-
sions. The la e e e s o emissions p oduced by using c opland o ene gy pu poses and, he e o e, indi ec ly
inc easing land compe i ion elsewhe e in he wo ld o mee global ood demand, po en ially eplacing land
wi h high ca bon s ocks, such as na u al o es s7–10. Fo example, he li e a u e es ima es ha he indi ec land
compe i ion induced by liquid bio uels in de eloped egions leads o global land clea ing and associa ed iLUC
emissions highe han he emission sa ings achie ed by eplacing gasoline by hese bio uels du ing 30 yea s11–13.
Fo sou ces o enewable ene gy o he han bioene gy, land equi emen s and he associa ed en i onmen al
impac s emain unde s udied in he li e a u e om a quan i a i e poin o iew1,10. In he case o sola ene gy, he
land compe i ion elemen is usually expec ed o be negligible due o i s highe ela i e ene gy densi y compa ed o
bioene gy and he possibili y o in eg a e i in u ban a eas o non-p oduc i e land7,14–16, and as such is cu en ly
excluded om o icial s a is ical epo ing and in eg a ed assessmen models (IAMs). Howe e , ecen s udies
based on sa elli e iews o u ili y-scale sola ene gy (USSE) unde ope a ion, ei he in he o m o pho o ol aics
(PV) o concen a ed sola powe (CSP), show ha hei land use e iciency (LUE) is up o six imes lowe han
ini ial es ima es17–19. Applying such obse ed LUEs acco dingly educes he po en ial con ibu ion o sola on
oo op space1,20,21.
The ins alla ion o USSE on land is subjec o a di e si y o cons ain s: sola esou ce cons ain s, which a e
ela ed o he sola i adiance in a ce ain a ea; geog aphical cons ain s such as he slope and he exis ing use o
he land; and egula o y cons ain s, e.g. he p o ec ed s a us o he land, o en ela ed o ecosys em and wildli e
p ese a ion21–27. The e o e, whe e a ailable, dese s and d y sc ubland wi h high sola i adiance and which a e
gene ally no sui able o human ac i i ies, a e used o planned o be used o sola ene gy26–28. Howe e , beyond
ha d es ic ions, o he ea u es such as he lack o oad, elec ici y and wa e in as uc u es, and he dis ance
OPEN
1Basque Cen e o Clima e Change (BC3), Edi icio Sede 1-1, Pa que Cien í ico de UPV/EHU, Ba io Sa iena
S/N, 48940 Leioa, Spain. 2Resea ch G oup on Ene gy, Economy and Sys em Dynamics, Escuela de A qui ec u a,
Uni e si y o Valladolid, A Salamanca, 18, 47014 Valladolid, Spain. 3Depa men o Economic Analysis,
ARAID-A agonese Agency o Resea ch and De elopmen , Ag i ood Ins i u e o A agon (IA2), Uni e si y o
Za agoza, Za agoza, Spain. 4Join Global Change Resea ch Ins i u e, Paci ic No hwes Na ional Labo a o y,
5825 Uni e si y Resea ch Cou , Sui e 3500, College Pa k, MD 20740, USA. 5Uni e si y o he Basque Coun y
(UPV/EHU), Ba io Sa iena s/n, 48940 Leioa, Spain. *email: dj.[email p o ec ed]
2
Vol:.(1234567890)
Scien i ic Repo s | (2021) 11:2907 | h ps://doi.o g/10.1038/s41598-021-82042-5
www.na u e.com/scien i ic epo s/
om human se lemen s complica e he la ge scale cons uc ion, ope a ion and main enance o sola powe in
hese a eas22. On op o ha , spa ial ic ions migh occu i land which is made a ailable o sola ene gy by
na ional o local go e nmen s is in eali y a biodi e si y ho spo 29,30 o he home o human communi ies31,32.
Recen de elopmen s show ha USSE in densely popula ed coun ies is o en ins alled on a able land ha is
used o po en ially sui able o o he p oduc i e uses such as ag icul u e o o es y17,26,33,34, in ensi ying land
compe i ion o he same easons as he sp awl o bioene gy does. Fu he mo e, clea ing cu en ly ege a ed
land o USSE also has local impac s on biodi e si y, ca bon cycling and aes e ics25,30,35.
The sha e o sola ene gy in global elec ici y scena ios ha a e la gely o ully deca bonized by 2050 usually
a y om abou 20% o 60%36,37. Fo speci ic egions, hese pene a ion le els can e en ange up o 90%37. Due
o he po en ial ele ance and ela i ely low powe densi y o sola ene gy in a deca bonized u u e, and gi en
ha PV in u ban a eas will only be able o co e a sha e o he o al demand1,21, his pape aims o quan i y he
po en ial land occupa ion o sola ene gy ins alled up o 2050, and he ela ed di ec and indi ec impac s on
ca bon cycles, wi hin a con ex o global clima e ac ion as p oposed in he Pa is Ag eemen . We concen a e on
h ee egions wi h he e ogeneous ea u es whe e u u es wi h a high sola ene gy pene a ion ha e been iden i-
ied in he li e a u e as likely o induce land compe i ion: he Eu opean Union (EU), India and join ly Japan
and Sou h-Ko ea. Unce ain ies in e ms o u u e sola module e iciency imp o emen s up o 2050 (20, 24,
28%)a e aken in o accoun , as well as sola land managemen op ions and hei di e en associa ed impac s
on local ca bon cycles: depending on how he land below and a ound sola ene gy ins alla ions is managed,
and on he land use p io o he con e sion o sola land, land ans o ma ion o hos ing USSE can cause a ne
elease o ca bon ha was s o ed in soil and ege a ion, o can lead o ne ca bon up ake38. See Sec ion2 o he
Supplemen a y Ma e ial (SM) o an o e iew o he scena ios designed o his s udy.
Resul s
A no el me hod has been speci ically designed in his wo k which allows dynamically accoun ing o he land
occupa ion o sola ene gy, depending on he geog aphical loca ion and yea o ins alla ion and based on eal-
wo ld LUEobse a ions1,17, wi hin a s a e-o - he-a In eg a ed Assessmen Model (IAM) ha links ene gy, land,
socioeconomic and clima e sys ems (see “Me hods” sec ion) and ha has also been applied in o he s udies o
measu e he e es ial ca bon leakage induced by bioene gy in a clima e change mi iga ion con ex 9,39,40. Th ough
his model, a ange o elec ici y mix pene a ion scena ios a e simula ed o sola ene gy echnologies (and bio-
ene gy o compa ison). Based on he spa ially de ined LUE o sola ene gy, as well as he iden i ied po en ial o
sola ene gy in u ban a eas, dese s and d y sc ublands, land use o sola ene gy compe es wi h o he land uses
h ough he inhe en ela i e p o i abili y o each land use. The induced global land co e changes and ela ed
LUC emissions a e hen compa ed wi h scena ios whe e he same emission educ ion a ge s in he elec ici y
sec o a e achie ed wi hou sola and bioene gy, o isola e he addi ional land equi emen s, land co e impac s
and ela ed LUC emissions p o oked by sola and bioene gy.
Sola land occupa ion. Table1 shows he ob ained esul s o absolu e and ela i e land equi emen s
o sola ene gy, based on land ha is (po en ially) sui able o comme cial p oduc ion (i.e. c ops, animal hus-
band y, and o es y, so excluding he use o oo ops dese s and d y sc ublands), o he simula ed scena ios
a pene a ion a es anging om 26 o 79% o he elec ici y mix, and o he ange o u u e sola PV module
e iciencies. Due o he lowe i adiance and highe la i ude o Eu ope, absolu e land use o pe uni o sola ou -
pu is almos wice as high as in Japan and Sou h-Ko ea and h ee imes highe as in India (see Fig S6 in he SM).
This a io inc eases wi h highe pene a ion a es, due o he sa ia ion o he po en ial o gene a e sola ene gy
on oo ops (see also Figu eS12 in he SM) in combina ion wi h he dec easing ma ginal e u ns o land-based
sola ene gy. Wi h sola ene gy accoun ing o 25 o 80% o he elec ici y mix, land occupa ion by USSE is
p ojec ed o be signi ican , anging om 0.5 o 2.8% o o al e i o y in he EU, 0.3 o 1.4% in India, and 1.2 o
5.2% in Japan and Sou h-Ko ea. This occupa ion is unequally sp ead wi hin each o he egions, as a eas ha a e
ela i ely a ac i e o sola ene gy a e p io i ized in each egion, such as sou he n Eu ope, no h wes e n India,
and sou he n Japan and Sou h-Ko ea (see Fig.1).
The u u e land equi emen s o sola ene gy ob ained o each scena io and egion can be pu in pe spec i e
compa ed, o example, o he cu en le el o buil -up a ea and ag icul u al c opland. In he h ee egions, a la ge
pa o he o al buil -up a ea (u ban and sola land) will consis o sola PV panels o CSP helios a s by 2050 i
a leas hal o he p oduced elec ici y comes om sola powe . Land o sola would amoun o o e 50% o he
cu en EU u ban land, o e 85% o India, and o e 75% in Japan and Sou h-Ko ea. F om a di e en pe spec i e,
a signi ican pa o he sunligh cap u ed o comme cial use would be used o elec ici y gene a ion ins ead o
g owing c ops, especially in Japan and Sou h-Ko ea (29–39%) and he EU (8–10%). The ela i e p ojec ed land
a ea dedica ed o ei he c ops o sola ene gy s ongly di e s wi hin each egion, wi h po en ial localecosys em
and landscape implica ions (see Figu eS16 in he SM).
Land co e changes. Sola ene gy in as uc u e cu en ly occupies a negligible amoun o land globally.
Ou esul s show ha his changes in scena ios wi h a high sha e o sola ene gy in he u u e elec ici y mix.
Figu e2 shows he ob ained land co e changes ela ed o inc easing sola ene gy (see Table1), wi hin each o
he h ee egions (uppe pa o he igu e), and indi ec ly d i en land co e changes ou side o hese egions in
he es o he wo ld (lowe pa ). Based on assump ions on economic and sui abili y cons ain s (see Sec ion1c
in SM), sola ene gy expansion in he h ee egions is ound o p edominan ly eplace (o a oid u u e land con-
e sion o) land used o comme cial pu poses, such as c opland o comme cial o es (e.g. o imbe p oduc s
o biomass). Ins ead, sola ene gy pene a ion is no ound o signi ican ly a ec he co e o unmanaged land in
each o he h ee egions. Howe e , he displacemen o comme cial land wi hin each o he h ee ocus egions
3
Vol.:(0123456789)
Scien i ic Repo s | (2021) 11:2907 | h ps://doi.o g/10.1038/s41598-021-82042-5
www.na u e.com/scien i ic epo s/
would incen i ise he use o cu en ly unused a able land in o he egions, while also boos ing he comme ciali-
sa ion o unmanaged land, indi ec ly leading o he loss o na u al land co e . The magni ude o his indi ec
land co e impac depends on he c op and o es y p oduc i i y in egions whe e sola ene gy pene a ion
akes place: ela i ely high c op p oduc i i ies in he EU, Japan and Sou h-Ko ea mean ha he displacemen o
c opland om hese egions o egions wi h lowe c op p oduc i i ies would indi ec ly inc ease global c opland
co e , ampli ying he impac o sola ene gy expansion in hese egions on global land compe i ion by up o 22%.
This e ec is lowe a lowe sola ene gy pene a ion le els (e en nega i e in he EU), as sola ene gy is p ojec ed
o displace he mos ma ginal c opland i s . In India, whe e cu en and p ojec ed c op p oduc i i ies a e below
he global a e age, he impac o sola expansion on global land compe i ion is less signi ican .
Figu e2 shows ha , ei he di ec ly o indi ec ly, expansion in sola ene gy p edominan ly educes non-
comme cial land co e on a global scale: o e e y 100 hec a es o sola land in he EU, we ind ha , depending
on he sola pene a ion le el, 31 o 43 hec a es o unmanaged o es may be clea ed h oughou all he wo ld.
The same amoun o sola land in India would clea 27 o 30 hec a es o unmanaged o es , and o Japan and
Sou h-Ko ea, he a io is 49 o 54 hec a es.
Table 1. Land occupa ion cha ac e is ics a di e en sola pene a ion le els by 2050. Ranges show esul s o
wi h di e en u u e sola module e iciencies wi h le alues wi hin each sola - ela ed column ep esen ing
28% e iciency and igh alues 20%. Resul s o bioene gy scena ios included o compa ison. a These a e
ealized pene a ion le els. See Sec ion1b in he SM o mo e in o ma ion. b Land sui able o comme cial
pu poses does no include he use o oo op space, dese s o d y sc ublands ha a e no sui able o c op,
pas u e o o es cul i a ion. Dese s and d y sc ublands in India hos abou 11.5–12% o sola ene gy
h oughou all pene a ion scena ios o sola ene gy in India (see Figu eS8 in he SM). See he “Me hods”
sec ion and Sec ion2b in he SM o modelling de ails wi h espec o loca ion choices. c These columns
compa e he “Land sui able o comme cial pu poses by 2050” o o al land, u ban land (in 2010; u u e u ban
expansion is no modelled) and o al c op a ea (abs ac ed om he same modelled scena ios).
Focus egion
Sola pene a ion
by 2050a
Roo op PV
gene a ion sha e
o sola (2050) Occupa ion o land sui able o
comme cial pu poses (b) by 2050 Rela i e sola land occupa ion by 2050c
Occupa ion o
land sui able
o comme cial
pu poses (b)
% o o al
elec ici y (PWh
in 2050)
% o sola
pene a ion by
2050
Sola ene gy Bioene gy (%
wi hin egion) % o o al land
a ea
Compa ed o
u ban a ea in
2010 (%)
Compa ed o
c op a ea in 2050
(%)
km2 pe TWh
sola (a e age
2020–2050)1000 km2
Eu opean Union
26 (1.19) 24.3–23.0 21–28 366 (45) 0.5–0.7 20–27 1.9–2.5 19.4–24.2
53 (2.54) 12.3–11.6 53–69 614 (38) 1.3–1.7 50–66 4.8–6.3 22.1–28.0
79 (3.87) 8.1–7.6 85–111 969 (32) 2.1–2.8 81–106 7.7–10 23.5–29.7
India
30 (1.8) 10.6–9.9 10–14 596 (16) 0.3–0.5 46–62 0.6–0.9 6.4–8.2
54 (3.29) 5.9–5.5 20–26 1051 (12) 0.7–0.9 88–118 1.2–1.6 6.5–8.5
78 (4.88) 3.6–3.3 30–41 1516 (10) 1.0–1.4 137–182 1.9–2.5 6.9–8.8
Japan and Sou h-
Ko ea
28 (0.5) 25.0–22.1 5–6 185 (17) 1.2–1.6 36–48 8.3–11 12.9–15.6
46 (0.8 PWh) 15.6–13.8 9–12 279 (13) 2.3–3 68–89 16–21 13.3–16.4
74 (1.3 PWh) 9.0–8.1 16–21 429 (10) 4–5.2 120–157 29–39 13.9–17.1
Figu e1. Geog aphical dis ibu ion o he sha e o o al land occupied by sola ene gy wi hin each egion,
by ag o-ecological zone. See “Me hods” sec ion and Figu eS1 o he SM o mo e in o ma ion on he spa ial
esolu ion used in his s udy. Sou ce: Au ho s´ own elabo a ion wi h he A c GIS 10.5.1 Desk op (Es i) so wa e.
4
Vol:.(1234567890)
Scien i ic Repo s | (2021) 11:2907 | h ps://doi.o g/10.1038/s41598-021-82042-5
www.na u e.com/scien i ic epo s/
Impac on e es ial ca bon s ocks. The land co e changes inFig.2 imply ha sola expansion leads
oLUC emissions, such as iLUC emissions ela ed o inc easing global land compe i ion, emissions ela ed o
ege a ion loss i o es and sc ubland makes place o sola land (ei he di ec ly h ough de o es a ion o indi-
ec ly by a oiding u u e a o es a ion), and ca bon elease om soil and ege a ion di ec ly below he ins alled
panels, whe e sunligh is much educed35. Howe e , an impo an pa o he emission balance is ela ed o he
land managemen egime applied in sola land. I all ege a ion is clea ed and a oided o eg ow h ough he
applica ion o he bicides, which is a common p ac ice in a ious coun ies41, LUC emissions om sola expan-
sion a e u he ampli ied. In con as , i a able land plo s a e con e ed o sola pa ks whose su ace is managed
as pas u es, he e will be a ne ca bon seques a ion in ege a ion and soil in he decades ollowing upon he
con e sion (apa om he land di ec ly below he panels, whe e pho osyn hesis is la gely blocked)35, o se ing
some o all o he ine i able LUC emissions caused by land compe i ion. In eali y, he applica ion o a pa icula
land managemen p ac ice depends on a di e si y o local ac o s (policies, clima e, e c.). See “Me hods” sec ion
o a de ailed explana ion o each land managemen egime.
Figu e3 and Table2 show he ob ained LUC emissions pe uni o sola ene gy ins alled om 2020 o 2050
associa ed o he di e en simula ed sola pene a ion and module e iciency scena ios, and o di e en man-
agemen egimes o he land in sola pa ks. They show ha sola expansion scena ios un il 2050 will mos likely
lead o ne LUC emissions, al hough he e can be a ne ca bon seques a ion in India when managing he land
in sola pa ks as pas u es. The seques a ion e ec is ampli ied i delayed pos -2050 impac s on local ca bon
cycles a e aken in o accoun (see Table2). In he absence o land managemen p ac ices speci ically aiming a
ca bon seques a ion, land co e change due o he expansion o sola ene gy in he EU would cause 13 o 53g
o CO2 pe p oduced kilowa -hou (kWh) o elec ici y, abou 4 o 16% o he CO2 emissions om na u al gas
i ed elec ici y. Sola ene gy in India in ol es signi ican ly less land co e change pe uni o ou pu (see Fig.2),
and es ima ed LUC emissions pe kWh a e below 12g o CO2 o all scena ios. In Japan and Sou h-Ko ea, LUC
emissions ela ed o he expansion o sola ene gy a e 11 o 35g o CO2 pe kWh. When using ela i ely e icien
PV echnologies such as monoc ys alline and mul ic ys alline silicon (made om a single c ys al o silicon and
om many silicon agmen s mel ed oge he , espec i ely) (lowe ange o es ima ed LUC emissions, highe
ange o non-land li e cycle emissions), ou esul s show ha LUC emissions a e compa able o abou 10 o 50%
he cu en non-land li e cycle emissions o such echnologies. Ins ead, when using less space-e icien bu mo e
esou ce-e icien PV echnologies such as hin- ilm Cadmium ellu ide (CdTe) made by deposi ing one o mo e
hin laye s o pho o ol aic ma e ial on a glass, plas ic o me al subs a e (highe ange o LUC emissions, lowe
ange o non-land li e cycle emissions), we es ima e LUC emissions in he ange o 50 o 150% o he non-land
li e cycle emissions. I sola land is seeded wi h he bs and managed as pas u e, ne LUC emissions d op by mo e
han 50% in mos cases.
Figu e2. Global land-co e changes by 2050 due o sola expansion, o a ange o sola ene gy pene a ion
le els and o an a e age e iciency o ins alled sola modules o 24% by 2050. The uppe g aphs shows o al land
co e changes by 2050 ela i e o 2015 wi hin each egion and he lowe side shows he land co e changes in
he es o he wo ld (leaking), indi ec ly d i en by he pene a ion o sola land wi hin he egion. Posi i e land
co e changes e e o inc eases and nega i e o land co e loss. See Sec ion3b in he SM o agg ega ed global
land co e changes. No e ha land co e changes do no co espond wi h land use changes: his igu e compa es
o al land co e in di e en scena ios o land-based sola ene gy pene a ion, bu does no show which speci ic
ypes o land con e o sola land (o any o he ype o land). No e ha hese land co e changes a e based on
simula ed land use decisions d i en by economic op imisa ion. See “Me hods” sec ion o mo e de ails.
5
Vol.:(0123456789)
Scien i ic Repo s | (2021) 11:2907 | h ps://doi.o g/10.1038/s41598-021-82042-5
www.na u e.com/scien i ic epo s/
Table2 also shows he ob ained emissions pe m2 o land occupa ion by sola ene gy, which e lec he alue
o he used land in e ms o i s po en ial o seques e ca bon: ei he di ec ly by i s capaci y o seques e ca bon in
soil and ege a ion, o indi ec ly by i s ag icul u al p oduc i i y which, i being displaced by sola land, will lead
o con e sion o non-comme cial land o ag icul u al land elsewhe e. Since in ou simula ions land o USSE
p edominan ly eplaces comme cial land g owing c ops o imbe p oduc s wi hin each egion(see Fig.2), sola
ene gy expansion displaces comme cial imbe p oduc ion o o he egions, indi ec ly inc easing ca bon seques-
a ion ou side he egion by incen i ising cu en ly deg aded o es o o he a able land o be comme cialised
o imbe p oduc ion. A highe sola pene a ion a es howe e , inc easing land p essu e causes mo e na u al
o es s o be used o imbe o c op p oduc ion, leading o highe land use change emissions ou side he egion.
This e ec is bes isible o sola pene a ion scena ios in he EU, due o he high absolu e amoun o land use.
Sola ene gy e sus bioene gy. IAMs which link ene gy, economy, land and clima e modules end o
ely s ongly on he cul i a ion o dedica ed bioene gy c ops (such as swi chg ass and miscan hus) in global
clima e change mi iga ion scena ios43. As he land use impac s o bioene gy ha e been ex ensi ely analysed in
o he s udies, using he same model9,39, we p oceed o compa e he land occupa ion and ela ed LUC emissions
o elec ici y p oduc ion om sola ene gy and bioene gy, wi h he pu pose o imp o ing he compa abili y o
he ob ained esul s.
Table1 shows ha land equi emen s o eaching ce ain le els o elec ici y pene a ion wi h sola ene gy
a e abou a magni ude lowe han land equi emen s o mee hose same le els wi h bioene gy. Compa ing he
addi ional global LUC emissions un il 2100 as a esul o eaching ce ain sha es o bioene gy in he elec ici y
mix o 2050 in he egions in his s udy, we obse e om Table2 ha emissions pe dedica ed m2 a e in many
cases lowe han o sola ene gy a he same pene a ion le el in he elec ici y mix. Howe e , he ene gy densi y
o sola ene gy is a magni ude highe han ha o bioene gy. By compa ing he o alLUC emissions om one
uni o sola and bioene gy o he “a oided” pe iodicalcombus ion emissions om na u al gas i ed elec ici y, we
calcula e he “CO2 payback pe iod” o hese enewable al e na i es o elec ici y p oduc ion, which is a common
me hod o compa e LUC emission impac s o di e en ypes o bioene gy13,44. Table2 shows ha he payback
pe iod o bioene gy is signi ican ly highe (~ 4yea s) han ha o sola ene gy (< 8mon hs), as he highe land
equi emen s o bioene gy mo e han o se he lowe emissions pe m2 ound in mos cases. Howe e , since he
physical cha ac e is ics o bioene gy allow o ade o e la ge dis ances, compa able o ossil uels and in con as
o elec ici y om sola ene gy, only a limi ed pa o he land equi emen s and ela ed LUC emissions d i en
by bioene gy expansion is p ojec ed o be wi hin he EU, India, Japan and Sou h-Ko ea. No e ha hese esul s
do only ocus a sola and bioene gy based in land wi h po en ial comme cial use. Sola ene gy in u ban a eas,
Figu e3. Land use change emissions ela ed o land occupa ion pe kWh o sola ene gy om 2020 o 2050, o
he h ee sola land managemen egimes applied (see “Me hods” sec ion o mo e de ails), and ela i e o o he
li e cycle emissions o PV sys ems (depend on loca ion o ins alla ion) and emissions om na u al gas i ed
elec ici y (independen o loca ion). Unce ain y bounds e lec sola module e iciency scena ios ( eaching
a e age e iciencies o 20, 24 and 28% o modules ins alled in 2050; see Sec ion2c in SM). 1 Non-land li e cycle
emissions o PV a e based on a ange o PV echnologies, including mono and mul ic ys alline silicon (highe
ange), hin- ilm CdTe (lowe ange), CIS and a-Si sys ems as calcula ed in Liu & an den Be gh (2020)42, and
based on an a e age global ca bon in ensi y o elec ici y (0.48kg CO2/kWh). The ange is calcula ed by di iding
he egionally weigh ed sola elec ici y ou pu pe m2 as used in his s udy, by CO2 emissions pe m2 panel
su ace.

6
Vol:.(1234567890)
Scien i ic Repo s | (2021) 11:2907 | h ps://doi.o g/10.1038/s41598-021-82042-5
www.na u e.com/scien i ic epo s/
dese s and d y sc ublands, as well as bioene gy om was e o ag icul u al and o es y esidue, a e assumed
no o con ibu e o LUC emissions no ca bon seques a ion.
Discussion
By ep esen ing he land equi emen s o sola ene gy wi hin an IAM ha in eg a es ene gy, land, socioeconomic
and clima e sys ems, we we e able o, o he i s ime in he li e a u e o ou knowledge, es ima e he land co e
impac s and ela ed LUC emissions o sola ene gy wi hin clima e change mi iga ion scena ios up o 2050. The
ob ained esul s ep esen a con ibu ion o he no el ield o esea ch which analyses he en i onmen al impac s
o signi ican ly up scaling enewables o he han biomass45,46.
A combina ion o echnical and geopoli ical easons complica es he ins alla ion o sola ene gy a om
consump ion poin s. The e o e, a high sha e o sola gene a ion in he ene gy mix in ela i ely densely popula ed
egions wi h high pe capi a ene gy demands can equi e a signi ican sha e o domes ic land, compa able o he
cu en buil -up a ea in hese egions. The mos ele an ac o s in luencing he land use pe uni o sola ene gy
a e sola i adia ion, la i ude, and u u e sola module e iciencies. A he domes ic le el, sola ene gy is ound
o p edominan ly compe e o land wi h c opland and managed o es s, while on a global scale, 27 o 54% o
he land equi ed o sola ene gy is ound o indi ec ly displace unmanaged o es s, p edominan ly ou side he
Table 2. Land use change emissions and payback pe iods o sola pene a ion and sola land managemen
scena ios. Ranges show esul s o wi h di e en u u e sola module e iciencieswi hle alues wi hin
each sola - ela ed column ep esen ing 28% e iciency and igh alues 20%. Resul s o bioene gy scena ios
included o compa ison. a CLEAR: pe manen ly clea ing land ege a ion; MAINT: Main ain/ es o e p e ious
ege a ion (up o 30cm); SEED: Seeding and managemen as pas u es. See “Me hods” sec ion o a de ailed
desc ip ion o hese land managemen scena ios. b Di iding all LUC emissions om 2020 o 2050 o he o al
amoun o gene a ed elec ici y (including non-land-based sou ces, such as sola oo ops, unp oduc i e land
o was e- o-ene gy plan s o bioene gy). c Di iding all LUC emissions om 2020, including delayed ca bon
elease o seques a ion un il 2100, by he o al land a ea dedica ed o sola and bioene gy by 2050 (maximum
poin ). Nega i e alues indica e ne ca bon seques a ion o e e y m2 o sola land. d Calcula ed assuming a
he mal e iciency o 50% o na u al gas powe plan s. Only aking accoun di ec combus ion emissions.
Sola pene a ion
le el
Land managemen
scena io in
sola landa
Di ec and indi ec land use change (LUC)
emissions due o sola ene gybLUC emissions pe
occupied m2c
LUC CO2 payback pe iod
when eplacing gas- i ed
elec ici y d
% and PWh in 2050
elec . mix
Wi hin egion Ou side egion To al Sola ene gy Bio-ene gy Sola ene gy Bio-ene gy
G ams o CO2 pe kWh o sola elec ici y ou pu
be ween 2020 and 2050 (a e age) kg CO2 (2020–2100) pe m2
o dedica ed land in 2050 Mon hs
Eu opean Union
26% (1.19 PWh)
CLEAR 28.1 o 38.6
− 6.4 o − 4.6
21.8 o 34.0 4.4 o 5.2
3
4.7 o 7.1
46.9MAINT 19.7 o 27.1 13.3 o 22.5 3.1 o 3.7 3.2 o 5.2
SEED 9.0 o 12.7 5.0 o 8.1 1.1 o 1.7 1.2 o 2.4
53% (2.54 PWh)
CLEAR 33.0 o 43.5
3.3 o 4.7
36.3 o 48.3 6.7 o 6.8
3.1
6.2 o 8.2
49.2MAINT 24.0 o 30.9 27.3 o 36.6 5.3 o 5.4 4.8 o 6.5
SEED 11.7 o 15.4 15.0 o 20.1 ~ 3.1 2.9 o 3.8
79% (3.87 PWh)
CLEAR 34.6 o 46.7
5.6 o 6.9
40.2 o 53.6 7.2 o 7.3
3
6.4 o 8.3
49.3MAINT 25.8 o 35.5 31.4 o 42.4 ~ 5.9 5.2 o 6.7
SEED 12.7 o 18.1 18.3 o 24.9 3.5 o 3.6 3.2 o 4.0
India
30% (1.8 PWh)
CLEAR 10.0 o 12.5
− 4.0 o − 1.6
6.0 o 10.8 1.3 o 2.4
2.3
0.3 o 0.8
41.7MAINT 8.1 o 10.0 4.1 o 8.4 0.6 o 1.7 0.2 o 0.6
SEED 1.1 o 1.3 − 2.7 o − 0.6 − 5.2 o − 6.2 − 1.5 o − 1.7
54% (3.29 PWh)
CLEAR 10.8 o 13.0
− 2.8 o − 1.3
8.0 o 11.7 2.5 o 3.1
2.3
0.6 o 1.0
43.2MAINT 9.1 o 10.8 6.3 o 9.5 1.9 o 2.4 0.5 o 0.8
SEED 3.0 o 3.2 0.4 o 1.7 − 4.5 o − 5.0 − 1.2 o − 1.5
78% (4.88 PWh)
CLEAR 9.7 o 11.7
− 2.1 o − 0.9
7.6 o 10.8 2.9 o 3.2
2.4
0.7 o 1.1
43.9MAINT 8.0 o 9.5 5.9 o 8.6 2.1 o 2.5 0.5 o 0.8
SEED 2.3 o 2.5 0.4 o 1.4 − 4.4 o − 4.8 − 1.2 o − 1.5
Japan and Sou h-
Ko ea
28% (0.5 PWh)
CLEAR 18.9 o 25.8
− 3.2 o 0.8
15.7 o 26.6 4.8 o 6.8
2.7
2.9 o 5.2
47.7MAINT 12.8 o 17.5 9.6 o 18.2 2.2 o 4.1 1.4 o 3.1
SEED 6.8 o 10.0 3.6 o 10.7 − 0.3 o − 2.2 − 0.2 o − 1.3
46% (0.8 PWh)
CLEAR 23.0 o 30.6
− 2.1 o 0.6
20.9 o 31.1 6.1 o 7.2
2.6
3.3 o 5.0
47.3MAINT 16.9 o 20.5 14.8 o 21.1 3.9 o 4.3 2.1 o 2.9
SEED 9.9 o 12.5 7.7 o 13.1 − 0.2 o − 1.2 − 0.2 o − 0.7
74% (1.3 PWh)
CLEAR 25.6 o 33.3
0.0 o 2.3
25.6 o 35.6 7.1 o 7.7
2.7
3.6 o 5.0
48.9MAINT 15.6 o 20.0 15.6 o 22.2 3.5 o 3.9 1.8 o 2.6
SEED 10.0 o 12.9 10.0 o 15.2 − 0.5 o 0.0 − 0.3 o 0.0
7
Vol.:(0123456789)
Scien i ic Repo s | (2021) 11:2907 | h ps://doi.o g/10.1038/s41598-021-82042-5
www.na u e.com/scien i ic epo s/
egion whe e he sola ene gy is consumed. No e ha his iLUC has been documen ed o happen o bio uels11–13,
al hough he s eng h o his e ec is no compa able o sola ene gy gi en ha he powe densi y o sola is
much highe han ha o bio uels. S ill, we do ind a non-negligible e ec in his s udy. The ob ained land co e
change imply en i onmen al consequences such as g eenhouse gas emissions and biodi e si y loss47. Howe e ,
he impac o USSE on local en i onmen al impac s depends s ongly on how his new sola land will be man-
aged. I all p e ious ege a ion is pe manen ly clea ed, he o al (di ec and indi ec ) LUC emissions ela ed o
he expansion o sola ene gy om 2020 o 2050 co espond o 5 o 16% o emissions om na u al gas combus-
ion o powe gene a ion in de eloped egions such as he EU, Japan and Sou h-Ko ea, and abou 2.5–3.5% in
India, whe e condi ions o sola ene gy a e mo e a ou able and c op yields a e lowe , implying less indi ec
emissions. Howe e , i sola land is seeded wi h he bs and managed as pas u es, o al LUC emissions pe kWh o
elec ici y in he s udied pe iod a e 3 o 5 imes lowe , and could e en be nega i e (i.e., becoming ne sou ces o
ca bon seques a ion) in India, Japan and Sou h-Ko ea, i long- e m e ec s (pos 2050) a e aken in o accoun .
Nume ous Li e Cycle Assessmen s (LCA) ha e been pe o med o sola ene gy, es ima ing he li e cycle
emissions o sola ene gy sys ems depending on many ac o s, such as he yea and loca ion o cons uc ion, sola
module e iciency, moun ing sys em, loca ion o inpu p oduc ion, among o he s42,48. Compa ing he non-land
li e cycle emissions om LCAs o he LUC emissions es ima ed in his s udy, we can conclude ha LUC emis-
sions (which a e no mally no included in LCAs) inc ease o al li e cycle emissions o new USSE p ojec s by 10
o 150% in he absence o land managemen p ac ices ocused on seques e ing ca bon in sola land, depending
mainly on he egion whe e he in as uc u e is ins alled and he ype o echnology used. While his is a no a-
ble inc ease in li e cycle emissions, i is also impo an o conside ha LUC emissions will no epea i a sola
plan is enewed o upg aded a e he ini ial cons uc ion phase, and he e o e a e age LUC emissions o sola
ene gy will be lowe in he u u e. Also, his e es ialpa o sola ene gyli e cycle emissions could be a oided
by applying land managemen p ac ices ocused on ca bon seques a ion in sola land.
Using an exis ing IAM o s udy he po en ial land impac s o sola ene gy expansion, we we e bound o he
limi a ions o his model. One o hese was he di ision o land zones in he model (co esponding o Ag o-
Ecological zones, see “Me hods” sec ion), which de e mine he bounda ies o he geog aphical compe i ion o
hos sola ene gy wi hin each egion. This p e-de ined dis ibu ion was o iginally designed o cap u e a ia ions
in c op yields, and is no ideal o de ining he geog aphical di e si y o sola ene gy “yields” wi hin a egion.
Al hough a gene al good co espondence is ound, he e a e also excep ions (see Figu eS6 in he SM). This
limi a ion could be dampened in u u e wo k by using/de eloping a land co e laye ha ma ches be e wi h
geog aphical di e ences in sola i adia ion and la i ude. We we e also no able o accoun o he sui abili y o
land o sola ene gy limi ed by he slope o he p o ec ion o he land24. The e o e, we implici ly assumed ha
hose hec a es ha a e con e ed o sola land in ou scena ios a e indeed sui able o hos ing sola ene gy. In
con as , some land is sui able o sola ene gy and no o comme cial c ops o o es s, such as d y sc ubland
and dese s, which a e by de aul excluded om land compe i ion in he applied model. The inclusion o a
sola po en ial on iden i ied “was elands” in India (see “Me hods” sec ion) should ha e la gely ci cum en ed
his inhe en limi a ion in he applied me hod. To ex end he analysis pe o med in his s udy o o he egions,
i is impo an o ha e a well-quan i ied po en ial o sola ene gy in a eas ha a e no sui able o hos o he
comme cial land uses such as ag icul u e and o es y. Finally, we ha e no aken in o accoun he po en ial o
in eg a e sola sys ems in ag icul u al land (ag i ol aic sys ems), a echnique ha is cu en ly in an ea ly s age
o esea ch and de elopmen and o which he la ge-scale pe o mance is s ill unce ain49.
To da e, land use o sola ene gy is negligible compa ed o o he human land uses. Howe e , he ob ained
esul s show ha in u u e scena ios, wi h a la gely deca bonized elec ici y sys em, high pene a ion a es o
sola ene gy will equi e signi ican amoun s o land o be occupied by sola powe plan s. Fu he wo k applying
ecological ools should be ocused owa ds in es iga ing he implica ions o hese addi ional land occupa ion
le els -including he addi ional ansmission powe lines- in e ms o habi a agmen a ion and ecosys em dis-
u bance. Si ing policies o USSE should a oid ad e se land impac s and limi land compe i ion, o example
by excluding high yield c opland as al eady pe o med in some coun ies50, maximising he use o u ban a eas
and deg aded a able land22, o by seeding sola land wi h he bs and managing hese lands as common pas u es
(e.g. by allowing ex ensi e animal g azing), con e ing sola land o a ne sou ce o ca bon seques a ion35. Such
echno-ecological syne gies a e c ucial o minimising he unin ended consequences o sola expansion38, such
as he po en ial impac s on land co e change and LUC emissions as measu ed in his s udy. The esul s in his
s udy also indica e ha minimum e iciency s anda ds o sola modules help o educe sola land equi emen s
and limi land compe i ion, al hough he e migh be a ade-o wi h non-land li e cycle impac s, which end
o be highe o high-e iciency sola modules. Finally, he inclusion o his new ype o land use in in eg a ed
ene gy-land-clima e models, as has been done in his pape , will be use ul o cap u e a la ge ange o implica-
ions o speci ic ene gy ansi ion scena ios.
Me hods
The Global Change Assessmen Model (GCAM), e sion 4.3, has been used as a base o his s udy51. GCAM is
a dynamic- ecu si e model wi h echnology- ich ep esen a ions o he economy, ene gy sec o and land use
linked o a clima e model ha can be used o explo e clima e change mi iga ion policies including ca bon axes,
ca bon ading, egula ions and accele a ed deploymen o ene gy echnologies. See Sec ion1a o he SM o
a wide desc ip ion o he model. The backg ound scena io o he model exe cises in his s udy consis o he
“Middle o he Road” Sha ed Socioeconomic Pa hway (SSP 2)52 wi h global CO2 educ ion a ge s as de ined by
he Na ionally De e mined Con ibu ions (NDCs) wi h inc eased ambi ions a e 203053.
In o de o iden i y he e ec s ha sola ene gy and bioene gy pa hways ha e on land use and land use change
emissions, h ee pa hways ha e been modelled achie ing a de ined pene a ion le el in he elec ici y mix om
8
Vol:.(1234567890)
Scien i ic Repo s | (2021) 11:2907 | h ps://doi.o g/10.1038/s41598-021-82042-5
www.na u e.com/scien i ic epo s/
2020 o 2050, using di e en elec ici y gene a ion echnologies (see Sec ion2b in he SM on how he di e en
pene a ion le els ha e been modelled):
• Sola ene gy pa hway (S): land-based PV, oo op-based PV, CSP
• Bioene gy pa hway (B): Con en ionalbiomassand biomass gasi ica ion(wi h and wi hou Ca bon Cap u e
and S o age), Biomass-d i enCombined Hea and Powe .
• Non-land-occupying pa hway (NL): wind, geo he mal, oo op-based PV (and nuclea in scena ios whe e
pene a ion le el canno be eached wi h he i s 3 echnologies oge he )
The land occupa ion o sola and bioene gy (Figs.1 and 2, Table1) is iden i ied using Eq. (1), land use change
emissions pe uni o ou pu om 2020 o 2050 ( o Fig.3 and Table2) om 2020 o 2050 ha e been calcula ed
using Eq.(2), and he CO2 payback pe iod (Table2) has been calcula ed using Eq.(3). In hese equa ions, he
subsc ip de ines he egion, p he elec ici y pene a ion le el, i he echnologies included in ei he he sola -
o bioene gy pa hway, NL de ines non-land-occupying ene gy echnologies and i(l) ep esen s land-compe ing
sola - o bioene gy, so no aking in o accoun sola ene gy based on oo ops, dese s o d y sc ublands o
bioene gy om was e o ag icul u al esidues. The pa ame e a de ines he CO2 emission ac o pe uni o elec-
ici y ou pu o he al e na i e he mal gene a ion echnology (i.e. na u al gas). Scena ios a e un un il 2050,
bu delayed e ec s on ca bon elease o seques a ion in ege a ion and soils can be abs ac ed un il 2100. The
impac om land managemen egimes ha e been calcula ed h ough o -model calcula ions, as such egimes
a e assumed no o a ec he alloca ion p ocedu e o new sola ene gy. See Sec ion2d o he SM o mo e de ails.
Land compe i ion in GCAM. Land use and ag icul u al ou pu in GCAM e sion 4.3 a e calib a ed o
p e-de ined Ag o-Ecological Zones (AEZs), which sub-di ide geo-poli ical egions in 18 di e en ypes o land
egions, based on di e ences in clima e zones ( opical, empe a e, bo eal) and he leng h o g owing pe iods o
c ops54. See Figu eS1 in he SM o an o e iew o he AEZs wi hin he h ee ocus egions o his s udy.
Land use in GCAM has been di ided in di e en nodes ha a ec he le el o compe i ion be ween di e en
uses (see Figu eS3 in SM). Those land-use ca ego ies (e.g. co n, whea , bioene gy) belonging o he same node
(c ops in his example) a e assumed o compe e mo e di ec ly wi h each o he han wi h hose land-uses in o he
nodes (e.g. o es o pas u e). Fo each land use, assump ions on ca bon s ocks in he ege a ion and he soil a e
made (see TableS1 in SM). A change in land co e ei he leads o posi i e o nega i e LUC emissions, d i en by
he di e ence in he assumed ca bon s ocks (in ege a ion and soil) be ween he o iginal and he new land use.
Based on he p o i abili y o each land use, which depends on assumed yields, p oduc ion cos s and commodi y
p ices, land owne s choose be ween di e en land uses o maximise p o i . Such land use decisions a e based
on he logi model o sha ing, aking in o accoun he he e ogenei y o local ci cums ances wi hin each AEZ,
and a oiding ex eme “winne - akes-i -all” ou comes55. See Sec ion1a in he SM o mo e de ails, and see Wise
e al.56 o a de ailed explana ion on he app oach and design o he land module in GCAM.
Sola land‑use module. An addi ional module has been de eloped o he GCAM model o link he con-
sump ion o sola ene gy wi h land use, compe ing wi h o he comme cial (c ops, imbe and in ensi e pas-
u es) and non-comme cial (na u al o es , g assland, sc ubland) land uses. Speci ically, he sola land ca ego y
is included in he “C ops” land node (Figu eS3 in SM), which means ha demand o sola land will p ima -
ily compe e wi h used, deg aded and po en ial c opland (including chemically e ilised meadows). Indi ec ly,
sola land also compe es wi h o he land uses such as o es , g ass- and sc ubland. This s uc u e is based on
obse ed endencies o sola si ing in Eu ope, India, Japan and Sou h-Ko ea (see TableS2 in SM), showing
ha mainly a able land is used o cu en USSE p ojec s, and suppo ed by academic li e a u e17,33,34,57,58 and
sola indus y epo s59,60. Also, he op imal mic oclima e o sola ene gy p oduc ion (based on insola ion, ai
empe a u e, wind speed and humidi y) is ound o e land ha is cu en ly used as c opland61, suppo ing he
assump ion ha u u e in es o s will ha e a sligh p e e ence o c opland (in use o allow) o he alloca ion
o sola ene gy p ojec s, among o he ac o s such as la ness and connec i i y in e ms o oads and elec ici y
g ids22. Ne e heless, an impo an d i e o land use decisions in he model is land p o i abili y: e en i land
co e ed by c op cul i a ion is pe cei ed as he mos sui able by in es o s in sola ene gy, high obse ed o po en-
ial p o i abili y o c op cul i a ion on such land could o ce in es o s o ocus on o he land ypes.
To de ine he alue o land o hos ing sola ene gy, a yield in e ms o ene gy ou pu pe uni o land has been
de ined o e e y AEZ. Equa ion(4) de ines his yield o each AEZ, which depends on a e age sola i adia ion
(I) pe AEZ, a e age e iciency o sola powe plan s ( 1) a he yea o ins alla ion ( ), he a e aged pe o mance
a io o e he li e cycle o he sola powe plan ( 2) and he land occupa ion a io ( 3)1,17. To es ima e I pe AEZ,
(1)
Land occupa ioni
,
p
,
=
land o ii
,
p
,
−
land o iNL
,
p
,
(2)
LUC pe ou pu uni
i,p, =

2020 o 2050
p, (LUCi−LUCNL)

2020 o 2050
p
,
(ou pu i
i
−ou pu i
NL)
(3)
CO
2payback pe iodi(l),p, ,a=

2020 o 2100
p, (LUCi−LUCNL
)
ou pu 2050=max
i(l)
∗a
9
Vol.:(0123456789)
Scien i ic Repo s | (2021) 11:2907 | h ps://doi.o g/10.1038/s41598-021-82042-5
www.na u e.com/scien i ic epo s/
we o e lapped he sola i adiance annual a e age da a62 ( il adia ion, i.e. he posi ion whe e he il coincides
wi h he la i ude, which is he op imal posi ion o PV panels o ake ad an age o he sola esou ce a each loca-
ion) wi h each AEZ and geopoli ical egion in GCAM 4.3 using a GIS ool. The land occupa ion a io, de ined
by Eq.(5), depends on he packing ac o (PF) and he Gene a o - o-sys em a ea (GSR). PF is he a io be ween
he PV panels o helios a s and he g ound a ea equi ed o a ays’ ins alla ion including sepa a ion o a oid
excessi e sel -shading, while GSR ep esen s he sha e o he ull a ea enclosed by he si e bounda y o he powe
plan which is co e ed by he PV panels and helios a s including he sepa a ion be ween hem. Hence, wi h ela-
ion o he PF, he GSR accoun s o he addi ional space equi ed o hos physical in as uc u e such as access
oads, subs a ions se ice buildings, and o he in as uc u e, as well as land no being able o be di ec ly used
due o o og aphy and une enness o he plo p e en ing he op imiza ion o he layou o he sola a ays. GSR
is dependen on he size and shape o he e ain and plo s and should be analysed on a case by case basis. GSR
anges o 0.7–0.85 ha e been epo ed63 al hough la ge plan s end o ha e lowe GSR due o mo e di icul
use-op imiza ion o land plo s a la ge scale p oduc ions, hence he e we ake a GSR o 0.7 assuming ha he
deploymen o scale o sola powe plan s on land will likely be based on la ge -size plan s due o he incen i es
o economics o scale17,19. The packing ac o again depends on he a e age la i ude o each AEZ and is de ined
by Eq.(6): he u he om he equa o , he mo e space is needed be ween he di e en panels o helios a s o
a oid sel -shading, so he lowe he packing ac o . The heo e ical equa ion o PF dependen on he sun ele a-
ion, he sun azimu h and he il angle, which can be simpli ied assuming ha il coincides wi h he la i ude
(β = ∅) and aking he conse a i e shading c i e ion o a oiding shading only a noon63. This o mula is only
alid o la i udes < 66.5° ( o ensu e PF > 0), bu in his s udy we cons ained sola deploymen in high la i udes
a eas since low sola i adiance in hese a eas make sola powe uneconomical (see Sec ion1c o he SM). Fo
simplici y, we ha e based he PF es ima ion on ixed acking PV sys ems on la land. Sola yields can sligh ly
di e (abou 25% in bo h ways) o 1- o 2-axis PV acking sys ems o o CSP sys ems19. See TableS5 in he
SM o he assumed alues o he pa ame e s in Eq.(4) o he ocus egions o his s udy.
Figu eS6 o he SM de ines he sola yield pe AEZ. No e ha his igu e only ep esen s he land inpu s pe
uni o ene gy ou pu . The capi al inpu s pe uni o ou pu depend only on IAEZ, 1 and 2 and since capi al cos s
end o be la ge han land cos s, in es o s in sola ene gy end o choose he loca ion p edominan ly based on
sola i adiance ins ead o he sola ene gy yield pe land uni . Consis en ly expo ing o impo ing la ge sha es
o sola ene gy be ween geog aphically and/o poli ically dis inc egions aces bo h echnical and geopoli ical
challenges. The e o e, we ha e chosen a conse a i e assump ion ha sola ene gy mus be p oduced and con-
sumed in he same geopoli ical GCAM egion.
Impac o sola ene gy in as uc u e on local ca bon cycle. The impac o USSE in as uc u es
on local mic oclima es is a ield in ea ly esea ch s ages, al hough some case s udies ha e been pe o med. In
he case o sola ene gy on pas u es in we clima es, a signi ican loss o ca bon in ege a ion and soils can be
expec ed in he land below he in as uc u e ha is pe manen ly blocked om sunligh , bu he yea - ound ca -
bon cycle in gap a eas be ween ows o sola panels will be ha dly a ec ed35. Howe e , in semi-a id pas u es wi h
we win e s, opposi e e ec s a e obse ed, and mic oclima es below panels seem o enhance ege a ion g ow h64.
E idence om ag i ol aic sys ems show ha yea - ound lea g ow h below sola in as uc u e is abou 20%
lowe below sola modules, and ha his e ec is s onges in summe 65. Wha can be concluded om hese
s udies is ha local mic oclima es depend on many ac o s, which a e ye oo unce ain o d aw obus assump-
ions om. Howe e , i also shows ha he design and managemen o sola pa ks is o high impo ance o
he ca bon cycle in such pa ks. Fo example, highe placed modules a e bene icial o ege a ion g ow h below
he modules, as i allows mo e sunligh o each he ege a ion. Bu a c ucial aspec o local ca bon cycles is
he decision on how o manage he land below he sola ene gy in as uc u e. Tha is why we ha e amed he
unce ain y ela ed o local ca bon cycle impac s on managemen decisions, based on h ee di e en egimes
iden i ied h ough li e a u e e iew:
• Land clea ing: Clea ing and g ubbing o soil and oo s, opsoil s ipping and s ockpiling, land g ading and
le elling, and soil compac ion. Exis ing ege a ion ha suppo s habi a is emo ed and any o he ege a ion
is o en discou aged; weeds and o he unwan ed ege a ion a e gene ally managed wi h he bicides and by
co e ing he g ound wi h g a el; his is a common p ac ice in a ious coun ies41,66. Modules a e placed a
g ound le el, which is cheape , and he absence o ege a ion a oids shading e ec s.
• Main aining p e ious ege a ion: Vege a ion as in p e ious land uses is as much as possible main ained, so
a able land s ays a able and pas u es s ay pas u es. All ege a ion in p e ious land co e abo e 30cm heigh ,
such as ees, bushes and high g ass, will be emo ed such ha he ege a ion ha is le is simila o ha in
pas u es. A eas di ec ly below sola modules a e clea ed o he cons uc ion phase, bu weeds migh g ow
a e ha phase. Modules a e placed sligh ly highe o a oid po en ial shading om ege a ion. This egime
(4)
ρe
AEZ =
I
AEZ ·
1
·
2
·
3
AEZ
(5)
3AEZ =GSR ·PFAEZ
(6)
PF
AEZ =(cosβAEZ +
sinβAEZ
an

66.55◦·(�
180
◦)−∅AEZ

)
−
1
;βand ∅
in adians