Using ai bo ne LiDAR and enhanced-geoloca ed GEDI me ics o map
s uc u al ai s o e a Medi e anean o es
Aa on Ca denas-Ma inez
a,*
, Ad ian Pascual
b
, Emilia Guisado-Pin ado
a
,
Vic o Rod iguez-Galiano
a
a
Depa amen o de Geog a ía Física y An´
alisis Geog ´
a ico Regional, Uni e sidad de Se illa, 41004, Se ille, Spain
b
Depa men o Geog aphical Sciences, Uni e si y o Ma yland, College Pa k, MD, Uni ed S a es
ARTICLE INFO
Da ase link: Using ai bo ne LiDAR and
enhanced-geoloca ed GEDI me ics o map
s uc u al ai s o e a Medi e anean o es
(O iginal da a)
Keywo ds:
Spacebo ne LiDAR
Ecological mapping
Geoloca ion
GEDI
Fo es s uc u e
ABSTRACT
The es ima ion o h ee-dimensional (3D) ege a ion me ics om space-bo ne LiDAR allows o cap u e spa io-
empo al ends in o es ecosys ems. S uc u al ai s om he NASA Global Ecosys em Dynamics In es iga ion
(GEDI) a e i al o suppo o es moni o ing, es o a ion and biodi e si y p o ec ion. The Medi e anean Basin is
home o elic o es species acing he consequences o in ensi ied clima e change e ec s and whose habi a s
ha e been p og essi ely sh inking o e ime. We used wo sou ces o 3D-s uc u al me ics, LiDAR poin clouds
and ull-wa e o m space-bo ne LiDAR om GEDI o es ima e o es s uc u e in a p o ec ed a ea o Sou he n
Spain, home o elic species in jeopa dy due o ecen ex eme wa e -s ess condi ions. We locally calib a ed
GEDI spacebo ne measu emen s using disc e e poin clouds collec ed by Ai bo ne Lase Scanne (ALS) o adjus
he geoloca ion o GEDI wa e o m me ics and o p edic GEDI s uc u al ai s such as canopy heigh , oliage
heigh di e si y o lea a ea index. Ou esul s showed signi ican imp o emen s in he e ie al o ecological
indica o s when using da a colloca ion be ween ALS poin clouds and compa able GEDI me ics. The bes esul s
o canopy heigh e ie al a e colloca ion yielded an RMSE o 2.6 m, when limi ed o o es -classi ied a eas and
la e ain, compa ed o an RMSE o 3.4 m wi hou colloca ion. T ends o oliage heigh di e si y (FHD; RMSE
=2.1) and lea a ea index (LAI; RMSE =1.6 m
2
/m
2
) we e less consis en han hose o canopy heigh bu
con i med he enhancemen de i ed om colloca ion. The wall- o-wall mapping o GEDI ai s amed o e ALS
su eys is cu en ly a ailable o moni o Medi e anean spa se moun ain o es s wi h su iciency. Ou esul s
showed ha combining di e en LiDAR pla o ms is pa icula ly impo an o mapping a eas whe e access o in-
si u da a is limi ed and especially in egions wi h ab up changes in ege a ion co e , such as Medi e anean
moun ainous o es s.
1. In oduc ion
The assessmen o h ee-dimensional (3D) e ical ege a ion s uc-
u e s ands as a key elemen in moni o ing e es ial ecosys ems, whe e
canopy heigh eme ges as a lagship indica o o nume ous moni o ing
ecosys em s a egies, modelling s udies and en i onmen al policies
(Bas os e al., 2022; Li e al., 2023). Fo ins ance, i s signi icance ex ends
o he es ima ion o abo eg ound biomass (AGB), which is a key
pa ame e in he assessmen and modelling o global ca bon luxes
(Dubayah e al., 2022; F iedlings ein e al., 2022; Ma e al., 2023).
Addi ionally, canopy heigh plays a pi o al ole in cha ac e izing habi a
s uc u al he e ogenei y as an impo an ac o in explaining biodi e -
si y spa ial pa e ns (Hakkenbe g e al., 2023; Ma selis e al., 2022;
To esani e al., 2023). Endemic o es s ep esen one o he global
biodi e si y ho spo s and mus -p ese ed ecosys ems (Dela aux e al.,
2023), bu clima e change and human p essu e a e jeopa dizing he
capabili y o species o adap as enough o esis dis u bances due o
s and eplacemen o p olonged hea wa es (Ande egg e al., 2015;
Ha mann e al., 2018). In he Medi e anean basin, he landscape is
unde going ans o ma ions d i en by d ough s, ex eme hea episodes
and inc easingly ecu en wild i es, impac ing ca bon luxes and
h ea ening he habi a s o endemic species (G ünig e al., 2023; Mo ei a
e al., 2011; Ru aul e al., 2020). Baseline o es maps o e hese
i eplaceable biodi e si y ecosys ems could help ace he e ec s o
clima e and land-co e change (Goe z e al., 2022; Ha is e al., 2021;
Po apo e al., 2021; Schimel e al., 2015).
* Co esponding au ho .
E-mail add ess: [email p o ec ed] (A. Ca denas-Ma inez).
Con en s lis s a ailable a ScienceDi ec
Science o Remo e Sensing
jou nal homepage: www.sciencedi ec .com/jou nal/science-o - emo e-sensing
h ps://doi.o g/10.1016/j.s s.2025.100195
Recei ed 22 Oc obe 2024; Recei ed in e ised o m 10 Janua y 2025; Accep ed 10 Janua y 2025
Science o Remo e Sensing 11 (2025) 100195
A ailable online 12 Janua y 2025
2666-0172/© 2025 Published by Else ie B.V. This is an open access a icle unde he CC BY-NC-ND license ( h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/ ).
Accu a ely moni o ing and unde s anding all he complexi y o
e es ial ecosys em p ocesses, dynamics, and ulne abili ies, as well as
de ining success ul managemen s a egies, la gely depend on he
a ailabili y o imely and high- esolu ion da a abou 3D ege a ion
s uc u e pa ame e s (G assi e al., 2017; Xu e al., 2021). The e olu ion
o emo e sensing ools, spanning om op ical and ada images o lase
scanne s, has signi ican ly enhanced he capabili y o map o es and
ee a ibu es (Ma in e al., 2016). Abo eg ound me ics om LiDAR
su eys a e solid desc ip i e pa ame e s o ege a ion e ical p o ile
and obus p oxies o es ima e o es s uc u e and AGB a mul iple
scales (Asne e al., 2014; Beland e al., 2019; Be gen e al., 2009).
Mo eo e , canopy co e me ics and o he ecological me ics such as
Lea A ea Index (LAI), Plan A ea Index (PAI) o Foliage Heigh Di e si y
(FHD) a e being inc easingly used o epo clima e change e ec s on
ege a ion and o in o m on o es dynamics (Rishmawi e al., 2021;
Schneide e al., 2020; Wang e al., 2022). These s uc u al ai s, along
wi h many o he ege a ion me ics, ha e been es ima ed globally using
space-bo ne LiDAR echnology h ough he NASA GEDI mission
(Dubayah e al., 2020, 2022).
O e he pas ou yea s, he GEDI lase s ha e p o iled o es can-
opies consis en ly ac oss opical, sub opical and empe a e o es s o
de i e o es canopy heigh , e ain condi ions and he e ical dis i-
bu ion o o es s (i.e. GEDI lase oo p in da a measu emen s in he
comple ed 2019/23 pe iod exceed he 25 billion). The consolida ion o
he GEDI mission owa ds 2030 is a majo suppo o o es ecology and
clima e science (Goe z e al., 2022). The calib a ion and alida ion o
GEDI ecological indica o s ind suppo in da a c osso e s o GEDI acks
and high- esolu ion 3D da a, sui able “labo a o ies” o compa e GEDI
s uc u al ai s and ecological indexes o simila , and some imes com-
pa able, lase -based me ics de i able om poin cloud da a ei he
ai bo ne-collec ed (e.g. Li e al., 2023; Pascual and Gue a-He nandez,
2023) o e es ial-based (Calde s e al., 2020).
S udies on he compa ison be ween GEDI ull wa e o m and ALS
poin clouds de i ed me ics can be spli in o wo g oups, conside ing
whe he hey ha e applied geoloca ion co ec ion me hods o no (Roy
e al., 2021; Tang e al., 2023). In spa se, open ecosys ems, some s udies
ha e used he GEDI simula o (Hancock e al., 2019) o co ec he
geoloca ion o on-o bi GEDI measu emen s h ough ALS, helping o
educe he 10-m (1-sigma) ho izon al unce ain y in he e sion 2
p oduc (Li e al., 2023; Pascual e al., 2023). Howe e , se e al s udies
ha e omi ed his c i ical s ep, acknowledging ha unce ain y in hei
esul s (e.g., Dha gay e al., 2022; Hue e mann e al., 2022; Pule i
e al., 2020; Wang e al., 2022). The cos o missing p ope geoloca ion
co ec ion me hods o GEDI measu emen s is highe in spa se, open
o es s whe e a subs an ial p opo ion o he g ound may be exposed. In
hese low-co e ecosys ems, a ho izon al o se o a ew me e s can
signi ican ly impac he accu a e e ie al o o es heigh e ical p o-
iles and co e me ics (e.g., Do ado-Roda e al., 2021; Po apo e al.,
2021). Assessing GEDI me ics o e spa se o es s is impo an as he
GEDI ins umen was designed o measu ing canopy e ical p o iles in
closed-canopy ecosys ems (condi ions o 95%–98% canopy co e , as
s a ed in Dubayah e al., 2020) and no o he moni o ing o spa se
o es ecosys ems.
The sou h o he Ibe ian Peninsula, wi hin he Medi e anean
con ex , is an in e es ing mosaic o discon inuous woody ege a ion
s uc u es ha a y g ea ly in e ical and spa ial a angemen and
change due o clima ic, ecological and managemen impac s
(Gonz´
alez-´
A ila e al., 2023). Mo eo e , Spain has a consolida ed ALS
su ey p og am wi h a high po en ial o cu en and u u e calib a ions
and alida ions o GEDI science mission p oduc s in spa se o es s,
among o he o es ypes (Pascual e al., 2023). Mos o he s udies
compa ing GEDI o lase poin clouds we e mos ly ocused on alida ing
ele a ion and ege a ion ela i e heigh (RH) me ics and less, espe-
cially o e spa se o es s, on o he ecological indica o s i.e., LAI, PAI o
FHD. In his s udy, we aim o assess he capabili y o on-o bi GEDI da a
o es ima e s uc u al ai s in moun ainous Medi e anean spa se
o es s by compa ing ends be ween on-o bi GEDI and simula ed GEDI
wa e o ms de i ed om ALS. By using high- esolu ion ALS da a o
co ec he geoloca ion o GEDI wa e o ms, we explo ed he ela ions
be ween s uc u al ai s o ege a ion, highligh ing he ole o
enhancing he geoloca ion o GEDI. This app oach seeks o o e insigh s
in o he unce ain y o GEDI me ics due o geoloca ion issues in hese
ecosys ems. O he speci ic objec i es add essed in his s udy a e: (1) o
explo e he e ec s o sou ces o unce ain y such as land-co e and
opog aphy in he es ima ions; and (2) o p oduce wall- o-wall maps o
o es s uc u e using ALS su eys as ec o s o assess GEDI’s s uc u al
ai s dominance a he landscape le el.
2. Ma e ial and me hods
2.1. S udy a ea
The s udy was conduc ed in a Medi e anean o es loca ed in
Sou he n Spain (Malaga p o ince in Andalusia, Spain; 36◦44
′
N, 4◦59
′
W). The selec ed s udy si e was Sie a de las Nie es (SN), a 230-km
2
ese oi o biodi e si y p o ec ed as a Na ional Pa k in 2021 (Fig. 1).
This a ea hos s a ema kable di e si y o plan species, wi h up o 1387
axa (Cabezudo e al., 2022), 79 o hem endemic o his egion. Among
hem, he mos dis inc i e species is he Abies pinsapo Boiss., which has
in SN i s la ges popula ion (~5800 ha). The A. pinsapo is a elic species
om he Te ia y cha ac e ized by i s py amidal shape which can each
up o 30 m in op-o -canopy, and whose p esence is ypically limi ed o
no h- acing slopes o e 1000 m abo e sea le el. He e, he species inds
he op imal condi ions o humidi y and empe a u e: he mean annual
empe a u e in SN is ~11 ◦C and he annual p ecipi a ion eaches 1400
mm (M´
endez-Cea e al., 2023), wi h in ense summe d ough s om June
o Oc obe . Below 1000 m abo e sea le el, A. pinsapo mingles wi h
Medi e anean coni e s such as Pinus halepensis Mill. and Pinus pinas e in
he No h-Eas (Lina es e al., 2011). O he species also p esen in SN
include Que cus aginea and Junnipe us communis, and unde s o y species
such as Ulex pe i lo us, Rubus ulmi olius and Sal ia osma inus.
Topog aphy in SN is pa icula ly challenging o he e ie al o
ege a ion s uc u al ai s due o he p esence o e y s eep slopes and
canyons ha exace ba e he complexi y o e ie ing accu a e p o iles o
o es ege a ion. The s uc u al complexi y o he canopy, wi h la ge
a ia ions in ee species heigh s and he p esence o a dense unde s o y,
mus also be conside ed. These condi ions ep esen a challenge o he
cha ac e iza ion o s uc u al ai s o ege a ion wi h emo e sensing
echniques and enabled o es he usabili y o GEDI in complex Medi-
e anean en i onmen s (Do ado-Roda e al., 2021).
2.2. Ai bo ne lase scanning da a
ALS da a acquisi ion was pe o med in he no heas o SN, o e a
mixed- o es a ea known as “Pue o Saucillo”, cha ac e ized by hos ing
one o he main A. pinsapo o es s in he sou he n Ibe ian Peninsula
(M´
endez-Cea e al., 2023). The al i ude o Pue o Saucillo (app oxi-
ma ely 1000–1100 m abo e sea le el) ma ks he ansi ional zone be-
ween he sou hwes e n a eas o he s udy si e, whe e A. pinsapo
domina es (Na a o-Ce illo e al., 2022), and he a eas whe e i is ound
in mixed o es s alongside pine species such as P. halepensis y P. pinas e
(de G´
al ez-Mon a˜
nez e al., 2024). This makes he s udy a ea he
lowes -al i ude zone wi hin he Na ional Pa k whe e A. pinsapo is p e-
sen (Lina es e al., 2011) and a ep esen a i e a ea o i s o es s. The
a ea (10.96 km
2
) was su eyed in Feb ua y 2020 using he Leica ALS60
lase scanne (Leica Geosys ems AG, Hee b ugg, Swi ze land) moun ed
on a Cessna-337 ai c a . Fligh al i ude was se o 300 m and 46.3 m/s
as nominal ligh speed o ensu e op imal op ical co e age o e s eep
slopes. The maximum scan angle o ±9◦ om nadi and abou 30% ligh
s ip o e lap led o an a e age poin densi y o 7.7 p m
−2
(Table 1). The
esul ing ALS da a accu acy was 30 cm in he ho izon al and 15 cm in
he e ical.
A. Ca denas-Ma inez e al.
Science o Remo e Sensing 11 (2025) 100195
2
The acqui ed lase poin cloud da a was p ocessed using Te ascan
(Te asolid, 2022) and he lidR 3.1. package (Roussel e al., 2020)
a ailable in he R s a is ical so wa e (R Co e Team, 2022). Isola ed
LiDAR poin s we e classi ied by iden i ying poin s wi h ewe neigh-
bou s wi hin a sea ch adius o 5 m, while low poin s we e classi ied by
iden i ying indi idual poin s o g oups o poin s lowe han a h eshold
o 0.5 m wi hin a 2D adius o 5 m. The lidR package was used o iden i y
g ound e u ns and c ea e he digi al e ain model (DTM) ollowing
s anda d ou ines in ALS-based o es in en o y (see Gue a-He n´
andez
and Pascual, 2021; Pascual e al., 2020). The Clo h Simula ion Fil e
(Zhang e al., 2016) was applied o de i e e ain ele a ion. Then, an
in e se dis ance weigh ed algo i hm was used o c ea e a 1-m esolu ion
DTM and a ine-g ained slope map. The DTM was used o no malize he
ALS poin clouds o abo e-g ound heigh s and p oduce a 0.5-m esolu-
ion canopy heigh model (CHM) (Fig. 2).
2.3. GEDI da a
GEDI da a wi hin he s udy a ea was e ie ed using he ALS
co e age o selec co- egis e ed GEDI oo p in s. Foo p in a iables
included e ain and o es canopy heigh me ics (GEDI L2A, Dubayah
e al., 2021a), canopy co e and densi y me ics (GEDI L2B, Dubayah
e al., 2021b) and he es ima es o abo eg ound biomass densi y (AGBD)
included in he L4A p oduc (GEDI L4A, Dubayah e al., 2022). The
p o ocol o he selec ion o high-quali y GEDI oo p in s in his s udy
was as ollows: wa e o m ideli y in he selec ed oo p in s exceeded
0.95 and was g ea e han canopy co e . The selec ion o he GEDI L2A
algo i hm o g ound inding was op imized o each oo p in . Foo -
p in s we e il e ed ou i he absolu e di e ences be ween he ele a ion
o he cen e o he lowes mode ela i e o e e ence ellipsoid and he
in e pola ed ele a ion o he TanDEM-X global DTM used in he GEDI
mission exceeded 50 m. We imposed a maximum h eshold o L4A
biomass es ima es (500 Mg ha
−1
) o emo e ou lie s passing he il e s
due o dense og and ugged opog aphy ha challenge he op imal
e ie al o GEDI ele a ion me ics. A e he quali y il e ing, 862 GEDI
oo p in s we e selec ed. These oo p in s a e con ained wi hin 10 GEDI
acks.
2.4. The GEDI simula o
The GEDI wa e o m simula o ool was de eloped o he p e-launch
calib a ion o GEDI and was used o he calib a ion o GEDI L4A models
(Duncanson e al., 2022). A comp ehensi e desc ip ion o he simula o
can be ound in Hancock e al. (2019). B ie ly, he simula o ope a es by
gene a ing i ual wa e o ms o gi en oo p in loca ions using disc e e
poin cloud da a. In his s udy, s uc u al me ics we e calcula ed wi h
Fig. 1. O e iew o he s udy a ea showing di e en o es ypes and condi ions cap u ed wi h he ai bo ne LiDAR su ey mos ly o e endemic o es si es domina ed
by Abies pinsapo and Pinus Halepensis.
Table 1
Ai bo ne Lase Scanning senso speci ica ions and ligh pa ame e s.
Fligh da e Feb ua y 2020
Senso Leica ALS60
Su eyed a ea (km
2
) 10.96
Fligh al i ude abo e g ound le el (m) ~300
Beam di e gence (m ad) 0.15
Wa eleng h (nm) 1064
O e lap (%) ~30
FOV (◦) 20
Poin densi y (p s m
−2
) 7.7
A. Ca denas-Ma inez e al.
Science o Remo e Sensing 11 (2025) 100195
3
espec o g ound in ull GEDI-like wa e o m simula ion. I should be
no ed ha RH we e de e mined in ela ion o g ound ele a ion, de i ed
di ec ly om ALS da a. This calcula ion used he cen e o g a i y o
poin s classi ied as g ound e u ns, a he han inding g ound ele a ion
om a Gaussian i o he lowes in lec ion poin i o he simula ed
wa e o m, as desc ibed in Duncanson e al. (2022). This app oach
allowed mi iga ing unce ain ies ela ed o he in e ac ions be ween he
RH signal and g ound iden i ica ion du ing he simula ion, especially in
a eas wi h s eep slopes (Liu e al., 2021).
2.4.1. Geoloca ion o GEDI high-quali y oo p in s
The calcula ion o he o se be ween GEDI oo p in s and ALS da a
ollowed he app oach p esen ed in Blai and Ho on (1999) and
implemen ed in he Colloca eWa es ool as pa o he GEDI simula o
(see Hancock e al., 2019 and he simula o ins uc ions). This me hod
uses he Pea son co ela ion o ind he bes a ine ans o ma ion in X, Y
and Z o align he la ge- oo p in GEDI da ase o a small- oo p in ALS
da ase . In ou s udy, he colloca ion o GEDI oo p in s was pe o med
using he ollowing pa ame e se ings as s a ing poin o ind he bes
ans o ma ions: geoe o swi ch wi h 20 m o expec ed e o and 0.5 m
o co ela ion dis ance; check co e o emo e he oo p in s wi h less
han 66% ALS co e age; and a GEDI beam sensi i i y o a leas 0.9 o
pe o m he colloca ion. The ou pu o he simula o a e wa e o ms o
each oo p in and h ee-dimensional co ec ion ac o by in e sec ing
o bi o co ec he geoloca ion o he oo p in s (Fig. 3).
To measu e he impac o geoloca ion co ec ion we compa ed pai s
o RH me ics (on-o bi e sus simula ed GEDI) o RH98 and wo GEDI
L2B me ics. RH98 was used as a p oxy o canopy heigh , being one o
he GEDI L2A RH me ics used in GEDI L4A models due o i s impo ance
in abo eg ound biomass densi y es ima ion (i.e., his p edic o is used in
GEDI biomass models o e Eu ope and many Wo ld egions as discussed
by Kellne e al. (2023). The GEDI L2B me ics included in ou s udy
we e FHD and LAI. These me ics a e ex ac ed om each GEDI wa e-
o m and a e based on he di ec ional gap p obabili y p o ile de i ed
om he L1B wa e o m (Tang and A ms on, 2019). Thei inclusion in
ou s udy e lec s hei ole as s uc u al ai s ha ep esen di e en
main a ibu es o o es s uc u e. FHD desc ibes he dis ibu ion o
oliage densi y ac oss e ical canopy laye s (MacA hu and MacA hu ,
1961; Valbuena e al., 2012) and se es as an indica o o he e ical
s a i ica ion and s uc u al complexi y o he canopy (A kins e al.,
Fig. 2. O e iew o he esea ch expe imen using GEDI high-quali y on-o bi da a and ai bo ne LiDAR da a. The esul ing LiDAR poin cloud da a was used o map
canopy heigh and slope e ain.
A. Ca denas-Ma inez e al.
Science o Remo e Sensing 11 (2025) 100195
4
2023). Meanwhile, LAI is de ined as he p ojec ed lea a ea wi hin a
canopy pe ho izon al g ound a ea (Asne e al., 2003), p o iding a
measu e o canopy a ea and densi y (A kins e al., 2023). I should be
no ed ha while on-o bi GEDI measu es PAI (p ojec ed plan a ea
wi hin a canopy pe ho izon al g ound a ea), i is eplaced in he GEDI
simula o by LAI. Ne e heless, bo h me ics a e closely ela ed, gi en
ha he unk and b anches ba ely con ibu e o he o al plan a ea
su ace (Kucha ik e al., 1998). Taking his in o conside a ion, we
ea ed bo h me ics as compa able, as done p e iously by Pimmasa n
e al. (2020) and Hue e mann e al. (2023). Fu he mo e, al hough he
GEDI Simula o allows o he calcula ion o LAI by heigh laye s, his
s udy conside ed he measu emen o LAI o he en i e e ical column.
The widesp ead use o FHD and LAI in s udies ocused on he es ima ion
o s uc u al and unc ional ai s o ege a ion is well-documen ed o
bo h ALS (e. g. Schneide e al., 2017; Zheng e al., 2021, 2022) and o
GEDI (e. g. Bouche e al., 2020; Dha gay e al., 2022; Hi schmugl e al.,
2023; Schneide e al., 2020). This b oad applicabili y allowed us o
assess GEDI’s pe o mance o bo h ai s in compa ison wi h o he
ecosys ems.
The geoloca ion enhancemen was e alua ed conside ing land-co e
he e ogenei y and e ain s eepness in SN. We used he 10-m V200 land-
co e p oduc om he Eu opean Space Agency (Zanaga e al., 2022)
now ope a ional in GEDI (L4B biomass, Dubayah e al., 2023) o classi y
oo p in s in o o es s, sh ublands and g asslands and he NASA Shu le
Rada Topog aphy Mission (SRTM) digi al ele a ion model o calcula e
e ain slope.
2.4.2. Simula ion o GEDI me ics a landscape-le el
The GEDI simula o was also used o simula e GEDI-like wa e o m
me ics o e wall- o-wall iles o ALS da a. The su eyed a ea using ALS
was g idded in o 25-m iles (14,575 iles) o simula e wa e o ms and
p edic on-o bi condi ions o he s udy a ea. The p edic ion was ach-
ie ed by es ablishing ela ionships be ween on-o bi and simula ed
condi ions h ough he compa ison o alues o he same GEDI me ics
(RH98, FHD, and LAI). These ela ionships we e cap u ed wi hin he
high-quali y GEDI oo p in s used o he analyses.
2.5. S uc u al ai s using ALS da a
To u he compa e GEDI p oduc s, we used ALS-de i ed ai s o
show di e ences in da a dis ibu ion and issues when p edic ing GEDI
s uc u al ai s o e complex o es s uc u es in s eep condi ions. Th ee
canopy- ela ed s uc u al ai s widely used in desc ibing o es s uc-
u al di e si y and measu able h ough ALS (e.g., Gelabe e al., 2020;
Schneide e al., 2017; Zheng e al., 2021) we e applied in his s udy. We
selec ed he 98 h pe cen ile heigh (P98; indica i e o canopy heigh ),
FHD (as p e iously desc ibed in Sec ion 2.4.1 as indica i e o he e -
ical dis ibu ion o canopy laye ing) and LAI (also desc ibed in Sec ion
2.4.1 as he p ojec ed su ace a ea o plan ma e ial pe uni g ound
a ea), s uc u al ai s ep esen a i e o heigh , canopy s uc u al
complexi y and ege a ion densi y espec i ely (A kins e al., 2023;
Valbuena e al., 2020). The ALS P98 was compa ed o GEDI RH98 and
used, along wi h FHD and LAI, in he subsequen spa ial p edic ion o
GEDI s uc u al ai s a he landscape le el (See Sec ion 2.6.). To de i e
P98, he ALS poin cloud was i s il e ed o e ain only he i s e u ns
using he il e _poi unc ion, and he 98 h quan ile heigh was hen
calcula ed using he g id_me ics unc ion a ailable in lidR. To es ima e
FHD, we used he unc ion FHD published in lea R R package 0.3.5
(Almeida e al., 2021). FHD was e ie ed om abundances conside ed
as pe - oxel ela i e lea a ea densi y (LAD) alues by applying he
Fig. 3. Figu e o 3 sub- igu es showing non-ALS-colloca ed posi ions (on-o bi ) and ALS-colloca ed posi ions wi h he co ec ions a e using he simula o (enhanced
on-o bi ). The ALS-based canopy heigh model is p esen ed in he backg ound. The 2D mean on-o bi geoloca ion co ec ion was 8m.
A. Ca denas-Ma inez e al.
Science o Remo e Sensing 11 (2025) 100195
5

Shannon–Weine di e si y unc ion, as desc ibed in MacA hu and
MacA hu (1961):
FHD = − ∑
i
pi*ln pi(1)
whe e pi is he p opo ion o he o al ege a ion (in his case, ALS
e u ns) ha is in he i h laye , s uc u ed in 2-m e ical in e als.
Finally, LAI was calcula ed in lea R using he me hod p oposed by
Almeida e al. (2019), based on he applica ion o he MacA hu -Ho n
equa ion (MacA hu and Ho n, 1969) o LAD o each 1-m oxel
popula ed wi h ege a ion:
⎧
⎪
⎨
⎪
⎩
LAD =ln(pulsesin
pulsesou )*1
K
LAI =∑LAD
(2)
whe e pulsesin and pulsesou ep esen he pulses ha en e ed each oxel
and passed h ough i , espec i ely. K ep esen s he Bee -Lambe Law
ex inc ion coe icien and depends p ima ily on he oliage dis ibu ion
and o ien a ion and he hickness o lea es and o es canopy (Kamoske
Fig. 4. Wo k low diag am showing all s eps in he me hodology applied. F om op o down: da a collec ion using UAV LiDAR, e ie al o GEDI obse a ions,
il e ing o ALS poin clouds and gene a ion o p oduc s, colloca ion o GEDI oo p in s using ALS o co ec geoloca ion e o , ALS-simula ion o GEDI me ics a
landscape scale and e alua ion o g idded maps o h ee indica o s: o es canopy heigh , Foliage Heigh Di e si y (FHD) and Lea A ea Index (LAI).
A. Ca denas-Ma inez e al.
Science o Remo e Sensing 11 (2025) 100195
6
e al., 2019; Weiss e al., 2004). K alue should be adjus ed in o de o
calib a e es ima ed LAI in o an independen LAI measu emen as ec-
ommended by Almeida e al. (2019), based on he use o ield mea-
su emen s. Following hese conside a ions, we employed a limi ed se o
six LAI ield measu emen s collec ed in June 2023 using a LAI-2200C
Plan Canopy Analize du ing a ield campaign o p o ide a calib a-
ion o K. The o al explained a iance (R
2
) was 0.757 be ween LAI
ALS
and LAI
ield
(see Supplemen a y Figu e s1). Taking in o accoun ha his
o mula assumes ha each LiDAR pulse is e ically inciden , we
imposed a maximum o 10◦o -nadi iew in he poin cloud (~75% o
he poin s had less han 5◦o -nadi iew), conside ing he ecom-
mended alues in Liu e al. (2018) and he limi a ions o ou LiDAR
su ey. T ai s we e ob ained o he en i e ALS ligh a ea a a 25-m
esolu ion co esponding wi h he GEDI oo p in .
2.6. Accu acy assessmen
Fo he assessmen o enhanced geoloca ion, we used linea models
and calcula ed he absolu e and ela i e oo mean squa ed e o
(RMSE), o al explained a iance (R
2
), and bias be ween dis ibu ions o
simula ed GEDI and on-o bi GEDI me ics (i.e., RH98, FHD and LAI).
The ALS-based benchma k me ics a sho le el we e also assessed using
on-o bi GEDI as e e ence. Fo he es ima ion o RH98 and FHD, all
a ailable on-o bi GEDI oo p in s we e used. Meanwhile, since LAI is
no only a s uc u al bu also a biophysical ai , i is mo e sensi i e o
changes due o clima e and o es dis u bances (Heiskanen e al., 2013;
Wu e al., 2018). The e o e, linea models de i ed om he GEDI oo -
p in s co esponding o he closes yea s (2019 and 2020) o he ALS
ligh we e es ed o es ima ing LAI. The spa ial p edic ions a 25-m
esolu ion using ALS-based es ima es and p edic ed on-o bi GEDI
alues de i ed om linea models using simula ed GEDI and on-o bi
GEDI oo p in s we e simila ly assessed a landscape le el. To isu-
alize endencies o GEDI s uc u al ai s in he s udy a ea, alues o
o es canopy heigh - RH98, FHD and LAI we e escaled be ween 0 and
1, using a min-max no maliza ion app oach o cap u e he dominance o
each indica o and i s spa ial a ia ion ac oss SN. We used an RGB
colou composi e o he s uc u al ai s. The eby, ed a eas we e
de ined as alues o RH98 >0.5, FHD <0.5 and LAI <0.5; g een a eas as
RH98 <0.5, FHD >0.5 and LAI <0.5; and blue a eas as RH98 <0.5, FHD
<0.5 and LAI >0.5. Finally, small whi e a eas esul ing om he com-
bina ion o high alues o each s uc u al ai s we e de ined as RH98,
FHD and LAI >0.75. A wo k low diag am summa izing all he s eps
ollowed in he me hodology is shown in Fig. 4.
3. Resul s
3.1. E ec o geoloca ion co ec ion o es ima e GEDI canopy heigh
Dis ibu ions o ALS-colloca ed and non-colloca ed (i.e., on-o bi
posi ions o he oo p in s) showed Pea son co ela ion alues abo e
0.85 o 98.7% o he oo p in s (See Table 2). The a e age 2D dis ance
be ween he pai o enhanced and non-enhanced geoloca ed oo p in s
was 9.4 m (median), 3.4 m (mode, he lowe o se co ec ion o he
in e sec ing GEDI L1B acks) and 8.0 m (mean). To compu e hese
summa ies, we a e aged he geoloca ion o se o high-quali y oo p in s
in e sec ing he s udy a ea.
The ALS-colloca ion o GEDI obse a ions educed he RMSE in
o es canopy heigh (RH98) by almos a me e and inc eased he R
2
om 0.52 o 0.62 (Table 3). Thus, RMSE in GEDI RH98 using he ALS-
colloca ed me hod was 4.36 m when compa ing simula ed s on-o bi
GEDI. Wi hou geoloca ion co ec ion, he alue inc eased o 5.3 m
(Table 3, Figu e s2). Gi en ha many s udies ha e compa ed ALS heigh
pe cen iles o GEDI ene gy-based ela i e heigh me ics, i is ele an o
show ha , ega dless o he geoloca ion co ec ion me hod, compa ing
ALS P98 o GEDI RH98 p oduced la ge disc epancies. Ou alues o
RMSE and bias we e simila , abo e 7 m and 5 m, espec i ely, in bo h
si ua ions. The quan ile dis ibu ions showed an imp o emen in he
alignmen be ween simula ed GEDI and on-o bi GEDI on he igh -side
(i.e., abo e he 95 h quan ile) o he RH98 spec um when using da a
colloca ion (Fig. 5). The compa ison o GEDI RH98 e sus ALS P98
con i med he a o e-desc ibed null e ec and e ealed a sho e domain
o ALS pe cen iles compa ed o bo h simula ed and on-o bi GEDI.
3.2. Impac o land-co e and slope on GEDI canopy heigh es ioma ion
Mo e han 90% o he selec ed oo p in s we e iden i ied as o es ed
a eas acco ding o he 10-m ESA land-co e p oduc implemen ed in
GEDI. Colloca ed wa e o ms o e o es s showed sys ema ically lowe
e o s in canopy heigh es ima ion compa ed o oo p in s anging o e
sh ublands o g asslands (Fig. 6). Speci ically, ALS-GEDI colloca ion
o e o es ed a eas led o a dec ease in RMSE om 5 m o 4.1 m.
Meanwhile, o g asslands and sh ublands, he RMSE dec eased om
7.6 m o 6.3 m.
Slope es ima es om he SRTM global p oduc highligh he sub-
s an ial imp o emen in GEDI accu acy o o es s in la condi ions
(slope below 10◦). He e, he RMSE was down o 2.6 m and he R
2
eached 0.83 when using he ALS-GEDI colloca ion (Fig. 7). Towa ds
mode a e (10–30◦) and s eep condi ions (abo e 30◦) he e o in he
e ie al o canopy heigh was 4.2 m and 5 m, espec i ely. The absence
o GEDI geoloca ion co ec ion showed a decline in R
2
by 10 poin s o
la condi ions and up o 20 poin s o s eep condi ions, whe e he RMSE
eached 6.4 m, ma king a 20% inc ease compa ed o ALS-colloca ed
condi ions (5.1 m) as shown in Fig. 7. I is no ewo hy ha mos ob-
se a ions (~70%) o his s udy a ea ange in mode a e e ain s eep-
ness (10–30◦).
3.3. Es ima ion and wall- o-wall mapping o GEDI s uc u al ai s
The ag eemen be ween on-o bi and simula ed GEDI es ima es o
Table 2
Geoloca ion o se s (ALS e sus on-o bi GEDI). The numbe s o each GEDI ack
ID indica e he Julian Da e and speci ic hou o acquisi ion.
GEDI L1B
T ack ID
dX
Eas ing
dY
No hing
dZ
Ele a ion
Pea son
co ela ion
GEDI
oo p in s
(n)
2019170142546 8.00 −5.00 0.00 0.921 197
2019338194614 8.00 2.00 0.00 0.879 76
2020052055355 2.12 −2.88 0.13 0.667 4
2020060024632 6.00 11.00 0.00 0.936 184
2020188065523 −5.29 −12.24 −0.27 0.940 57
2020314215842 2.65 2.10 0.23 0.943 207
2021197193255 8.00 −7.00 0.00 0.858 49
2022021233617 11.15 13.41 0.10 0.372 7
2022144225451 −1.35 0.89 0.39 0.896 75
2022336121243 1.00 −2.00 0.00 0.886 6
Table 3
Pe o mance o GEDI a es ima ing o es canopy heigh showing he e ec o
GEDI geoloca ion co ec ion using he GEDI simula o . Fi ing s a is ics we e
compu ed o pai s o on-o bi and simula ed ela i e heigh 98 (RH98) and
heigh pe cen ile 98 o ALS dis ibu ions.
S uc u al
indica o
Fi ing
S a is ic
ALS-colloca ed GEDI No geoloca ion
co ec ion
GEDI
eal
s
GEDI
sim
GEDI
eal
s ALS
GEDI
eal
s
GEDI
sim
GEDI
eal
s ALS
Rela i e
heigh 98
Canopy
heigh
(RH98, m)
R
2
0.624 0.383 0.519 0.354
RMSE 4.348 7.112 5.253 7.275
RMSE (%) 12.2 30.6 21.8 30.7
Bias −2.316 −5.273 −3.179 −5.338
Foo p in s
(n)
862 862 850 850
A. Ca denas-Ma inez e al.
Science o Remo e Sensing 11 (2025) 100195
7
FHD was weake compa ed o canopy heigh , as expec ed (Table 4). The
di e ence be ween colloca ed and non-colloca ed dis ibu ions o GEDI
FHD es ima ion was small, and bo h dis ibu ions showed R
2
alues
below 0.6 o GEDI- o-GEDI ends, which doubled he alues in he
GEDI- o-ALS es ima ion o FHD (See Figu e s3). On-o bi e sus simu-
la ed GEDI dis ibu ions o LAI showed poo p edic i e capabili y and
an impo an misma ch be ween simula ed da a and eal on-o bi mea-
su emen s (Table 4, Figu e s4). This misma ch was e iden h ough R
2
alues below 0.25 in all cases, e en when compa ed o ALS-based es i-
ma ion. The LAI p edic ion was pe o med using GEDI oo p in s in
2019 ha ou pe o med accu acies compu ed o he yea 2020: R
2
=
0.237 and 0.098, espec i ely, bo h using ALS-colloca ed GEDI
oo p in s.
GEDI s uc u al ai dis ibu ions we e also assessed o measu e he
impac o land-co e and slope on he es ima ions. Simila o canopy
heigh , colloca ed wa e o ms o e o es ed a eas consis en ly showed
lowe e o s o FHD compa ed o he es o he oo p in s (See
Figu e s5). Fu he mo e, he impac o he slope on GEDI FHD es ima es
o e o es ed a eas a ied signi ican ly depending on he colloca ion o
he wa e o ms (Figu e s6). FHD es ima es using he colloca ed wa e-
o ms showed a good pe o mance o la a eas (R
2
=0.657), dec easing
mode a ely o slopes g ea e han 10◦(R
2
=0.56 o mode a e slopes
and R
2
=0.6 o s eep slopes). Ne e heless, al hough he RMSE
emained simila (~2), FHD dis ibu ions using he non-colloca ed
wa e o ms showed g ea e a ia ion, anging om R
2
o 0.577 o la
a eas o an R
2
o 0.404 o s eep condi ions. On he o he hand, mo e
han 96% o GEDI oo p in s used o LAI es ima ion we e conside ed
o es ed a eas, making he land-co e analysis less consis en due o he
educed numbe o obse a ions in sh ublands and g asslands
(Figu e s7). The esul s showed ha he colloca ion o GEDI wa e o ms
sys ema ically imp o ed he pe o mance o he LAI es ima ion in all
land-co e classes when compa ing bo h simula ed GEDI and ALS o on-
o bi GEDI. He e, he bes pe o mances we e obse ed when compa ing
colloca ed on-o bi and simula ed GEDI LAI es ima es, wi h R
2
=0.235
in o es ed a eas and R
2
=0.291 in non- o es ed condi ions. In u n, he
impac o slope on GEDI LAI es ima es was lowe han he land-co e
and a ec ed by he small numbe o oo p in s in la a eas and hose
wi h slopes g ea e han 30◦. Thus, he ag eemen be ween on-o bi and
simula ed GEDI was weake o la a eas han hose wi h mode a e
slopes (Figu e s8). LAI es ima es o e mode a e slopes expe ienced also
he only imp o emen in he pe o mance a e applying he colloca ion
bo h o simula ed GEDI and ALS.
Linea ela ionships be ween on-o bi and simula ed GEDI oo p in s
we e used o p edic GEDI s uc u al ai s o he en i e s udy a ea
(Figs. 8 and 9). Fo canopy heigh , we obse ed a sys ema ic de ia ion in
he end line compa ing p edic ed GEDI RH98 o ALS P98. The co e-
la ion was ema kably high (abo e 0.95), bu he bias o 2.6 m
con i med wha was obse ed in he oo p in s (Fig. 8). Fo he case o
LAI o FHD, he alignmen o p edic ed GEDI- and ALS-based da a dis-
ibu ions was s ill high (R
2
>0.7) al hough he bias in LAI was
Fig. 5. Quan ile dis ibu ions showing he ela ionship be ween on-o bi GEDI es ima es o ela i e heigh 98, ALS-colloca ed GEDI es ima es o enhanced geo-
loca ion and he ALS 98 h heigh pe cen ile. Resul s a e p esen ed o GEDI colloca ion using ALS o co ec geoloca ion e o (colloca ed) and no geoloca ion
co ec ion, keeping on-o bi posi ions (non-colloca ed). The mean 2D o se be ween pai s o da a was 8 m.
A. Ca denas-Ma inez e al.
Science o Remo e Sensing 11 (2025) 100195
8
pa icula ly s ong, and es ima ions we e subs an ially displaced om
he 1:1 line. Dis ibu ions showed a cu -o alue in he low end due o
an o e es ima ion o GEDI, imposing a igh cons ain o desc ibe FHD
and especially LAI domains (Fig. 9). Despi e his a e ac , mos o he
obse a ions o FHD ollowed he 1:1 endline and he ela i e RMSE is
below 20%.
We used model p edic ions o GEDI and s uc u al ai s e ie ed
om ALS a 25 m o map di e ences. Fo he case o canopy heigh , he
di e ence be ween RH98 and P98 was sys ema ic ac oss he s udy a ea
(Fig. 8), wi h an o e es ima ion o RH98 compa ed o ALS P98. Fo FHD,
we obse ed ansi ion a eas whe e ew pa ches did no show he o e all
unde es ima ion o GEDI compa ed o ALS-based es ima es o FHD
(Fig. 9). The spa ial layou o LAI es ima es showed he highes spa ial
a iabili y in he es ima es: sys ema ically high alues o GEDI non-
o es ed a eas and wi h a canopy heigh alue o 0 in he ALS-de i ed
CHM p oduc explain he la ges di e ences, while GEDI p edic ed
low LAI compa ed o ALS o e dense o es a eas.
3.4. Assessing he dominance o s uc u al ai s
The in eg a ed ep esen a ion o o es s uc u e a 25-m esolu ion
using canopy heigh , FHD and LAI was use ul in de ec ing ansi ion
a eas showing ab up changes in he dominance o each ai (Fig. 10).
Red ones show high canopy heigh and low alues o bo h FHD and
LAI, co esponding o P. halepensis eaching up o 34 m all. These pine-
speci ic s ands plan ed o es o a ion decades ago show less s uc u al
a iabili y, especially in he uppe canopies, as con i med by low GEDI
es ima ion alues o FHD and LAI. Simila ly, high alues o RH98 and
FHD can be obse ed in he no he n pa (mixed s ands o P. halepensis
and P. Pinas e ). A eas ep esen ed in blue and g een ones in he
sou he n pa (A. pinsapo s ands) show high alues o LAI and FHD. In
his case, A. pinsapo o es s a e cha ac e ized by a high canopy densi y
and laye ing, o en associa ed wi h a dense unde s o y o J. communis o
R. ulmi olius. This phenomenon could be explained by he es o a ion
e o s ocussed on c ea ing he e ogeneous condi ions be ween old-
g ow h s ands and younge pa ches. Fu he , obse ed high alues o
LAI in cen al sec ions o he s udy a ea co espond o P. halepensis and
A. Pinsapo mixed s ands.
4. Discussion
S uc u al ai maps ob ained om GEDI and ALS can p o ide apid
baselines o suppo ing o es moni o ing and conse a ion. Ac ually,
he ca e ul selec ion o GEDI oo p in s o aining used p oduc s be-
ween op ical da a, ada da a, ai bo ne LiDAR and GEDI is a imely
esea ch opic (F ancini e al., 2022; Po apo e al., 2021; Qi e al., 2025;
Zhao e al., 2024). Fo example, F ancini e al. (2022) and Po apo e al.
(2021) used GEDI and Landsa o map changes in o es biomass due o
o es dis u bances. O he s udies as Qi e al. (2019) employed simula ed
GEDI and TanDEM-X InSAR da a o imp o e o es s uc u al mapping
ac oss se e al moun ainous and non-moun ainous o es ed a eas in he
Ame icas.
Al hough GEDI lase s we e no designed o ope a e in spa se
moun ain o es s, GEDI da a ha e been consis en ly used in sa annas
and discon inuous ege a ion be o e (Do ado-Roda e al., 2021; Ho ´
en
e al., 2023; Li e al., 2023). In his s udy, we assess he accu acy o
enhanced geoloca ed on-o bi GEDI o e a Medi e anean moun ainous
o es , being o ou knowledge he i s a emp o alida e GEDI canopy
heigh measu emen s in hese spa se, moun ainous Medi e anean en-
i onmen s. Ou esul s disen angle key ac o s a ec ing he accu acy o
GEDI measu emen s in hese ecosys ems. The e o e, we con ibu e o he
GEDI s a e-o - he-a , adding new insigh s om Medi e anean ecosys-
ems: spa se moun ain o es s wi h ab up ansi ions in e ms o species
dominance, o es co e and e ain s eepness.
4.1. Re ie al o o es s uc u e om GEDI
Nume ous s udies ha e compa ed GEDI me ics o ALS wi hou
add essing impo an issues on geoloca ion accu acy (Pule i e al.,
2020; Rishmawi e al., 2021; Zhu e al., 2022). Di ec ly compa ing
dis ibu ions o ALS heigh pe cen iles o GEDI ene gy-based pe cen iles
migh no be he mos app op ia e me hod o assess he GEDI
Fig. 6. Sca e plo s showing he accu acy o GEDI canopy heigh es ima es using ela i e heigh 98 (RH98). On-o bi colloca ed and non-colloca ed GEDI mea-
su emen s and simula ed da a using ALS a e compa ed. The dis ibu ion o ALS heigh pe cen iles is also p esen ed. Resul s a e p esen ed o land-co e class
“ o es s” and combining classes “sh ublands” and “g asslands” (ESA V200 10-m map).
A. Ca denas-Ma inez e al.
Science o Remo e Sensing 11 (2025) 100195
9
Goe z, S., Dubayah, R., Duncanson, L., 2022. Re isi ing he s a us o o es ca bon s ock
changes in he con ex o he measu emen and moni o ing needs, capabili ies and
po en ial o add essing educed emissions om de o es a ion and o es
deg ada ion. En i on. Res. Le . 17 (11), 111003. h ps://doi.o g/10.1088/1748-
9326/ac9c1d.
Gonz´
alez-´
A ila, S., O ega, E., Ma ín, B., 2023. Mul iscale agmen a ion o o es ypes
in Spain. Fo . Ecol. Manag. 546, 121317. h ps://doi.o g/10.1016/j.
o eco.2023.121317.
G assi, G., House, J., Den ene , F., Fede ici, S., den Elzen, M., Penman, J., 2017. The key
ole o o es s in mee ing clima e a ge s equi es science o c edible mi iga ion.
Na . Clim. Change 7 (3), 220–226. h ps://doi.o g/10.1038/nclima e3227.
G ünig, M., Seidl, R., Sen , C., 2023. Inc easing a idi y causes la ge and mo e se e e
o es i es ac oss Eu ope. Global Change Biol. 29 (6), 1648–1659. h ps://doi.o g/
10.1111/gcb.16547.
Gue a-He n´
andez, J., Pascual, A., 2021. Using GEDI lida da a and ai bo ne lase
scanning o assess heigh g ow h dynamics in as -g owing species: a showcase in
Spain. Fo es Ecosys ems 8 (1), 14. h ps://doi.o g/10.1186/s40663-021-00291-2.
Gue a-He n´
andez, J., Pe ei a, J.M.C., S o all, A., Pascual, A., 2024. Impac o i e
se e i y on o es s uc u e and biomass s ocks using NASA GEDI da a. Insigh s om
he 2020 and 2021 wild i e season in Spain and Po ugal. Sci. Rem. Sens. 9, 100134.
h ps://doi.o g/10.1016/j.s s.2024.100134.
Hakkenbe g, C.R., Tang, H., Bu ns, P., Goe z, S.J., 2023. Canopy s uc u e om space
using GEDI lida . F on . Ecol. En i on. 21 (1), 55–56. h ps://doi.o g/10.1002/
ee.2585.
Hancock, S., A ms on, J., Ho on, M., Sun, X., Tang, H., Duncanson, L.I., Kellne , J.R.,
Dubayah, R., 2019. The GEDI simula o : a la ge- oo p in wa e o m lida simula o
o calib a ion and alida ion o spacebo ne missions. Ea h Space Sci. 6 (2),
294–310. h ps://doi.o g/10.1029/2018EA000506.
Ha is, N.L., Gibbs, D.A., Baccini, A., Bi dsey, R.A., de B uin, S., Fa ina, M., Fa oyinbo, L.,
Hansen, M.C., He old, M., Hough on, R.A., Po apo , P.V., Sua ez, D.R., Roman-
Cues a, R.M., Saa chi, S.S., Slay, C.M., Tu ubano a, S.A., Tyuka ina, A., 2021.
Global maps o wen y- i s cen u y o es ca bon luxes. Na . Clim. Change 11 (3),
234–240. h ps://doi.o g/10.1038/s41558-020-00976-6.
Ha mann, H., Mou a, C.F., Ande egg, W.R.L., Rueh , N.K., Salmon, Y., Allen, C.D.,
A nd , S.K., B eshea s, D.D., Da i, H., Galb ai h, D., Ru h o , K.X., Wunde , J.,
Adams, H.D., Bloemen, J., Caille e , M., Cobb, R., Gessle , A., G ams, T.E.E.,
Jansen, S., e al., 2018. Resea ch on ie s o imp o ing ou unde s anding o
d ough -induced ee and o es mo ali y. New Phy ol. 218 (1), 15–28. h ps://doi.
o g/10.1111/nph.15048.
Heiskanen, J., Rau iainen, M., S enbe g, P., M˜
o us, M., Vesan o, V.-H., 2013. Sensi i i y
o na owband ege a ion indices o bo eal o es LAI, e lec ance seasonali y and
species composi ion. ISPRS J. Pho og amme y Remo e Sens. 78, 1–14. h ps://doi.
o g/10.1016/j.isp sjp s.2013.01.001.
Hi schmugl, M., Lippl, F., Sobe, C., 2023. Assessing he e ical s uc u e o o es s using
ai bo ne and spacebo ne LiDAR da a in he Aus ian alps. Rem. Sens. 15 (3), 664.
h ps://doi.o g/10.3390/ s15030664.
Ho ´
en, R., Lamelas, M.T., de la Ri a, J., Domingo, D., Mon ealeg e, A.L., Ga cía-
Ma ín, A., Re illa, S., 2023. Assessing GEDI-NASA sys em o o es uels
classi ica ion using machine lea ning echniques. In . J. Appl. Ea h Obs. Geoin .
116, 103175. h ps://doi.o g/10.1016/j.jag.2022.103175.
Hue e mann, S., Jones, S., So o-Be elo , M., Hislop, S., 2022. In e compa ison o eal
and simula ed GEDI obse a ions ac oss scle ophyll o es s. Rem. Sens. 14 (9), 2096.
h ps://doi.o g/10.3390/ s14092096.
Hue e mann, S., Jones, S., So o-Be elo , M., Hislop, S., 2023. Using Landsa ime se ies
and bi- empo al GEDI o compa e spec al and s uc u al ege a ion esponses a e
i e. In . J. Appl. Ea h Obs. Geoin . 122, 103403. h ps://doi.o g/10.1016/j.
jag.2023.103403.
Ilangakoon, N., Glenn, N.F., Schneide , F.D., Dash i, H., Hancock, S., Spae e, L.,
Goulden, T., 2021. Ai bo ne and spacebo ne lida e eal ends and pa e ns o
unc ional di e si y in a semi-a id ecosys em. F on ie s in Remo e Sensing 2. h ps://
doi.o g/10.3389/ sen.2021.743320.
Kamoske, A.G., Dahlin, K.M., S a k, S.C., Se bin, S.P., 2019. Lea a ea densi y om
ai bo ne LiDAR: compa ing senso s and esolu ions in a empe a e b oadlea o es
ecosys em. Fo . Ecol. Manag. 433, 364–375. h ps://doi.o g/10.1016/j.
o eco.2018.11.017.
Kellne , J.R., A ms on, J., Duncanson, L., 2023. Algo i hm heo e ical basis documen o
GEDI oo p in abo eg ound biomass densi y. Ea h Space Sci. 10 (4). h ps://doi.
o g/10.1029/2022EA002516.
Kucha ik, C.J., No man, J.M., Gowe , S.T., 1998. Measu emen s o b anch a ea and
adjus ing lea a ea index indi ec measu emen s. Ag ic. Fo . Me eo ol. 91 (1–2),
69–88. h ps://doi.o g/10.1016/S0168-1923(98)00064-1.
Li, X., Wessels, K., A ms on, J., Duncanson, L., U bazae , M., Naidoo, L., Ma hieu, R.,
Main, R., 2024. E alua ion o GEDI oo p in le el biomass models in Sou he n
A ican Sa annas using ai bo ne LiDAR and ield measu emen s. Science o Remo e
Sensing 10, 100161. h ps://doi.o g/10.1016/j.s s.2024.100161.
Li, X., Wessels, K., A ms on, J., Hancock, S., Ma hieu, R., Main, R., Naidoo, L.,
E asmus, B., Scholes, R., 2023. Fi s alida ion o GEDI canopy heigh s in A ican
sa annas. Remo e Sensing o En i onmen 285, 113402. h ps://doi.o g/10.1016/j.
se.2022.113402.
Lina es, J.C., Delgado-Hue as, A., Ca ei a, J.A., 2011. Clima ic ends and di e en
d ough adap i e capaci y and ulne abili y in a mixed Abies pinsapo–Pinus
halepensis o es . Clima ic Change 105 (1–2), 67–90. h ps://doi.o g/10.1007/
s10584-010-9878-6.
Liu, A., Cheng, X., Chen, Z., 2021. Pe o mance e alua ion o GEDI and ICESa -2 lase
al ime e da a o e ain and canopy heigh e ie als. Remo e Sensing o
En i onmen 264, 112571. h ps://doi.o g/10.1016/j. se.2021.112571.
Liu, J., Skidmo e, A.K., Jones, S., Wang, T., Heu ich, M., Zhu, X., Shi, Y., 2018. La ge o -
nadi scan angle o ai bo ne LiDAR can se e ely a ec he es ima es o o es
s uc u e me ics. ISPRS J. Pho og amme y Remo e Sens. 136, 13–25. h ps://doi.
o g/10.1016/j.isp sjp s.2017.12.004.
Ma, L., Hu , G., Tang, H., Lamb, R., Lis e , A., Chini, L., Dubayah, R., A ms on, J.,
Campbell, E., Duncanson, L., Healey, S., O’Neil-Dunne, J., O , L., Poul e , B.,
Shen, Q., 2023. Spa ial he e ogenei y o global o es abo eg ound ca bon s ocks and
luxes cons ained by spacebo ne lida da a and mechanis ic modeling. Global
Change Biol. 29 (12), 3378–3394. h ps://doi.o g/10.1111/gcb.16682.
MacA hu , R.H., Ho n, H.S., 1969. Foliage p o ile by e ical measu emen s. Ecology 50
(5), 802–804. h ps://doi.o g/10.2307/1933693.
MacA hu , R.H., MacA hu , J.W., 1961. On bi d species di e si y. Ecology 42 (3),
594–598. h ps://doi.o g/10.2307/1932254.
Mandl, L., S i ih, A., Seidl, R., Ginzle , C., Sen , C., 2023. Spacebo ne LiDAR o
cha ac e izing o es s uc u e ac oss scales in he Eu opean Alps. Remo e Sensing in
Ecology and Conse a ion. h ps://doi.o g/10.1002/ se2.330.
Ma selis, S.M., Keil, P., Chase, J.M., Dubayah, R., 2022. The use o GEDI canopy s uc u e
o explaining a ia ion in ee species ichness in na u al o es s. En i on. Res. Le .
17 (4), 045003. h ps://doi.o g/10.1088/1748-9326/ac583 .
Ma in, D.C., Koh, L.P., Lynam, A.J., Wich, S., Da ies, A.B., K ishnamu hy, R.,
S okes, E., S a key, R., Asne , G.P., 2016. In eg a ing echnologies o scalable
ecology and conse a ion. Global Ecology and Conse a ion 7, 262–275. h ps://doi.
o g/10.1016/j.gecco.2016.07.002.
M´
endez-Cea, B., Ga cía-Ga cía, I., Lina es, J.C., Gallego, F.J., 2023. Wa ming appea s as
he main isk o non-adap edness o wes e n Medi e anean elic i o es s unde
expec ed clima e change scena ios. F on . Plan Sci. 14. h ps://doi.o g/10.3389/
pls.2023.1155441.
Mo ei a, F., Viedma, O., A ianou sou, M., Cu , T., Kou sias, N., Rigolo , E., Ba ba i, A.,
Co ona, P., Vaz, P., Xan hopoulos, G., Mouillo , F., Bilgili, E., 2011. Landscape –
wild i e in e ac ions in sou he n Eu ope: implica ions o landscape managemen .
J. En i on. Manag. 92 (10), 2389–2402. h ps://doi.o g/10.1016/j.
jen man.2011.06.028.
Næsse , E., 1997. De e mina ion o mean ee heigh o o es s ands using ai bo ne lase
scanne da a. ISPRS J. Pho og amme y Remo e Sens. 52 (2), 49–56. h ps://doi.
o g/10.1016/S0924-2716(97)83000-6.
Na a o-Ce illo, R.M., Gonz´
alez-Mo eno, P., Ruiz-G´
omez, F.J., S´
anchez-Cues a, R.,
Gazol, A., Cama e o, J.J., 2022. D ough s ess and pes s inc ease de olia ion and
mo ali y a es in ulne able Abies pinsapo o es s. Fo . Ecol. Manag. 504, 119824.
h ps://doi.o g/10.1016/j. o eco.2021.119824.
Oli ei a, P.V.C., Zhang, X., Pe e son, B., Ome o, J.P., 2023. Using simula ed GEDI
wa e o ms o e alua e he e ec s o beam sensi i i y and e ain slope on GEDI L2A
ela i e heigh me ics o e he B azilian Amazon Fo es . Science o Remo e Sensing
7, 100083. h ps://doi.o g/10.1016/j.s s.2023.100083.
Pascual, 2019. Using ee de ec ion based on ai bo ne lase scanning o imp o e o es
in en o y conside ing edge e ec s and he Co- egis a ion ac o . Rem. Sens. 11 (22),
2675. h ps://doi.o g/10.3390/ s11222675.
Pascual, A., Gue a-He nandez, J., 2023. Co ec ion o phenology-induced e ec s in
o es canopy heigh models based on ai bo ne lase scanning da a. Insigh s om he
deciduous moun ain o es s in Picos de Eu opa Na ional Pa k in Spain. Ecol. In . 75,
102092. h ps://doi.o g/10.1016/j.ecoin .2023.102092.
Pascual, A., Gue a-He n´
andez, J., A ms on, J., Mino , D.M., Duncanson, L., May, P.B.,
Kellne , J.R., Dubayah, R., 2023. Assessing he pe o mance o NASA’s GEDI L4A
oo p in abo eg ound biomass densi y models using Na ional Fo es In en o y and
ai bo ne lase scanning da a in Medi e anean o es ecosys ems. Fo . Ecol. Manag.
538, 120975. h ps://doi.o g/10.1016/j. o eco.2023.120975.
Pascual, A., Gue a-He n´
andez, J., Cosenza, D.N., Sando al, V., 2020. The ole o
imp o ed g ound posi ioning and o es s uc u al complexi y when pe o ming
o es in en o y using ai bo ne lase scanning. Rem. Sens. 12 (3). h ps://doi.o g/
10.3390/ s12030413.
Pimmasa n, S., T ipa hi, N.K., Ninsawa , S., Sasaki, N., 2020. Applying LiDAR o quan i y
he plan a ea index along a successional g adien in a opical o es o Thailand.
Fo es s 11 (5), 520. h ps://doi.o g/10.3390/ 11050520.
Po apo , P., Li, X., He nandez-Se na, A., Tyuka ina, A., Hansen, M.C., Komma eddy, A.,
Pickens, A., Tu ubano a, S., Tang, H., Sil a, C.E., A ms on, J., Dubayah, R., Blai , J.
B., Ho on, M., 2021. Mapping global o es canopy heigh h ough in eg a ion o
GEDI and Landsa da a. Remo e Sensing o En i onmen 253, 112165. h ps://doi.
o g/10.1016/j. se.2020.112165.
Pule i, N., G o i, M., Fe a a, C., Chianucci, F., 2020. Lida -based es ima es o
abo eg ound biomass h ough g ound, ae ial, and sa elli e obse a ion: a case s udy
in a Medi e anean o es . J. Appl. Remo e Sens. 14 (4). h ps://doi.o g/10.1117/1.
JRS.14.044501.
Qi, W., A ms on, J., Choi, C., S o all, A., Saa ela, S., Pa dini, M., Fa oyinbo, L.,
Papa hanassiou, K., Pascual, A., Dubayah, R., 2025. Mapping la ge-scale pan opical
o es canopy heigh by in eg a ing GEDI lida and TanDEM-X InSAR da a. Rem.
Sens. En i on. 318, 114534. h ps://doi.o g/10.1016/j. se.2024.114534.
Qi, W., Lee, S.-K., Hancock, S., Lu hcke, S., Tang, H., A ms on, J., Dubayah, R., 2019.
Imp o ed o es heigh es ima ion by usion o simula ed GEDI Lida da a and
TanDEM-X InSAR da a. Remo e Sensing o En i onmen 221, 621–634. h ps://doi.
o g/10.1016/j. se.2018.11.035.
Qui os, E., Polo, M.-E., F agoso-Campon, L., 2021. GEDI ele a ion accu acy assessmen : a
case s udy o sou hwes Spain. IEEE J. Sel. Top. Appl. Ea h Obs. Rem. Sens. 14,
5285–5299. h ps://doi.o g/10.1109/JSTARS.2021.3080711.
R Co e Team, 2022. R: A Language and En i onmen o S a is ical Compu ing. R
Founda ion o S a is ical Compu ing.
A. Ca denas-Ma inez e al.
Science o Remo e Sensing 11 (2025) 100195
16

Rishmawi, K., Huang, C., Zhan, X., 2021. Moni o ing key o es s uc u e a ibu es
ac oss he con e minous Uni ed S a es by in eg a ing GEDI LiDAR measu emen s and
VIIRS da a. Rem. Sens. 13 (3), 442. h ps://doi.o g/10.3390/ s13030442.
Roussel, J.-R., Au y, D., Coops, N.C., Tompalski, P., Goodbody, T.R.H., Meado , A.S.,
Bou don, J.-F., de Boissieu, F., Achim, A., 2020. lidR: an R package o analysis o
Ai bo ne Lase Scanning (ALS) da a. Remo e Sensing o En i onmen 251, 112061.
h ps://doi.o g/10.1016/j. se.2020.112061.
Roy, D.P., Kashongwe, H.B., A ms on, J., 2021. The impac o geoloca ion unce ain y on
GEDI opical o es canopy heigh es ima ion and change moni o ing. Science o
Remo e Sensing 4, 100024. h ps://doi.o g/10.1016/j.s s.2021.100024.
Ru aul , J., Cu , T., Mo on, V., T igo, R.M., Mouillo , F., Kou sias, N., Pimon , F.,
Ma in-S Paul, N., Ba be o, R., Dupuy, J.-L., Russo, A., Belhadj-Khedhe , C., 2020.
Inc eased likelihood o hea -induced la ge wild i es in he Medi e anean Basin. Sci.
Rep. 10 (1), 13790. h ps://doi.o g/10.1038/s41598-020-70069-z.
Schimel, D., Pa lick, R., Fishe , J.B., Asne , G.P., Saa chi, S., Townsend, P., Mille , C.,
F ankenbe g, C., Hibba d, K., Cox, P., 2015. Obse ing e es ial ecosys ems and he
ca bon cycle om space. Global Change Biol. 21 (5), 1762–1776. h ps://doi.o g/
10.1111/gcb.12822.
Schneide , F.D., Fe az, A., Hancock, S., Duncanson, L.I., Dubayah, R.O., Pa lick, R.P.,
Schimel, D.S., 2020. Towa ds mapping he di e si y o canopy s uc u e om space
wi h GEDI. En i on. Res. Le . 15 (11), 115006. h ps://doi.o g/10.1088/1748-
9326/ab9e99.
Schneide , F.D., Mo sdo , F., Schmid, B., Pe chey, O.L., Hueni, A., Schimel, D.S.,
Schaepman, M.E., 2017. Mapping unc ional di e si y om emo ely sensed
mo phological and physiological o es ai s. Na . Commun. 8 (1), 1441. h ps://
doi.o g/10.1038/s41467-017-01530-3.
S i ih, A., Seidl, R., Sen , C., 2023. Al e na i e s a es in he s uc u e o moun ain o es s
ac oss he {Alps} and he ole o dis u bance and eco e y. Landsc. Ecol. h ps://doi.
o g/10.1007/s10980-023-01597-y.
Tang, H., A ms on, J., 2019. Algo i hm Theo e ical Basis Documen (ATBD) o GEDI L2B
Foo p in Canopy Co e and Ve ical P o ile Me ics.
Tang, H., S oke , J., Lu hcke, S., A ms on, J., Lee, K., Blai , B., Ho on, M., 2023.
E alua ing and mi iga ing he impac o sys ema ic geoloca ion e o on canopy
heigh measu emen pe o mance o GEDI. Remo e Sensing o En i onmen 291,
113571. h ps://doi.o g/10.1016/j. se.2023.113571.
Te asolid, 2022. Te ascan (021.005).
To esani, M., Rocchini, D., Albe i, A., Moud ý, V., Heym, M., Thou e ai, E., Kacic, P.,
Tomelle i, E., 2023. LiDAR GEDI de i ed ee canopy heigh he e ogenei y e eals
pa e ns o biodi e si y in o es ecosys ems. Ecol. In . 76, 102082. h ps://doi.o g/
10.1016/j.ecoin .2023.102082.
U bazae , M., Hess, L.L., Hancock, S., Sa o, L.Y., Ome o, J.P., Thiel, C., Dubois, C.,
Heckel, K., U ban, M., Adam, M., Schmullius, C., 2022. Assessmen o e ain
ele a ion es ima es om ICESa -2 and GEDI spacebo ne LiDAR missions ac oss
di e en land co e and o es ypes. Science o Remo e Sensing 6, 100067. h ps://
doi.o g/10.1016/j.s s.2022.100067.
Valbuena, R., O’Conno , B., Zellwege , F., Simonson, W., Vihe aa a, P., Mal amo, M.,
Sil a, C.A., Almeida, D.R.A., Danks, F., Mo sdo , F., Chi ici, G., Lucas, R.,
Coomes, D.A., Coops, N.C., 2020. S anda dizing ecosys em mo phological ai s om
3D in o ma ion sou ces. T ends Ecol. E ol. 35 (8), 656–667. h ps://doi.o g/
10.1016/j. ee.2020.03.006.
Valbuena, R., Packalen, P., Ma ın-Fe nandez, S., Mal amo, M., 2012. Di e si y and
equi abili y o de ing p o iles applied o s udy o es s uc u e. Fo . Ecol. Manag. 276,
185–195. h ps://doi.o g/10.1016/j. o eco.2012.03.036.
Vogele , J.C., Feke y, P.A., Ellio , L., Swayze, N.C., Filippelli, S.K., Ba y, B.,
Holb ook, J.D., Vie ling, K.T., 2023. E alua ing GEDI da a usions o con inuous
cha ac e iza ions o o es wildli e habi a . F on ie s in Remo e Sensing 4. h ps://
doi.o g/10.3389/ sen.2023.1196554.
Wang, C., Elmo e, A.J., Numa a, I., Coch ane, M.A., Shaogang, L., Huang, J., Zhao, Y.,
Li, Y., 2022. Fac o s a ec ing ela i e heigh and g ound ele a ion es ima ions o
GEDI among o es ypes ac oss he con e minous USA. GIScience & Remo e
Sensing 59 (1), 975–999. h ps://doi.o g/10.1080/15481603.2022.2085354.
Weiss, M., Ba e , F., Smi h, G.J., Jonckhee e, I., Coppin, P., 2004. Re iew o me hods o
in si u lea a ea index (LAI) de e mina ion. Ag icul u al and Fo es Me eo ology 121
(1–2), 37–53. h ps://doi.o g/10.1016/j.ag o me .2003.08.001.
Wu, J., Kobayashi, H., S a k, S.C., Meng, R., Guan, K., T an, N.N., Gao, S., Yang, W.,
Res epo-Coupe, N., Miu a, T., Oli ie a, R.C., Roge s, A., Dye, D.G., Nelson, B.W.,
Se bin, S.P., Hue e, A.R., Saleska, S.R., 2018. Biological p ocesses domina e
seasonali y o emo ely sensed canopy g eenness in an Amazon e e g een o es .
New Phy ologis 217 (4), 1507–1520. h ps://doi.o g/10.1111/nph.14939.
Xu, L., Saa chi, S.S., Yang, Y., Yu, Y., Pong a z, J., Bloom, A.A., Bowman, K., Wo den, J.,
Liu, J., Yin, Y., Domke, G., McRobe s, R.E., Woodall, C., Nabuu s, G.-J., De-
Miguel, S., Kelle , M., Ha is, N., Maxwell, S., Schimel, D., 2021. Changes in global
e es ial li e biomass o e he 21s cen u y. Science Ad ances 7 (27), eabe9829.
h ps://doi.o g/10.1126/sciad .abe9829.
Zanaga, D., Van De Ke cho e, R., Daems, D., De Kee smaecke , W., B ockmann, C.,
Ki ches, G., We e s, J., Ca us, O., San o o, M., F i z, S., Lesi , M., He old, M.,
Tsendbaza , N.-E., Xu, P., Ramoino, F., A ino, O., 2022. ESA Wo ldCo e 10 M 2021
V200. Zenodo. h ps://doi.o g/10.5281/zenodo.7254221.
Zhang, W., Qi, J., Wan, P., Wang, H., Xie, D., Wang, X., Yan, G., 2016. An easy- o-use
ai bo ne LiDAR da a il e ing me hod based on clo h simula ion. Remo e Sensing 8
(6), 501. h ps://doi.o g/10.3390/ s8060501.
Zhao, X., Chen, J.M., Zhang, Y., Jiao, Z., Liu, L., Qiu, F., Zang, J., Cao, R., 2024. Global
mapping o o es clumping index based on GEDI canopy heigh and complemen a y
da a. ISPRS Jou nal o Pho og amme y and Remo e Sensing 209, 1–16. h ps://doi.
o g/10.1016/j.isp sjp s.2024.01.014.
Zheng, Z., Zeng, Y., Schneide , F.D., Zhao, Y., Zhao, D., Schmid, B., Schaepman, M.E.,
Mo sdo , F., 2021. Mapping unc ional di e si y using indi idual ee-based
mo phological and physiological ai s in a sub opical o es . Remo e Sensing o
En i onmen 252, 112170. h ps://doi.o g/10.1016/j. se.2020.112170.
Zheng, Z., Zeng, Y., Schuman, M.C., Jiang, H., Schmid, B., Schaepman, M.E.,
Mo sdo , F., 2022. Indi idual ee-based s pixel-based app oaches o mapping
o es unc ional ai s and di e si y by emo e sensing. In e na ional Jou nal o
Applied Ea h Obse a ion and Geoin o ma ion 114, 103074. h ps://doi.o g/
10.1016/j.jag.2022.103074.
Zhu, X., Nie, S., Wang, C., Xi, X., Lao, J., Li, D., 2022. Consis ency analysis o o es
heigh e ie als be ween GEDI and ICESa -2. Remo e Sensing o En i onmen 281,
113244. h ps://doi.o g/10.1016/j. se.2022.113244.
Ziegle , A., Heisig, J., Ludwig, M., Reudenbach, C., Meye , H., Nauss, T., 2023. Using
GEDI as aining da a o an ongoing mapping o landscape-scale dynamics o he
plan a ea index. En i onmen al Resea ch Le e s 18 (7), 075003. h ps://doi.o g/
10.1088/1748-9326/acde8 .
Glossa y
AGBD: Abo eg ound Biomass Densi y
ALS: Ai bo ne Lase Scanning
CHM: Canopy Heigh Model
DTM: Digi al Te ain Model
FHD: Foliage Heigh Di e si y
GEDI: Global Ecosys em Dynamics In es iga ion
LAI: Lea A ea Index
NFI: Na ional Fo es In en o y
P: Pe cen ile
PAI: Plan A ea Index
RH: Rela i e Heigh
A. Ca denas-Ma inez e al.
Science o Remo e Sensing 11 (2025) 100195
17