RESEARCH Open Access
P edic ing he p o isioning po en ial o
o es ecosys em se ices using ai bo ne
lase scanning da a and o es esou ce maps
Ja i Vauhkonen
Abs ac
Backg ound: Remo e sensing-based mapping o o es Ecosys em Se ice (ES) indica o s has become inc easingly
popula . The esul ing maps may enable o spa ially assess he p o isioning po en ial o ESs and p io i ize he land
use in subsequen decision analyses. Howe e , he mapping is o en based on eadily a ailable da a, such as land
co e maps and o he publicly a ailable da abases, and igno ing he ela ed unce ain ies.
Me hods: This s udy es ed he po en ial o imp o e he obus ness o he decisions by means o local model i ing
and unce ain y analysis. The quali y o o es land use p io i iza ion was e alua ed unde wo di e en decision suppo
models: ei he using he de eloped models de e minis ically o in co po a ion wi h he unce ain ies o he models.
Resul s: P edic ion models based on Ai bo ne Lase Scanning (ALS) da a explained he a ia ion in p oxies o he sui abili y
o o es plo s o main aining biodi e si y, p oducing imbe , s o ing ca bon, o p o iding ec ea ional uses (be y picking
and isual ameni y) wi h RMSEs o 15%–30%, depending on he ES. The RMSEs o he ALS-based p edic ions we e 47%–97%
o hose de i ed om o es esou ce maps wi h a simila esolu ion. Due o applying a simila ield calib a ion s ep on bo h
o he da a sou ces, he di e ence can be a ibu ed o he be e abili y o ALS o explain he a ia ion in he ES p oxies.
Conclusions: Despi e he di e en accu acies, p oxy alues p edic ed by bo h he da a sou ces could be used o a
pixel-based p io i iza ion o land use a a esolu ion o 250 m
2
, i.e., in a conside ably mo e de ailed scale han equi ed
by cu en ope a ional o es managemen . The unce ain y analysis indica ed ha maps o he ES p o isioning
po en ial should be p epa ed sepa a ely based on expec ed and ex eme ou comes o he ES p oxy models o
ully desc ibe he p oduc ion possibili ies o he landscape unde he unce ain ies in he models.
Keywo ds: Fo es y decision making, Spa ial p io i iza ion, Ligh de ec ion and anging (LiDAR), Remo e sensing
Backg ound
Fo es y decision making equi es e alua ing po en ial
managemen al e na i es wi h espec o mul iple objec-
i es (Kangas e al. 2008). A undamen al decision is e-
la ed o which goods and se ices o p oduce: in
addi ion o con en ional imbe p oduc ion, he manage-
men objec i es may be ela ed o main aining habi a s,
p o iding ec ea ional and aes he ic oppo uni ies, and
ca bon s o age o seques a ion (e.g. Pukkala 2016).
These goods and se ices a e join ly called “mul iple
uses”(Kangas 1992) o , ollowing Cos anza e al. (1997),
Daily e al. (1997) and many o he s, “ecosys em se ices”
o o es . In he ollowing ex , I use ESs o abb e ia e
“Ecosys em Se ices”, e e ing mos essen ially o indi-
ca o s o o es - ela ed ESs ha can be de i ed om
Remo e Sensing (RS) o o he digi al map da a as
indi ec p oxies (And ew e al. 2014). The mapping o
hese p oxies allows spa ial p io i iza ion and o he
spa ially explici analyses o mul iple ESs a a ious
scales (e.g. Sch ö e e al. 2014; Räsänen e al. 2015; Sani
e al. 2016; Roces-Díaz e al. 2017). Acco ding o e iews
(Ma ínez-Ha ms and Bal ane a, 2012; Englund e al.
2017) and a collec ion o case s udies (Ba edo e al.
2015), howe e , such analyses can be expec ed o su e
om he lack o s anda dized e minology, me hodology
and da a. Inc eased a en ion should especially be
Co espondence: ja i. auhkonen@luke. i
Na u al Resou ces Ins i u e Finland (Luke), Bioeconomy and En i onmen
Uni , P.O. Box 68, Yliopis oka u 6, FI-80101 Joensuu, Finland
© The Au ho (s). 2018 Open Access This a icle is dis ibu ed unde he e ms o he C ea i e Commons A ibu ion 4.0
In e na ional License (h p://c ea i ecommons.o g/licenses/by/4.0/), which pe mi s un es ic ed use, dis ibu ion, and
ep oduc ion in any medium, p o ided you gi e app op ia e c edi o he o iginal au ho (s) and he sou ce, p o ide a link o
he C ea i e Commons license, and indica e i changes we e made.
Vauhkonen Fo es Ecosys ems (2018) 5:24
h ps://doi.o g/10.1186/s40663-018-0143-1
ocused on quan i ying and communica ing he esul ing
unce ain ies o he decision make s in o de o make
in o med decisions (see also Eigenb od e al. 2010;
Schulp e al. 2014; Foody 2015). Accoun ing o hese
aspec s, he p esen s udy examines he obus ness o
o es land-use p io i iza ion based on maps o he p o-
isioning po en ial o o es ESs (Vauhkonen and
Ruo salainen 2017a), i.e., he i ness o o es pa ches o
p o ide goods and se ices ypical o he ESs occu ing
in he s udied a ea, e-conside ing he me hodological
and da a wo k low p oposed in he ea lie s udy.
To esul in alid conclusions om RS-based decision
analyses, he es ima es should be accu a e al eady a he
le el o indi idual pixels. The use o ac i e RS such as
Ligh De ec ion and Ranging (LiDAR) is expec ed o
p oduce mo e accu a e in o ma ion compa ed o pas-
si e, op ical RS (Le sky e al. 2001; Coops e al. 2004;
Mal amo e al. 2006), especially, when using small pixels
(e.g., 200 m
2
as in Næsse 2002). Fo es s uc u e and
habi a ela ed in en o ies in pa icula bene i om he
abili y o LiDAR o p o ide h ee-dimensional in o ma-
ion, when ope a ed as Ai bo ne Lase Scanning (ALS;
Mal amo e al. 2014). Kanka e e al. (2015) e alua ed he
es ima ion accu acy o biomass a ibu es based on wo
di e en RS se ups in an a ea closely esembling o ha
p esen ly s udied. Acco ding o hei esul s, pixel-le el
p edic ions based on coa se o medium esolu ion sa el-
li e image y had a Roo Mean Squa ed E o (RMSE) o
47.7% o he o al biomass, which could be educed o
25.7% using ALS and local ield e e ence da a. The ALS
da a used we e acqui ed by he land su ey, and he
a ailabili y o such da a is inc easing due o la ge-a ea
acquisi ions o e ain ele a ion modelling. Such da a
ha e also been used o map a ibu es ela ed o habi a
(Melin e al. 2013,2016; Vauhkonen and Imponen
2016), s uc u al (Valbuena e al. 2016b; Vauhkonen and
Imponen 2016) and aes he ic (Vauhkonen and Ruo salainen
2017b) p ope ies o he o es .
O e all, when a ious o es ESs a e ca ego ized ac-
co ding o a ypology such as he Common In e na ional
Classi ica ion o Ecosys em Se ices (CICES) as in
Englund e al. (2017), he po en ial o ALS o assessing
he sui abili y o o es a eas o p o ide hese ESs can be
cha ac e ized as:
–Regula ion and main enance se ices: A e y high
numbe o s udies indica es ha he ege a ion
heigh and densi y p o iles p oduced by ALS a e
use ul o a de ailed quan i ica ion o a ia ions in
abo e-g ound biomass (Næsse and Gobakken 2008;
Zolkos e al. 2013; Popescu and Hauglin 2014) and,
hus, ca bon s o age (Pa enaude e al. 2004).
Essen ially, ALS p oduces a h ee-dimensional
desc ip ion o he o es s uc u e, which can be
ela ed o ecological p ope ies such as habi a ypes
(Bässle e al. 2011) o biological di e si y in gene al
(Mülle and Vie ling 2014) and employed o assess
sui abili y o o es s o be main ained as habi a s o
di e en species (Da ies and Asne 2014; Hill e al.
2014; Simonson e al. 2014).
–P o isioning se ices: Se e al s udies ca ied ou
especially in bo eal o es s uc u es indica e ALS
da a use ul o assessing p ope ies ela ed o wood
p oduc ion. Excep ha he me hods lis ed in he
p e ious pa ag aphs can be di ec ly used o assess
he p oduc ion po en ial o bulk biomass, also mo e
de ailed p edic ions o imbe asso men s
(Ko honen e al. 2008; Ko amaa e al. 2010;
Vauhkonen e al. 2014; Hou e al. 2016) o wood
ibe - ela ed a ibu es (Hilke e al. 2013; Lu he e
al. 2014) a e possible. Al hough he yield s udies a e
mos ly ela ed o wood-based biomass, he e also a e
examples o imp o ed assessmen s o he yield o
sh ub ui s (Ba be e al. 2016) o edible ungi
(Peu a e al. 2016) based on ALS.
–Cul u al se ices: The applicabili y o ALS highly
depends on he cul u al se ice o in e es . Fo
example, se e al a chaeological s udies indica e he
po en ial o imp o e he mapping o his o ical
emains in he o es using an ALS-based digi al
e ain model. Simila echniques o isualize he
e ain (Domingo-San os e al. 2011) o ees
(Lämås e al. 2015) could po en ially be used o
assess he aes he ic p ope ies o he o es . To da e,
he s udy o Vauhkonen and Ruo salainen (2017b),
which assessed he p e e ences on he isual ameni y
o a o es a ea based on cu ings simula ed o
iangula ed ege a ion poin clouds, appea s o be
he only ALS-based a emp owa ds his di ec ion.
The use o ALS can hus be mo i a ed by he po en ial
o ob ain a be e co espondence wi h o es biophysical
a ibu es and hese da a may be a ailable o some a eas
in a simila ex en as land co e maps and o he publicly
a ailable da a. Despi e he high po en ial, howe e , also
ALS-based in o ma ion may yield a high deg ee o
unce ain ies, i applied in expe models o mula ed
acco ding o con en ionally measu ed ield a ibu es.
Fo example, he sui abili y index p oposed by Pukkala
e al. (2012) o map po en ial habi a s o Sibe ian jay
(Pe iso eus in aus us L.) would equi e es ima ing he
a ailabili y o Vaccinium my illus (L.) be ies and epi-
phy ic lichens o ood and nes s. Al hough sub-models
o es ima e hese a ibu es a e p esen ed (Pukkala e al.
2012), also hose include s and age and si e e ili y,
which a e di icul o es ima e by ALS. Al hough some
esea che s ha e p edic ed e en unde s o ey- ela ed a -
ibu es, he esul s o Ko pela e al. (2012) indica e ha
Vauhkonen Fo es Ecosys ems (2018) 5:24 Page 2 o 19
di ec measu es a e di icul o ob ain due o ansmis-
sion losses occu ing in he uppe canopy (see also
Mal amo e al. 2005) and such es ima ions would be
e en mo e un eliable based on passi e op ical RS
me hods. E en he ecogni ion o dominan ee species
may be challenging in ALS-based in en o ies: despi e
p omising esul s based solely on ALS (Ø ka e al. 2013;
Vauhkonen e al. 2014), he esul s o Rä y e al. (2016)
sugges di icul ies in de ec ing species, which domina e
a mino p opo ion o an a ea o he wise homogeneous
in e ms o he species.
On he o he hand, ALS may allow p oducing o he
a ibu es wi h mo e ele ance om he o es manage-
men poin o iew. Fo example, o es s wi h mul i-
laye ed e ical s uc u e can be dis inguished based on
he da a (Zimble e al. 2003; Mal amo e al. 2005),
which can be u he employed in de ec ing he p e ail-
ing sil icul u al sys em (Bo alico e al. 2014), manage-
men in ensi y (S e d up-Thygeson e al. 2016;
Valbuena e al. 2016a), o de elopmen s age (Valbuena
e al. 2016b). E en mo e de ailed indices may be de el-
oped based on ecological a ionale (Lis opad e al.
2015) o a ho ough unde s anding o he p ope ies a -
ec ing he ALS esponse (Valbuena e al. 2013,2014).
Ea lie s udies ha e sugges ed ha he in o ma ion in
heALSda amaybecondensed oa ewme ics(Kane
e al. 2010;Lei e e e al.2015; Valbuena e al. 2017),
he pa i ioning o which will p o ide a s a i ica ion
co esponding closely o he s uc u al complexi y ob-
se ed in he ield (Pascual e al. 2008;Thompsone al.
2016; Vauhkonen and Imponen 2016).
E en hough p ope ies ela ed o indi idual ESs ha e
been ac i ely s udied, no s udies ha show how o sup-
po managemen decisions ela ed o he p o isioning
o mul iple o es ESs based on h ee-dimensional o es
s uc u e desc ip ion ob ained by ALS can cu en ly be
ound om he li e a u e. Ba bosa and Asne (2017) and
Rechs eine e al. (2017) de i ed in o ma ion om ALS
da a o p io i ize landscapes o ecological es o a ion
and species conse a ion planning, espec i ely. Packa-
lén e al. (2011) used ALS da a and spa ial op imiza ion
o de i e so called dynamic ea men uni s o guide he
managemen o pulpwood p oduc ion in a plan a ion
o es . Al hough a simila app oach could be ex ended
o he decision making o o he o mul iple ESs (Pukkala
e al. 2014), all ALS-based applica ions a e, o da e, o-
cused on single ESs.
The pu pose o his s udy is o es ALS da a o man-
agemen p io i iza ion o mul iple ESs in a bo eal o es
landscape. P oxies o pixel-wise p o isioning po en ial
o biodi e si y, ca bon, imbe , be ies, and ec ea ional
ameni ies we e o mula ed using ALS-based ea u es
and compa ed o in o ma ion ob ained om o es e-
sou ce maps wi h a esolu ion o 16 × 16 m
2
.The
quali y o land use p io i iza ion based on he ob ained
in o ma ion was e alua ed unde wo di e en decision
suppo models: ei he using he de eloped models de-
e minis ically o in co po a ion wi h he unce ain ies
o he models.
Me hods
A me hodological o e iew
Speci ically, he ALS da a a e es ed o p edic ing he
p o isioning po en ial o ESs (Vauhkonen and
Ruo salainen 2017a) in a spa ial p io i iza ion ame-
wo k, whe e land use decisions a e based on anking he
se o decision al e na i es in he conside ed loca ion(s)
and choosing he bes acco ding o he decision make s’
p e e ences (c ., Malczewski and Rinne 2015). When
applied o p io i ize o es s o single (e.g. Leh omäki e
al. 2015) o mul iple uses (e.g. Vauhkonen and
Ruo salainen 2017a) based on ES p oxy maps, a simpli-
ied wo k low o such analyses includes h ee me hodo-
logical s eps:
1) Da a acquisi ion, ea u e ex ac ion and/o expe
modelling o de i e p oxy alues o he analyzed
ESs.
2) Scaling and no maliza ion o he p oxy alues
de i ed om di e en sou ces o he same scale.
The esul ing alues can be called ‘p io i y’,‘bene i ’,
o ‘u ili y’ alue and used in di e en ways
depending on he li e a u e sou ce (see also Pukkala
2008; Pukkala e al. 2014; Malczewski and
Rinne 2015).
3) Decision analyses using he no malized da a a
selec ed spa ial scale(s).
Because he no malized p oxy maps esul ing om he
p e ious s eps ‘measu e’ he ESs in a same scale and ac-
coun o he alue ange o each ES in he en i e land-
scape, hey can be used (a) o mu ually ank ESs wi hin
a spa ial uni o subsequen ly p io i ize managemen o
p o ide mos sui able ESs in each uni ; and (b) o iden-
i y he mos impo an loca ions o speci ic ESs in he
landscape o be conside ed as managemen ho -spo s o
cold-spo s. Because he spa ial p io i iza ion is ca ied
ou a a sub-s and-le el using pixels o o he co e-
sponding map uni s, i is expec ed o allow a mo e e i-
cien use o he p oduc ion possibili ies o he o es
(Heinonen e al. 2007) and, o e all, ope a ionalize he
concep o ESs o landscape planning, which is u he
mo i a ed by de G oo e al. (2010).
The p esen s udy examines whe he changes o
each o he h ee s eps lis ed abo e could imp o e
pixel-wise analyses o he p o isioning po en ial o
o es ESs (c . he discussion sec ion o Vauhkonen
and Ruo salainen 2017a):
Vauhkonen Fo es Ecosys ems (2018) 5:24 Page 3 o 19
1) Wha da a o use o he expe models o he p o-
isioning po en ial: A consolida ed app oach o ob ain
g id-based, wall- o-wall p edic ions o he essella ed
landscapes would be o use o es esou ce maps based
on gene alizing ield sample plo measu emen s o la ge
a eas using coa se o medium esolu ion RS images and
o he nume ic map da a (Tomppo e al. 2008a,2008b,
2014). This app oach, e e ed o as Mul i-Sou ce
Na ional Fo es In en o y (MS-NFI), was used by
Vauhkonen and Ruo salainen (2017a). E en i ALS al-
lows mo e p edic ion possibili ies, as e iewed abo e, i
is p ac ically easoned o benchma k he accu acies
agains he pixel da a p o ided by he MS-NFI
app oach, because di e en o es esou ce maps a e
eadily a ailable in many coun ies (Tomppo e al.
2008b, Roces-Díaz e al. 2017; Vauhkonen and
Ruo salainen, 2017a).
2) How o scale he ESs o iginally measu ed in
di e en uni s o he join analyses: Vauhkonen and
Ruo salainen (2017a) used a simple no maliza ion o
con e he ES alues be ween 0 and 1:
ij ¼nij
N;ð1Þ
whe e
ij
is he no malized alue and n
ij
is he posi ion
o he j: h plo in ascending o de o he expe model
alues o he i: h ecosys em se ice among al oge he N
plo s. No ably, his no maliza ion p oduced alues in an
in e al scale, whe eas he a ios be ween he expe
model alues could also be assumed use ul o he p io -
i y anking. An al e na i e, a io-scale no maliza ion
could be compu ed as:
ij ¼ESij−min ESi
ðÞ
max ESi
ðÞ−min ESi
ðÞ
;ð2Þ
whe e
ij
is he alue (o p io i y o bene i o u ili y, de-
pending on li e a u e sou ce; see abo e) p oduced by he
i: h ES in plo j.
3) How o use he ob ained in o ma ion in decision
analyses: Vauhkonen and Ruo salainen (2017a) de e -
minis ically p io i ized each pixel o he ES wi h he
highes p edic ed p oxy alue, bu highligh ed he need
o conside unce ain ies a ound he p edic ions. I a
quan i ica ion o he unce ain ies is ob ained (e.g., by
app oxima ing esidual e o s o calib a ion models i -
ed o he da a), he decision analyses can conside dis-
ibu ions o unce ain y in addi ion o he expec ed
alues and p oduce sepa a e ecommenda ions o di -
e en decision make s acco ding o hei a i udes o-
wa ds isk (Pukkala and Kangas 1996). The e o e, in
addi ion o de e minis ic use o he p edic ed alues, his
s udy conside ed bo h he expec ed and ex eme ou -
comes o he p edic ions when selec ing he mos
sui able ES o a pixel. The p incipal idea o his analysis
is illus a ed in Fig. 1.
On his backg ound, he p esen s udy es ed he da a
sou ce (ALS o MS-NFI), p io i y alue unc ion o m
(Eqs. 1o 2), unce ain y managemen app oach, and
join implica ions o hese choices o he p edic ions o
he p o isioning po en ial o o es ESs and subsequen
managemen p io i iza ion decisions. Fo es ESs consid-
e ed we e selec ed based on wo c i e ia: likelihood o
occu in he s udied landscape and exis ence o expe
models o de i e p oxies o hei p o isioning po en ial
based on he ield measu emen s (Table 1). The ield
and MS-NFI da a con ained es ima es o o es a ibu e
ha could be di ec ly inse ed o he expe models.
Using ALS da a, eg ession analyses we e employed o
es ima e p edic i e ela ionships be ween ALS- ea u es
and ES p oxy alues o ully u ilize he di e en p ope -
ies o hese da a (c ., Sec ion “ALS-based models o he
p io i y alues o he ESs”below). In he absence o inde-
penden , wall- o-wall da a o alida ion, bo h he p e-
dic ions and alida ions we e ca ied ou a he le el o
indi idual o es plo s. The e alua ion is he e o e lim-
i ed o he local i ness o he ESs o a speci ic o es
pa ch in a single poin in ime and wi hou conside ing
hei spa ial o empo al con inuum. No decision make
was assumed in his s udy and he alues ob ained om
Fig. 1 A gene ic example o selec ing he bes decision al e na i e
based on di e en ou comes o model p edic ions (colo ed cu es).
The yellow cu e yields he highes p io i y alue based on he
expec ed (uppe ho izon al line) o wo s ou come o he model.
Howe e , i he decision make weigh s bes possible ou comes, he
al e na i e depic ed by he g ey cu e should be selec ed as i
p oduces he highes p io i y in he igh ail accumula ion poin ( he
in e cep ion o he cu e and he lowe ho izon al line)
Vauhkonen Fo es Ecosys ems (2018) 5:24 Page 4 o 19
bo h he Eqs. 1and 2we e he e o e ea ed wi h equal
weigh s, e en i hose could addi ionally be weigh ed ac-
co ding o he decision make s’p e e ence s uc u e.
S udy a ea and expe imen al da a
The s udy a ea is loca ed in E o, Finland (61.19°N,
25.11°E), which belongs o he sou he n bo eal o es
zone. The da a ex ended o e an a ea o app oxima ely
3 km × 6 km. The o es s ands in he a ea a y om in-
ensi ely managed o na u al o es s in e ms o hei
sil icul u al s a us. App oxima ely 84% o he g owing
s ock in he s udied plo s is domina ed by coni e ous
ee species Sco s pine (Pinus syl es is L.) and No way
sp uce (Picea abies [L.] H. Ka s .). Deciduous ee spe-
cies such as bi ches (Be ula spp. L.), aspen (Populus e-
mula L.), alde s (Alnus spp. P. Mill.), willows (Salix spp.
L.), and owan (So bus aucupa ia L.) occu in mixed
s ands and below he dominan canopy.
Da a se s used we e compiled om h ee ea lie s ud-
ies in he same a ea (Vauhkonen and Imponen 2016;
Niemi and Vauhkonen 2016; Vauhkonen and Ruo salai-
nen 2017a). Vauhkonen and Imponen (2016) down-
loaded and p ocessed ALS da a acqui ed by he Na ional
Land Su ey o Finland o s a i y he a ea acco ding o
o es s uc u al p ope ies. The ALS da a we e acqui ed
om a lying al i ude o 2200 m using Leica ALS 50
scanne on 7 May, 2012, o yield a nominal pulse densi y
o 0.8 m
−2
. Ci cula sample plo s (9 m adius) we e
placed by clus e ing he ALS da a wi h espec o o es
s uc u al ea u es, which was ound o be an e icien
s a egy o dis ibu e he sample ac oss he spa ial, size,
and age dis ibu ions o he ee s ock (Vauhkonen and
Imponen 2016). The ield measu emen s we e ca ied
ou in June–Augus , 2014. The species and
diame e -a -b eas heigh (DBH) we e measu ed o each
ee wi h a DBH ≥5 cm. Fo each ee species o he
plo , a ee wi h a DBH co esponding o he median
ee was measu ed o heigh and used o calib a e
heigh cu es o p edic ing he missing ee heigh s.
Plo -le el o es a ibu es we e compu ed om he
ee-le el measu emen s using s anda d equa ions and
me hods, which a e desc ibed in de ail in an open-access
a icle by Niemi and Vauhkonen (2016).
Publicly a ailable MS-NFI da a (Na u al Resou ces In-
s i u e Finland 2017) we e included o p o ide a bench-
ma k o he ALS da a. The MS-NFI maps a e he same
used Vauhkonen and Ruo salainen (2017a) and de ails
on hei p e-p ocessing a e gi en in ha pape . These
as e maps depic ed si e e ili y, g owing s ock olume
and biomass componen s by ee species, o al basal a ea
and mean diame e and heigh co esponding o hose
o he (basal a ea weigh ed) median ee, and hey we e
p oduced using a k-nea es neighbo (k-NN) es ima ion
me hod based on op imized neighbo and ea u e selec-
ion (Tomppo and Halme 2004; Tomppo e al. 2008a,
2014). The me hod used a ious sa elli e images om
2012 o 2014 and Na ional Fo es In en o y (NFI) ield
plo measu emen s om 2009 o 2013, which we e up-
da ed o co espond he si ua ion in mid-2013 using
g ow h models.
Al oge he 102 ield plo s we e co e ed by bo h ALS
and MS-NFI da a and we e included in he analyses.
Table 1 The ESs conside ed in his s udy and expe models o de i ing hei e e ence p oxy alues
Abb . ES Indica o , uni (ci a ion)
a
S and-le el o es a ibu es used as p edic o s
b
BIOD Biodi e si y Index alue based on expe opinion
(Leh omäki e al. 2015)
1
Si e e ili y, g owing s ock olume, diame e ,
dominan species
TIMB Timbe p oduc ion Soil expec a ion alue (SEV), €∙ha
−1
(Pukkala 2005)
2
Diame e , basal a ea, age, si e e ili y, species-speci ic
g owing s ock olume, numbe o ees, ope a ional
en i onmen ( empe a u e, in e es a e, imbe p ices)
CARB Ca bon s o age Es ima ed amoun o ca bon
3
, ∙ha
−1
(Ka jalainen and Kellomäki 1996)
G owing s ock olume
BILB Sui abili y o bilbe y picking Index alue based on expe opinion
(Ihalainen e al. 2002)
Age, basal a ea, heigh , species-speci ic g owing
s ock olume, si e e ili y
COWB Sui abili y o cowbe y picking Index alue based on expe opinion
(Ihalainen e al. 2002)
Age, species-speci ic g owing s ock olume, diame e ,
si e e ili y
AMEN Visual ameni y Index alue based on expe opinion
(Pukkala e al. 1988)
Diame e , numbe o ees, species-speci ic g owing
s ock olume, si e e ili y
a
When compu ing he alues o he p esen s udy, he ollowing de ails o excep ions compa ed o o iginal publica ions we e made:
1
The index alues a e o o m diame e × olume, scaled using dominan -species-speci ic ans o ma ion unc ions (Leh omäki e al. 2015) and maximum alues o
o es a ibu es in he s udy a ea, and mul iplied by si e e ili y speci ic weigh s (Leh omäki e al. 2015).
2
Values o ope a ional en i onmen ela ed pa ame e s we e ob ained as combina ions o e ec i e empe a u e sum ixed o 1300 deg ee days, in e es a es o
1%–4% and saw-wood/pulpwood p ices (uni s in €∙m
−3
) o 30/15, 30/25, 40/15, 40/25, 40/35, 50/25, and 50/35, and he SEV was ob ained as an a e age o hese
28 combina ions weigh ed by he p opo ions o species. All alues we e adop ed om he s udy by Pukkala (2005).
3
The es ima ed ca bon was ob ained based on con e sion ac o s om species-speci ic, o al s em olumes o ca bon con en s.
b
To s anda dize he compu a ion based on all da a se s, he ollowing simpli ica ions o g oupings we e used:
-Species g oups: pine, sp uce, deciduous ees.
-‘Diame e ’always e e ed o he basal-a ea weigh ed mean diame e .
Vauhkonen Fo es Ecosys ems (2018) 5:24 Page 5 o 19

The models o Table 1we e applied o p oduce
plo -speci ic e e ence alues o he p o isioning po en-
ial o he ESs based on ield da a. Acco ding o an ex-
plo a o y analysis, he expe models o Table 1, i wi h
many di e en da a se s, had conside ably di e en alue
anges o e he landscape. As a esul , a di ec
no maliza ion o he expe unc ion alues speci ically
wi h Eq. 2 esul ed o emphasizing one ES in he p io i y
ankings only because o he di e en shape and scale o
ini ial alue dis ibu ions, as elabo a ed upon in
Appendix 1. Fo his eason, he expe unc ion alues
o all ESs we e ans o med o ollow he no mal
dis ibu ion as closely as possible using he
Box-Cox- ans o ma ion (Appendix 1) p io o applying
Eqs. 1and 2. The o es a ibu e es ima es based on he
MS-NFI maps we e ans o med using he same pa am-
e e alues as wi h ield da a. This ans o ma ion did
no a ec he o de o he obse a ions, bu p oduced
app oxima ely equally shaped equency dis ibu ions o
e e y ES, as de ailed in Appendix 1.
ALS-based models o he p io i y alues o he ESs
P edic ion models wi h independen a iables ex-
ac ed om he ALS da a we e o mula ed o p edic
p io i y unc ion alues o he o m o Eq. 2.P io i y
unc ion alues co esponding o Eq. 1we e ob ained
by o de ing he a o emen ioned p edic ions, i.e., no
sepa a e models we e cons uc ed o he unc ion
o m o Eq. 2.
As easoned in he In oduc ion, he aim was no o
model he o es a ibu es used as he p edic o s o he
expe models, bu o iden i y and quan i y such p ope -
ies o he ALS poin clouds ha di ec ly explained he
a ia ion in he ES p oxies. As isualized in Fig. 2, he
poin clouds o he plo s wi h maximum p oxy alues
did no conside ably di e be ween he ESs in e ms o
he o al dis ibu ions. Howe e , when heigh alues o
p opo ions we e compu ed sepa a ely acco ding o echo
ca ego ies, ES-speci ic di e ences could be poin ed ou
(Fig. 2). The ea u es we e he e o e ex ac ed in echo
ca ego ies, which we e “only echoes”(su ix _only), “ i s
o many echoes”(_ i s ), “las o many echoes”(_las ),
“ i s echoes”(_FP), and “las echoes”(_LP), whe e he
las wo ca ego ies included “ i s o many”and “las o
many”echoes, espec i ely, wi h “only echoes”dupli-
ca ed in bo h. Fixed heigh alues o 0.5 m, 5 m, and
below o abo e an adap i e heigh alue de e mined as
he heigh o he 60 h pe cen ile we e used as he
Fig. 2 ALS heigh p o iles and desc ip i e cha ac e is ics o he ield plo s conside ed o be mos impo an loca ions o he ESs in he da a
s udied (p io i y alue o 1 based on Eq. 2). Fo compa ison, he lowe igh panel shows a plo ha had low p io i y alues o he conside ed ESs.
The black, g een, and blue symbols indica e only, i s -o -many, and las -o -many ALS echoes, espec i ely. G ey ho izon al lines indica e he mean
heigh s o hese echo ca ego ies and all echoes and a e d awn o illus a e he di e ences in e ms o hese me ics be ween he ESs
Vauhkonen Fo es Ecosys ems (2018) 5:24 Page 6 o 19
h esholds o g ound, sh ub, and supp essed o domin-
an canopy, espec i ely.
The ollowing ca ego ies o he ea u es we e
conside ed:
–Canopy heigh and densi y, which a e he basic
p edic o s used in ALS analyses (Næsse 2002) and
we e assumed o disc imina e be ween size-speci ic
a ibu es o he ESs: he maximum (hmax), he
mean (hmean), and he s anda d de ia ion (hs d) o
he heigh alues abo e he g ound h eshold; he
5 h, 10 h, 20 h, ..., 90 h, and 95 h pe cen iles
(hzz, whe e zz deno ed he pe cen ile alue); and
he co esponding p opo ional densi ies (dzz) we e
compu ed acco ding o Ko honen e al. (2008, pp.
502–503).
–P opo ion o echoes abo e a gi en h eshold o all
echoes, co esponding o a ege a ion co e es ima e
(Ko honen e al. 2011). This p opo ion was compu ed
in wo ways: using echoes o di e en ca ego ies abo e
he g ound (cc
X_g ound
,whe eXis he echo ca ego y)
o i s echoes abo e he sh ub laye h eshold (cc
sh ub
),
which co esponds o an a emp o quan i y he sh ub
laye hickness (c ., Vauhkonen and Imponen 2016).
–Absolu e di e ences be ween mean heigh s o
di e en echo ca ego ies. These ea u es we e
compu ed wi hou heigh h esholds and assumed o
disc imina e be ween p ope ies ela ed o
coni e ous- o deciduous-domina ed o es in he
ALS da a acqui ed du ing he lea -o pe iod
(Liang e al. 2007). These ea u es a e deno ed by
di
x–y
, whe e su ix x–y e e s o he heigh
di e ence o echo ca ego ies FP–LP, only–LP,
i s –only, o i s –las .
–P opo ions o he di e en echo ca ego ies, which
we e assumed o be a ec ed by he species and size
speci ic ES p ope ies in he canopy simila o he
ALS-in ensi y ea u es (Ø ka e al. 2012; Vauhkonen
e al. 2014). These ea u es a e deno ed by p op
X/Y_z
,
whe e X/Y indica ed he a io o wo echo ca ego ies
Xand Y, and zwas he heigh h eshold employed
o compu a ions.
–P edic o s ela ed o he sh ub and unde s o ey
laye s (Vauhkonen and Imponen 2016): he a io o
he echoes e lec ed abo e g ound bu below he
dominan canopy h eshold (
unde s o y
); he s anda d
de ia ion o he heigh alues o echoes e lec ed
abo e g ound bu below he dominan canopy
h eshold (s d
unde s o y
); and he a io o he echoes
e lec ed om he sh ub laye o all echoes (
sh ub
).
Fea u es x
i
,i=1, 2, …, 143, lis ed abo e o med he
ini ial se S
1
o candida e p edic o s. To accoun o use-
ul in e ac ions be ween he ea u es, he inal se Swas
ob ained as S
1
∪{x
i
×x
j
}∀i,j∈S
1
, which esul ed o
al oge he 10,296 candida e ea u es pe plo . Sepa a e
models o each ES we e cons uc ed by inse ing ea-
u es i e a i ely in o a model empla e:
^
yin ¼anþXN
n¼1bnxcn
jn;ð3Þ
whe e ^
yin is he ec o o p edic ed p io i y alues o
he i: h ES, x
jn
is he j: h ea u e o S, and a
n
,b
n
, and c
n
a e model pa ame e s a he n: h ound o N=1,2,3,4
i e a ion ounds. Pa ame e s a
n
,b
n
, and c
n
we e es i-
ma ed using he nls unc ion o R s a is ical compu ing
en i onmen (R Co e Team 2016). Tes ing e e y candi-
da e ea u e as x
jn
a e e y i e a ion ound, he RMSE
be ween he p edic ed and e e ence p io i y alues was
compu ed as:
RMSE ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Pn
i¼1^
yi−yi
ðÞ
2
n;
sð4Þ
whe e nis he numbe o obse a ions, and ^
yiand y
i
a e
he p edic ed and e e ence alues, espec i ely. The ea-
u e ha minimized he RMSE was e ained in he
model empla e and he i e a ions we e con inued un il
he model included a maximum o ou ea u es.
Howe e , mo e c i e ia we e employed o selec he
model o be used o he p io i iza ion analyses among
he models wi h one o ou ea u es:
1) The inal p edic o inse ed had o imp o e he
RMSE by a leas 1%.
2) The esidual e o s had o sa is y he null
hypo hesis ha he conside ed sample came om a
no mally dis ibu ed popula ion, which was
examined g aphically using sca e , esidual and
QQ-plo s, and nume ically using he es s a is ic
p oposed by Shapi o and Wilk (1965).
3) The model had o pass a “sensi i i y o
con e gence” es , in which he model was i
sepa a ely o each plo using Lea e-One-Ou -
C oss-Valida ion (LOOCV), i.e., no allowing he
plo in ques ion o be a ailable in he aining da a
o model i ing. Implica ions o including his es
a e u he desc ibed in he Resul s sec ion.
P edic ing he p io i y alues o he ESs based on he
MS-NFI maps
Benchma k p edic ions o hose based on ALS we e ob-
ained by inse ing he o es a ibu e es ima es om
he MS-NFI maps o Eq. 2. P io i y unc ion alues co -
esponding o Eq. 1we e ob ained by o de ing he a o e-
men ioned p edic ions (c ., p e ious sec ion). The
MS-NFI maps included es ima es o all o he independ-
en a iables excep he numbe o ees pe hec a e,
Vauhkonen Fo es Ecosys ems (2018) 5:24 Page 7 o 19
which was es ima ed by di iding he o al basal a ea by
he basal a ea co esponding o he mean diame e , i.e.,
assuming ha he esul ing numbe o a e age-sized
ees exis ed in a pixel. To compu e plo -wise es ima es,
he pixels o he o es esou ce maps in e sec ing wi h
he plo polygons we e iden i ied using a spa ial que y.
The es ima es o a plo we e ob ained om he in e -
sec ing pixels as weigh ed a e ages wi h he join a eas
o he plo s and pixels as he weigh s. Finally, o see i
amending he models based on ALS wi h he MS-NFI
laye s imp o ed he models, a simila ea u e selec ion
as wi h ALS da a was un including all MS-NFI-based
ES and o es a ibu e p oxies as addi ional ea u e
candida es.
Field calib a ion and e alua ion o he p edic ions
Following he me hod desc ibed in he p e ious sec ion,
po en ial es ima ion e o s in he MS-NFI maps p opa-
ga e o he p edic ed p io i y alues, whe eas simila
e o p opaga ion is a oided in he ALS-based analyses
due o local model i ing. An addi ional calib a ion s ep
was he e o e included o elimina e he con ibu ion o
he local ield sample o he p edic ions. Calib a ion
models yi¼ ð^
yiÞ, whe e y
i
was he e e ence p io i y
alue o he i: h ES and ^
yii s RS-based es ima e, o all
ESs we e i simul aneously as sys ems o linea equa-
ions. Due o he high in e -co ela ions (see Addi ional
ile 1, Table S1), he models we e i in wo s eps: i s ,
using O dina y Leas Squa es (OLS) o p oduce model
esiduals, and second, using Seemingly Un ela ed Re-
g ession (SUR) o accoun o he esidual e o co a i-
ance ma ices in he inal models. The compu a ions
we e ca ied ou in he LOOCV mode using he sys em-
i package o R (Henningsen and Hamann 2007). The
accu acies o he ALS- and MS-NFI-based p edic ions
we e compa ed using he RMSE and coe icien o de e -
mina ion (R
2
) compu ed be ween he e e ence alues
and p edic ions ob ained om he LOOCV models.
Decision analyses
The e ec s o he a o emen ioned p edic ion accu -
acies o he managemen decisions we e e alua ed by
compa ing he p io i y anking o he ESs in each in-
di idual plo . The ES wi h he highes p io i y alue,
based on Eqs. 1o 2applied o he ield e e ence
da a, was assumed o be he mos sui able ES o he
speci ic plo . The RS-based decision was conside ed
co ec , i he mos sui able ES based on he ield
da a and he RS p edic ion equaled. The deg ee o in-
co ec decisions was quan i ied using wo ap-
p oaches. Fi s , he co ec ness o e e y decision was
gi en a nume ical sco e (Gopal and Woodcock 1994):
si ua ions whe e RS and ield da a esul ed in he
same decision was gi en a sco e o 6; hose whe e
he RS-based se ice was he second bes acco ding
o he ield da a a sco e o 5; and so on, un il he
si ua ion whe e he RS-based se ice was he wo s
acco ding o ield da a, which was gi en a sco e o 1.
The dis ibu ions o hese “decision sco es”we e
compa ed be ween he di e en da a sou ces. Second,
he dispe sion in ield and RS-da a be ween he se -
ices selec ed as he mos sui able o he speci ic
plo was examined using con usion ma ices. The
p io i y anking o he less impo an ESs was no
e alua ed.
In addi ion o ‘de e minis ic’decision making de-
sc ibed abo e, he sensi i i y o he decisions was exam-
ined by inco po a ing he unce ain ies o he models o
he analyses (Fig. 1). Ins ead o using he expec ed alues
o he p io i y unc ions, he anking was ca ied ou as-
suming he p edic ions as ealized alues o a andom
a iable X~N(E,s
2
), whe e Ewas he expec ed alue
and s
2
was he mean squa ed e o o he model esid-
uals. A simila p io i y anking as wi h he expec ed
alues was ca ied ou wi h p edic ions ha we e among
he wo s and bes ou comes o he model, ob ained as
he alues o he 5 h and 95 h pe cen iles o he dis ibu-
ion o Xo each ES.
Resul s
ALS-based models o he p io i y alues o he ESs
The ALS ea u es conside ed as he p edic o s o he
eg ession models a e lis ed in Table 2. All ea u e
and echo ype ca ego ies and a wide ange o di e -
en heigh alues was employed when building he
models, o which eason only a ew speci ic obse a-
ions on he s uc u e o models can be made. All
ea u es selec ed we e p oduc s o o m ea u e
1
× ea-
u e
2
,whe e ea u e
1
was o en an absolu e heigh
alue (a mean heigh , pe cen ile, o heigh di e ence)
and ea u e
2
a p opo ion (ei he a canopy co e
p oxy o p opo ional densi y). This combina ion was
especially equen among he i s ea u es selec ed
o he models. In he models o TIMB, all selec ed
p edic o s we e such combina ions employing a ious
heigh alues and echo ca ego ies. The models o
BIOD and CARB used p opo ion ×p opo ion ypes
o in e ac ions and he a io o i s -o -many o only
and i s e u ns (p op
i s /FP_g ound
). The p edic o s o
BIOD (e.g., p op
i s /FP_g ound
;di
only–LP
;hs d
LP
)we e
mos di e se in e ms o desc ibing he canopy s uc-
u e wi h ea u es om di e en ca ego ies. The
models o BILB, COWB, and AMEN di e ed om
hose men ioned abo e in employing low pe cen ile
alues, las pulse p opo ions and ea u es such as
unde s o y
, di
only–FP
,p op
i s /FP_g ound
,andhs d
i s
.
Vauhkonen Fo es Ecosys ems (2018) 5:24 Page 8 o 19
O e all, he canopy co e p oxies we e he mos e-
quen ea u e ype, whe eas he compu ing heigh s
and echo ca ego ies o all ea u es a ied. Al hough a
wide ange o di e en heigh alues was used, a p e-
dic o wi h a pe cen ile alue abo e 70 was selec ed
only once.
The g aphical assessmen o model esiduals (de-
ailed esul s no shown) was mainly in line wi h he
es on esidual no mali y (Table 2): he QQ-plo s
showed hea y- ailed esiduals especially o he
models wi h s a is ically signi ican alues o he
Shapi o-Wilk es s a is ic. Howe e , he de ia ions o
no mali y we e ypically ela ed o one o wo plo s
wi h he highes o lowes alues, and no conside ed
p oblema ic o he u he analyses. When examined
in he same da a used o cons uc ing he models,
he pe o mance o e e y model could be sligh ly im-
p o ed by inc easing he numbe o p edic o s o he
maximum numbe allowed. Howe e , when he
models we e e- i using LOOCV, model pa ame e s
could no be sol ed o a leas one o he plo s in
he da a, esul ing o NA alues o his pe o mance
ac o in Table 2. Al hough his e ec could p obably
ha e been a oided by allowing a sligh ly wide ange
o ini ial pa ame e s when i ing he models, i was
also conside ed as a sensi i i y issue e lec ing an
o e -pa ame e iza ion o he ini ial model o ce ain
ypes o o es s uc u es.
The olume o deciduous ees and he MS-NFI
based p oxy o BILB would ha e eplaced he las
ALS-based ea u es in he models o CARB and BILB,
espec i ely, and in he models o COWB, he co e-
sponding MS-NFI p oxy would ha e been selec ed as
he second ea u e. Howe e , none o he a o emen-
ioned MS-NFI ea u es pe o med be e han he
ALS- ea u es o hese models in e ms o he ea u e se-
lec ion c i e ia. Based on he conside a ions abo e, he
ALS and MS-NFI da a se s we e always used sepa a ely.
Also, a di e en numbe o p edic o s was used in he
ALS-based models o he p io i y anking: BIOD and
COWB we e modeled using only one p edic o ( he
one selec ed i s ); TIMB, BILB, and AMEN using wo
p edic o s ( hose selec ed i s and second); and CARB
using h ee p edic o s selec ed i s .
Compa ison o ALS and MS-NFI o p edic ing he p io i y
alues o he ESs
The models based on ALS da a always ou pe o med
hose based on o es a ibu e es ima es de i ed om
he MS-NFI maps. As shown in Figs. 3and 4, he
ALS-based models gene ally explained mo e a ia ion
in he ES p oxies. The eg ession lines o he MS-NFI
da a based on he SUR calib a ion models also di e ed
mo e se e ely om he 0–1-lines. Using MS-NFI da a,
TIMB was p edic ed mos accu a ely wi h an RMSE o
30.4%. The RMSEs o o he ESs we e also close
(30.6%–33.2%), excep AMEN, which had an RMSE o
40.5% and BIOD, which was p edic ed wo s wi h an
RMSE o 41.6%. Using ALS, he ES p edic ed wo s
(COWB) had an RMSE o 29.8%, which is 97% o he
RMSE o he co esponding MS-NFI p edic ion. The
RMSEs o all o he ESs we e in o de o 21.7%–27.5%
(57%–83% o he RMSEs o MS-NFI p edic ions), ex-
cep CARB, which was p edic ed mos accu a ely wi h
an RMSE o 15.1% (47% o he RMSE o MS-NFI p e-
dic ion). The deg ee o de e mina ion o CARB also im-
p o ed mos due o using ALS ins ead o MS-NFI,
om R
2
= 0.11 o 0.81. The R
2
-imp o emen s o he
o he ESs we e close o his magni ude, excep o BILB
and COWB, which had R
2
alues close o each o he
based on bo h he da a sou ces. The esidual e o s o
models based on ALS and MS-NFI we e somewha co -
ela ed o BILB and COWB, bu no o he o he ESs
(Figs. 3and 4, igh column).
Table 2 The ea u es and pe o mance o ALS-based models o
p edic ing a io-scaled ES p oxy alues. W –Shapi o-Wilk es s a is ic
ES P edic o W
a
RMSE RMSE
LOOCV
BIOD cc
sh ub
× h40
i s
0.965
***
0.259 0.266
+ p op
i s /FP_g ound
× d50
i s
0.981 0.235 NA
+ di
only–LP
× hs d
LP
0.986 0.217 0.226
+ h95
i s
× h10
LP
0.977
*
0.203 NA
TIMB cc
only_g ound
× hmean
FP
0.977
*
0.220 0.228
+ h40
las
×cc
LP_g ound
0.970
**
0.202 0.213
+ h05
las
× d05
i s
0.989 0.182 NA
+cc
only_g ound
× h10
FP
0.988 0.174 NA
CARB cc
sh ub
× h60
i s
0.980 0.158 0.163
+ h20
las
×cc
only_g ound
0.988 0.144 0.152
+ d60
i s
× d30
LP
0.987 0.138 0.148
+ p op
i s /FP_g ound
× d05
i s
0.984 0.132 NA
BILB h60
i s
× h70
LP
0.987 0.279 0.286
+ d50
only
×cc
FP_g ound
0.984 0.255 0.274
+
unde s o y
× d05
FP
0.982 0.238 NA
+ d70
i s
× d50
only
0.984 0.222 0.267
COWB d20
LP
× h40
FP
0.966
***
0.281 0.295
+ di
only–FP 2
0.981 0.250 NA
+ p op
i s /FP_g ound
× d05
i s
0.983 0.233 NA
+ d60
i s
× h05
only
0.971
**
0.217 NA
AMEN h10
i s
× hmean
FP
0.993 0.239 0.246
+ d30
las
× d70
i s
0.991 0.219 0.229
+ h20
las
×cc
FP_g ound
0.994 0.204 NA
+ hs d
i s
× hmean
LP
0.991 0.187 NA
a
The as e isks e e o he signi icance o he es s a is ic a he 90% (*), 95%
(**), and 99% (***) con idence le el
Vauhkonen Fo es Ecosys ems (2018) 5:24 Page 9 o 19
Appendix 2
Con usion ma ices o he mos impo an ESs based on
expec ed ou comes
Appendix 3
Con usion ma ices o he mos impo an ESs based on
ex eme ou comes
Table 3 Con usion be ween ESs conside ed as mos impo an
based on he ield da a (obse ed) and MS-NFI maps (p edic ed)
using he expec ed alues o he p edic ed ES p oxies
P edic ed
BILB COWB AMEN BIOD CARB TIMB
Obse ed BILB 8 5 2 0 2 4
COWB 6 15 9 1 1 2
AMEN 1 1 0 0 0 0
BIOD 1 0 4 4 5 2
CARB 0 1 4 1 1 1
TIMB 3 4 2 1 3 8
Table 4 Con usion be ween ESs conside ed as mos impo an
based on he ield da a (obse ed) and MS-NFI da a calib a ed
wi h he local ield sample (p edic ed) using he expec ed alues
o he p edic ed ES p oxies
P edic ed
BILB COWB AMEN BIOD CARB TIMB
Obse ed BILB 11 7 0 0 1 2
COWB 9 18 0 2 2 3
AMEN 1 1 0 0 0 0
BIOD 10 0 0 5 1 0
CARB 1 2 0 1 3 1
TIMB 3 4 0 1 3 10
Table 5 Con usion be ween ESs conside ed as mos impo an
based on he ield da a (obse ed) and he expec ed alues o
he ALS-based models o ES p oxies (p edic ed)
P edic ed
BILB COWB AMEN BIOD CARB TIMB
Obse ed BILB 12 2 0 3 0 4
COWB 5 17 1 9 0 2
AMEN 1 0 0 1 0 0
BIOD 4 3 0 3 5 1
CARB 0 0 0 2 2 4
TIMB 1 3 0 3 5 9
Table 6 Con usion be ween ESs conside ed as mos impo an
based on he ield da a (obse ed) and MS-NFI da a calib a ed
wi h he local ield sample (p edic ed) using he wo s ou comes
o he p edic ed ES p oxies
P edic ed
BILB COWB AMEN BIOD CARB TIMB
Obse ed BILB 9 7 0 0 0 5
COWB 9 18 0 1 1 5
AMEN 0 1 0 0 0 1
BIOD 8 0 0 0 1 7
CARB 1 2 0 1 0 4
TIMB 3 4 0 0 1 13
Table 7 Con usion be ween ESs conside ed as mos impo an
based on he ield da a (obse ed) and he wo s ou comes o
he ALS-based models o ES p oxies (p edic ed)
P edic ed
BILB COWB AMEN BIOD CARB TIMB
Obse ed BILB 9 0 0 1 10 1
COWB 5 2 3 0 23 1
AMEN 1 0 0 0 1 0
BIOD 1 0 1 0 14 0
CARB 0 0 0 0 7 1
TIMB 1 1 1 0 13 5
Table 8 Con usion be ween ESs conside ed as mos impo an
based on he ield da a (obse ed) and MS-NFI da a calib a ed
wi h he local ield sample (p edic ed) using he bes ou comes
o he p edic ed ES p oxies
P edic ed
BILB COWB AMEN BIOD CARB TIMB
Obse ed BILB 8 4 0 7 2 0
COWB 8 9 0 16 0 1
AMEN 0 1 0 1 0 0
BIOD 2 0 0 14 0 0
CARB 0 0 0 8 0 0
TIMB 3 2 0 13 0 3
Vauhkonen Fo es Ecosys ems (2018) 5:24 Page 16 o 19

Abb e ia ions
ALS: Ai bo ne Lase Scanning; AMEN: Visual ameni y (one o he ecosys em
se ices conside ed, see Table 1); BILB: Sui abili y o bilbe y picking (one o
he ecosys em se ices conside ed, see Table 1); BIOD: Biodi e si y (one o
he ecosys em se ices conside ed, see Table 1); CARB: Ca bon s o age (one
o he ecosys em se ices conside ed, see Table 1); COWB: Sui abili y o
cowbe y picking (one o he ecosys em se ices conside ed, see Table 1);
DBH: Diame e -a -b eas -heigh ; ES: Ecosys em se ice; k-NN: k-Nea es
neighbo ; LOOCV: Lea e one ou c oss alida ion; MCDA: Mul iple c i e ia
decision analysis; MS-NFI: Mul i-Sou ce Na ional Fo es In en o y; RMSE: Roo
mean squa ed e o ; RS: Remo e sensing; TIMB: Timbe p oduc ion (one o
he ecosys em se ices conside ed, see Table 1)
Funding
The acquisi ion o he s udied da a was o iginally suppo ed by he Resea ch
Funds o Uni e si y o Helsinki.
A ailabili y o da a and ma e ials
All da a and ma e ials can be ob ained by eques ing om he au ho .
Au ho ’s con ibu ions
JV is he sole au ho . He ca ied ou all analyses and d a ed he manusc ip .
The au ho ead and app o ed he inal manusc ip .
E hics app o al and consen o pa icipa e
No applicable.
Compe ing in e es s
The au ho decla es ha he has no compe ing in e es s.
Recei ed: 14 Ma ch 2018 Accep ed: 24 May 2018
Re e ences
And ew ME, Wulde MA, Nelson TA (2014) Po en ial con ibu ions o emo e
sensing o ecosys em se ice assessmen s. P og Phys Geog 38:328–353
Ba be QE, Ba e CW, B aid ACR, Coops NC, Tompalski P, Nielsen SE (2016)
Ai bo ne lase scanning o modelling unde s o y sh ub abundance and
p oduc i i y. Fo Ecol Manag 377:46–54
Ba bosa JM, Asne GP (2017) P io i izing landscapes o es o a ion based on
spa ial pa e ns o ecosys em con ols and plan –plan in e ac ions. J Appl
Ecol 54:1459–1468
Ba edo JI, Bas up-Bi k A, Telle A, Onaindia M, de Manuel BF, Mada iaga I,
Rod iguez-Loinaz G, Pinho P, Nunes A, Ramos A, Ba is a M, Mimo S, Co do il C,
B anquinho C, G e -Regamey A, Bebi P, B unne SH, Weibel B, Koppe oinen L,
I konen P, Viinikka A, Chi ici G, Bo alico F, Pesola L, Vizza i M, Ga i V, An onello
L, Ba ba i A, Co ona P, Cullo a S, Giannico V, La o ezza R, Lomba di F, Ma che i
M, Nocen ini S, Riccioli F, T a aglini D, Sallus io L, Rosa io I, on Essen M,
Nicholas KA, Maguas C, Rebelo R, San os-Reis M, San os-Ma in F, Zo illa-Mi as
P, Mon es C, Benayas J, Ma in-Lopez B, Snall T, Be glund H, Beng sson J, Moen
J, Buse o L, San-Miguel-Ayanz J, Thu ne M, Bee C, San o o M, Ca alhais N,
Wu zle T, Schepaschenko D, Sh idenko A, Komp e E, Ah ens B, Le ick SR,
Schmullius C (2015) Mapping and assessmen o o es ecosys ems and hei
se ices –Applica ions and guidance o decision making in he amewo k o
MAES. Repo EUR 27751 EN, Join Resea ch Cen e, Eu opean Union. doi:
h ps://doi.o g/10.2788/720519
Bässle C, S adle J, Mülle J, Fö s e B, Gö lein A, B andl R (2011) LiDAR as a
apid ool o p edic o es habi a ypes in Na u a 2000 ne wo ks. Biodi e s
Conse 20:465–481
Bo alico F, T a aglini D, Chi ici G, Ma che i M, Ma chi E, Nocen ini S, Co ona P
(2014) Classi ying sil icul u al sys ems (coppices s. high o es s) in
Medi e anean oak o es s by ai bo ne lase scanning da a. Eu J Remo e
Sens 47:437–460
Box GEP, Cox DR (1964) An analysis o ans o ma ions. J Royal S a Soc Se B 26:
211–252
B okaw N, Len R (1999) Ve ical s uc u e. In: Hun e ML J (ed) Main aining
biodi e si y in Fo es ecosys ems. Camb idge Uni e si y P ess, Camb idge, pp
373–399
Coops NC, Wulde MA, Cul eno DS, S -Onge B (2004) Compa ison o o es
a ibu es ex ac ed om ine spa ial esolu ion mul ispec al and lida da a.
Can J Remo e Sens 30:855–866
Cos anza R, d’A ge R, de G oo R, Fa be S, G asso M, Hannon B, Limbu g K,
Naeem S, O’Neill RV, Pa uelo J, Raskin RG, Su on P, an den Bel M (1997)
The alue o he wo ld’s ecosys em se ices and na u al capi al. Na u e 387:
253–260
Daily GC, Alexande S, Eh lich PR, Goulde L, Lubchenco J, Ma son PA, Mooney
HA, Pos el S, Schneide SH, Tilman D, Woodwell GM (1997) Ecosys em
se ices: bene i s supplied o human socie ies by na u al ecosys ems. Issues
Ecol 2:1–16
Da ies AB, Asne GP (2014) Ad ances in animal ecology om 3D-LiDAR
ecosys em mapping. T ends Ecol E ol 29:681–691
de G oo RS, Alkemade R, B aa L, Hein L, Willemen L (2010) Challenges in
in eg a ing he concep o ecosys em se ices and alues in landscape
planning, managemen and decision making. Ecol Compl 7:260–272
Domingo-San os JM, de Villa án RF, Rapp-A a ás Í, de P o ens ECP (2011) The
isual exposu e in o es and u al landscapes: an algo i hm and a GIS ool.
Landscape U ban Plan 101:52–58
Duese RD, Shuga HH J (1978) Mic ohabi a s in a o es - loo small mammal
auna. Ecology 59:89–98
Eigenb od F, A mswo h PR, Ande son BJ, Heinemeye A, Gillings S, Roy DB,
Thomas CD, Gas on KJ (2010) The impac o p oxy-based me hods on
mapping he dis ibu ion o ecosys em se ices. J Appl Ecol 47:377–385
Englund O, Be ndes G, Cede be g C (2017) How o analyse ecosys em se ices in
landscapes –a sys ema ic e iew. Ecol Indic 73:492–504
Foody GM (2015) Valuing map alida ion: he need o igo ous land co e map
accu acy assessmen in economic alua ions o ecosys em se ices. Ecol
Econ 111:23–28
Gopal S, Woodcock C (1994) Theo y and me hods o accu acy assessmen o
hema ic maps using uzzy se s. Pho og amm Eng Remo e Sens 60:181–188
Hege schweile KT, Plum C, Fische C, B ändli UB, Ginzle C, Hunzike M (2017)
Towa ds a comp ehensi e social and na u al scien i ic o es - ec ea ion
moni o ing ins umen –a p o o ypical app oach. Landscape U ban Plan
167:84–97
Heinonen T, Ku ila M, Pukkala T (2007) Possibili ies o agg ega e as e cells
h ough spa ial op imiza ion in o es planning. Sil a Fenn 41:89–103
Henningsen A, Hamann JD (2007) Sys em i : a package o es ima ing sys ems o
simul aneous equa ions in R. J S a So w 23(4):1–40
Hilke T, F aze GW, Coops NC, Wulde MA, Newnham GJ, S ewa JD, an
Leeuwen M, Cul eno DS (2013) P edic ion o wood ibe a ibu es om
LiDAR-de i ed o es canopy indica o s. Fo Sci 59:231–242
Hill RA, Hinsley SA, B ough on RK (2014) Assessing habi a s and o ganism-habi a
ela ionships by ai bo ne lase scanning. In: Mal amo M, Næsse E,
Vauhkonen J (eds) Fo es y applica ions o ai bo ne lase scanning. Managing
Fo es ecosys ems, ol 27. Sp inge , Do d ech , pp 335–356
Hou Z, Xu Q, Vauhkonen J, Mal amo M, Tokola T (2016) Species-speci ic
combina ion and calib a ion be ween a ea-based and ee-based diame e
dis ibu ions using ai bo ne lase scanning. Can J Fo Res 46:753–765
Ihalainen M, Alho J, Kolehmainen O, Pukkala T (2002) Expe models o bilbe y
and cowbe y yields in Finnish o es s. Fo Ecol Manag 157:15–22
Kane VR, McGaughey RJ, Bakke JD, Ge sonde RF, Lu z JA, F anklin JF (2010)
Compa isons be ween ield-and LiDAR-based measu es o s and s uc u al
complexi y. Can J Fo Res 40:761–773
Table 9 Con usion be ween ESs conside ed as mos impo an
based on he ield da a (obse ed) and he bes ou comes o
he ALS-based models o ES p oxies (p edic ed)
P edic ed
BILB COWB AMEN BIOD CARB TIMB
Obse ed BILB 11 5 0 4 0 1
COWB 4 20 1 8 0 1
AMEN 0 1 0 1 0 0
BIOD 4 3 0 7 0 2
CARB 0 2 0 4 0 2
TIMB 1 6 0 8 0 6
Vauhkonen Fo es Ecosys ems (2018) 5:24 Page 17 o 19
Kangas A, Kangas J, Ku ila M (2008) Decision suppo o o es managemen .
Managing o es ecosys ems 16. Sp inge , Do d ech
Kangas A, Leskinen P, Kangas J (2007) Compa ison o uzzy and s a is ical
app oaches in mul ic i e ia decisionmaking. Fo Sci 53:37–44
Kangas J (1992) Mul iple-use planning o o es esou ces by using he analy ic
hie a chy p ocess. Scand J Fo Res 7:259–268
Kanka e V, Vauhkonen J, Holopainen M, Vas a an a M, Hyyppä J, Hyyppä H, Alho
P (2015) Spa se densi y, lea -o ai bo ne lase scanning da a in abo eg ound
biomass componen p edic ion. Fo es s 6:1839–1857
Ka jalainen T, Kellomäki S (1996) G eenhouse gas in en o y o land use change
and o es y in Finland based on in e na ional guidelines. Mi ig Adap S a
Glob Change 1:51–71
Kohle M, De aux C, G igulis K, Lei inge G, La o el S, Tappeine U (2017) Plan
unc ional assemblages as indica o s o he esilience o g assland ecosys em
se ice p o ision. Ecol Indic 73:118–127
Koi uniemi J, Ko honen KT (2006) In en o y by compa men s. In: Kangas A,
Mal amo M (eds) Fo es in en o y: me hodology and applica ions. Managing
Fo es ecosys ems, ol 10. Sp inge , Do d ech , pp 271–278
Ko honen L, Ko pela I, Heiskanen J, Mal amo M (2011) Ai bo ne disc e e- e u n
LIDAR da a in he es ima ion o e ical canopy co e , angula canopy
closu e and lea a ea index. Remo e Sens En i on 115:1065–1080
Ko honen L, Peuhku inen J, Malinen J, Su an o A, Mal amo M, Packalén P, Kangas
J (2008) The use o ai bo ne lase scanning o es ima e sawlog olumes.
Fo es y 81:499–510
Ko pela I, Ho i A, Mo sdo F (2012) Unde s o y ees in ai bo ne LiDAR da a -
selec i e mapping due o ansmission losses and echo- igge ing
mechanisms. Remo e Sens En i on 119:92–104
Ko amaa E, Tokola T, Mal amo M, Packalén P, Ku ila M, Mäkinen A (2010)
In eg a ion o emo e sensing-based bioene gy in en o y da a and op imal
bucking o s and-le el decision making. Eu J Fo Res 129:875–886
Lämås T, Sands öm E, Jonzén J, Olsson H, Gus a sson L (2015) T ee e en ion
p ac ices in bo eal o es s: wha kind o u u e landscapes a e we c ea ing?
Scand J Fo Res 30:526–537
Le sky MA, Cohen WB, Spies TA (2001) An e alua ion o al e na e emo e sensing
p oduc s o o es in en o y, moni o ing, and mapping o Douglas- i o es s
in wes e n O egon. Can J Fo Res 31:78–87
Leh omäki J, Tuominen S, Toi onen T, Leinonen A (2015) Wha da a o use o
o es conse a ion planning? A compa ison o coa se open and de ailed
p op ie a y o es in en o y da a in Finland. PLoS One. h ps://doi.o g/10.
1371/jou nal.pone.0135926
Lei e e R, Fu e R, Schaepman ME, Mo sdo F (2015) Fo es canopy-s uc u e
cha ac e iza ion: a da a-d i en app oach. Fo Ecol Manag 358:48–61
Liang X, Hyyppä J, Ma ikainen L (2007) Deciduous-coni e ous ee classi ica ion
using di e ence be ween i s and las pulse lase signa u es. In: Rönnholm P,
Hyyppä H, Hyyppä J (eds) P oceedings o ISPRS wo kshop on lase scanning
2007 and Sil iLase 2007. In a ch Pho og amm emo e Sens, ol XXXVI, pa
3/W52, pp 253–257
Lis opad CM, Mas e s RE, D ake J, Weishampel J, B anquinho C (2015) S uc u al
di e si y indices based on ai bo ne LiDAR as ecological indica o s o
managing highly dynamic landscapes. Ecol Indic 57:268–279
Lu he JE, Skinne R, Fou nie RA, an Lie OR, Bowe s WW, Co é JF, Hopkinson C,
Moul on T (2014) P edic ing wood quan i y and quali y a ibu es o balsam
i and black sp uce using ai bo ne lase scanne da a. Fo es y 87:313–326
MacA hu RH, MacA hu JW (1961) On bi d species di e si y. Ecology 42:594–598
Malczewski J, Rinne C (2015) Mul ic i e ia decision analysis geog aphic
in o ma ion science. Ad ances in Geog aphic In o ma ion Science Sp inge -
Ve lag, Be lin Heidelbe g
Mal amo M, Malinen J, Packalén P, Su an o A, Kangas J (2006) Nonpa ame ic
es ima ion o s em olume using ai bo ne lase scanning, ae ial
pho og aphy, and s and- egis e da a. Can J Fo Res 36:426–436
Mal amo M, Næsse E, Vauhkonen J (eds) (2014) Fo es y applica ions o ai bo ne
lase scanning - concep s and case s udies. Managing Fo es ecosys ems, ol
27. Sp inge , Do d ech
Mal amo M, Packalén P, Yu X, Ee ikäinen K, Hyyppä J, Pi känen J (2005)
Iden i ying and quan i ying s uc u al cha ac e is ics o he e ogeneous bo eal
o es s using lase scanne da a. Fo Ecol Manag 216:41–50
Ma ínez-Ha ms MJ, Bal ane a P (2012) Me hods o mapping ecosys em se ice
supply: a e iew. In J biodi Sci Ecosys Se Manage 8:17–25
Melin M, Meh ä alo L, Mie inen J, Tossa ainen S, Packalen P (2016) Fo es
s uc u e as a de e minan o g ouse b ood occu ence –an analysis linking
LiDAR da a wi h p esence/absence ield da a. Fo Ecol Manag 380:202–211
Melin M, Packalen P, Ma ala J, Meh ä alo L, Pusenius J (2013) Assessing and
modeling moose (Alces alces) habi a s wi h ai bo ne lase scanning da a. In J
Appl Ea h Obs Geoin o 23:389–396
Mülle J, Vie ling K (2014) Assessing biodi e si y by ai bo ne lase scanning. In:
Mal amo M, Næsse E, Vauhkonen J (eds) Fo es y applica ions o ai bo ne
lase Scanning.Managing Fo es ecosys ems, ol 27. Sp inge , Do d ech , pp
357–374
Næsse E (2002) P edic ing o es s and cha ac e is ics wi h ai bo ne scanning
lase using a p ac ical wo-s age p ocedu e and ield da a. Remo e Sens
En i on 80:88–99
Næsse E, Gobakken T (2008) Es ima ion o abo e- and below-g ound biomass
ac oss egions o he bo eal o es zone using ai bo ne lase . Remo e Sens
En i on 112:3079–3090
Na u al Resou ces Ins i u e Finland (2017) File se ice o publicly a ailable da a.
h p://ka a.me la. i/index-en.h ml. Accessed 16 Oc 2017
Niemi MT, Vauhkonen J (2016) Ex ac ing canopy su ace ex u e om ai bo ne
lase scanning da a o he supe ised and unsupe ised p edic ion o a ea-
based o es cha ac e is ics. Remo e Sens. h ps://doi.o g/10.3390/ s8070582
Ø ka HO, Dalpon e M, Gobakken T, Næsse E, Ene LT (2013) Cha ac e izing o es
species composi ion using mul iple emo e sensing da a sou ces and
in en o y app oaches. Scand J Fo Res 28:677–688
Ø ka HO, Gobakken T, Næsse E, Ene L, Lien V (2012) Simul aneously acqui ed
ai bo ne lase scanning and mul ispec al image y o indi idual ee species
iden i ica ion. Can J Remo e Sens 38:125–138
Packalén P, Heinonen T, Pukkala T, Vauhkonen J, Mal amo M (2011) Dynamic
ea men uni s in Eucalyp us plan a ion. Fo Sci 57:416–426
Pascual C, Ga cía-Ab il A, Ga cía-Mon e o LG, Ma ín-Fe nández S, Cohen WB
(2008) Objec -based semi-au oma ic app oach o o es s uc u e
cha ac e iza ion using lida da a in he e ogeneous Pinus syl es is s ands. Fo
Ecol Manag 255:3677–3685
Pa enaude G, Hill RA, Milne R, Ga eau DL, B iggs BBJ, Dawson TP (2004)
Quan i ying o es abo e g ound ca bon con en using LiDAR emo e
sensing. Remo e Sens En i on 93:368–380
Peu a M, Gonzalez RS, Mülle J, Heu ich M, Vie ling LA, Mönkkönen M, Bässle C
(2016) Mapping a ‘c yp ic kingdom’: pe o mance o lida de i ed
en i onmen al a iables in modelling he occu ence o o es ungi. Remo e
Sens En i on 186:428–438
Popescu SC, Hauglin M (2014) Es ima ion o biomass componen s by ai bo ne
lase scanning. In: Mal amo M, Næsse E, Vauhkonen J (eds) Fo es y
applica ions o ai bo ne lase scanning. Managing Fo es ecosys ems, ol 27.
Sp inge , Do d ech , pp 157–175
Pukkala T (2005) Me sikön uo oa on ennus emalli ki ennäismaan männiköille,
kuusikoille ja auduskoi ikoille (in Finnish o “p edic ion models o he
expec a ion alue o pine, sp uce and bi ch s ands on mine al soils”).
Me sä ie een Aikakauski ja 3(2005):311–322
Pukkala T (2008) In eg a ing mul iple se ices in he nume ical analysis o landscape
design. In: on Gadow K, Pukkala T (eds) Designing G een Landscapes.
Managing Fo es Ecosys ems, ol 15. Sp inge , Do d ech , pp 137–167
Pukkala T (2016) Which ype o o es managemen p o ides mos ecosys em
se ices? Fo es Ecosys . h ps://doi.o g/10.1186/s40663-016-0068-5
Pukkala T, Kangas J (1996) A me hod o in eg a ing isk and a i ude owa d isk
in o o es planning. Fo Sci 42:198–205
Pukkala T, Kellomäki S, Mus onen E (1988) P edic ion o he ameni y o a ee
s and. Scand J Fo Res 3:533–544
Pukkala T, Packalén P, Heinonen T (2014) Dynamic ea men uni s in o es
managemen planning. In: Bo ges JG, Diaz-Bal ei o L, McDill ME, Rod iguez
LCE (eds) The Managemen o Indus ial Fo es Plan a ions. Managing Fo es
ecosys ems, ol 33. Sp inge , Do d ech , pp 373–392
Pukkala T, Sulka a R, Jaakkola L, Lähde E (2012) Rela ionships be ween economic
p o i abili y and habi a quali y o Sibe ian jay in une en-aged No way
sp uce o es . Fo Ecol Manag 276:224–230
R Co e Team (2016) R: A language and en i onmen o s a is ical compu ing. R
Founda ion o S a is ical Compu ing, Vienna. h ps://www.R-p ojec .o g/.
Accessed 16 Oc 2017
Räsänen A, Lensu A, Tomppo E, Kui unen M (2015) Compa ing conse a ion
alue maps and mapping me hods in a u al landscape in sou he n Finland.
Landscape Online 44:1–19
Rä y J, Vauhkonen J, Mal amo M, Tokola T (2016) On he po en ial o
p ede e mine dominan ee species based on spa se-densi y ai bo ne lase
scanning da a o imp o ing subsequen p edic ions o species-speci ic
imbe olumes. Fo es Ecosys . h ps://doi.o g/10.1186/s40663-016-0060-0
Vauhkonen Fo es Ecosys ems (2018) 5:24 Page 18 o 19
Rechs eine C, Zellwege F, Ge be A, B eine FT, Bollmann K (2017) Remo ely
sensed o es habi a s uc u es imp o e egional species conse a ion.
Remo e Sens Ecol Conse 3:247–258
Roces-Díaz JV, Bu kha d B, K use M, Mülle F, Díaz-Va ela ER, Ál a ez-Ál a ez P (2017) Use
o ecosys em in o ma ion de i ed om o es hema ic maps o spa ial analysis o
ecosys em se ices in no hwes e n Spain. Landscape Ecol Eng 13:45–57
Sani NA, Ka aky SB, Pukkala T, Ma aji A (2016) In eg a ed use o GIS, emo e
sensing and mul i-c i e ia decision analysis o assess ecological land
sui abili y in mul i- unc ional o es y. J Fo Res 27:1127–1135
Sch ö e M, Rusch GM, Ba on DN, Blumen a h S, No dén B (2014) Ecosys em
se ices and oppo uni y cos s shi spa ial p io i ies o conse ing o es
biodi e si y. PLoS One. h ps://doi.o g/10.1371/jou nal.pone.0112557
Schulp CJE, Bu kha d B, Maes J, Van Vlie J, Ve bu g PH (2014) Unce ain ies in
ecosys em se ice maps: a compa ison on he Eu opean scale. PLoS One.
h ps://doi.o g/10.1371/jou nal.pone.0109643
Shapi o SS, Wilk MB (1965) An analysis o a iance es o no mali y (comple e
samples). Biome ika 52:591–611
Simonson WD, Allen HD, Coomes DA (2014) Applica ions o ai bo ne lida o he
assessmen o animal species di e si y. Me hods Ecol E ol 5:719–729
S e d up-Thygeson A, Ø ka HO, Gobakken T, Næsse E (2016) Can ai bo ne lase
scanning assis in mapping and moni o ing na u al o es s? Fo Ecol Manag
369:116–125
Thompson SD, Nelson TA, Giesb ech I, F aze G, Saunde s SC (2016) Da a-d i en
egionaliza ion o o es ed and non- o es ed ecosys ems in coas al B i ish
Columbia wi h LiDAR and RapidEye image y. Appl Geog 69:35–50
Tomppo E, Haakana M, Ka ila M, Pe äsaa i J (2008a) Mul i-sou ce na ional o es
in en o y –me hods and applica ions. Managing o es ecosys ems, ol 18.
Sp inge , Do d ech
Tomppo E, Halme M (2004) Using coa se scale o es a iables as ancilla y
in o ma ion and weigh ing o a iables in k-NN es ima ion: a gene ic
algo i hm app oach. Remo e Sens En i on 92:1–20
Tomppo E, Ka ila M, Mäkisa a K, Pe äsaa i J (2014) The mul i-sou ce na ional
o es in en o y o Finland - me hods and esul s 2011. Wo king Pape s o
he Finnish Fo es Resea ch Ins i u e, ol 319. h p://www.me la. i/julkaisu /
wo kingpape s/2014/mwp319.h m. Accessed 16 Oc 2017
Tomppo E, Olsson H, S åhl G, Nilsson M, Hagne O, Ka ila M (2008b) Combining
na ional o es in en o y ield plo s and emo e sensing da a o o es
da abases. Remo e Sens En i on 112:1982–1999
Valbuena R, Ee ikäinen K, Packalen P, Mal amo M (2016a) Gini coe icien
p edic ions om ai bo ne lida emo e sensing display he e ec o
managemen in ensi y on o es s uc u e. Ecol Indic 60:574–585
Valbuena R, Mal amo M, Ma ín-Fe nández S, Packalen P, Pascual C, Nabuu s
GJ (2013) Pa e ns o co a iance be ween ai bo ne lase scanning
me ics and Lo enz cu e desc ip o s o ee size inequali y. Can J
Remo e Sens 39(sup1):S18–S31
Valbuena R, Mal amo M, Meh ä alo L, Packalen P (2017) Key s uc u al ea u es o
bo eal o es s may be de ec ed di ec ly using L-momen s om ai bo ne lida
da a. Remo e Sens En i on 194:437–446
Valbuena R, Mal amo M, Packalen P (2016b) Classi ica ion o mul ilaye ed o es
de elopmen classes om low-densi y na ional ai bo ne lida da ase s.
Fo es y 89:392–401
Valbuena R, Vauhkonen J, Packalen P, Pi känen J, Mal amo M (2014) Compa ison
o ai bo ne lase scanning me hods o es ima ing o es s uc u e indica o s
based on Lo enz cu es. ISPRS J Pho og amm Remo e Sens 95:23–33
Vauhkonen J, Imponen J (2016) Unsupe ised classi ica ion o ai bo ne lase
scanning da a o loca e po en ial wildli e habi a s o o es managemen
planning. Fo es y 89:350–363
Vauhkonen J, Packalen P, Malinen J, Pi känen J, Mal amo M (2014) Ai bo ne lase
scanning based decision suppo o wood p ocu emen planning. Scand J
Fo Res 29(Suppl.1):132–143
Vauhkonen J, Ruo salainen R (2017a) Assessing he p o isioning po en ial o
ecosys em se ices in a Scandina ian bo eal o es : sui abili y and adeo
analyses on g id-based wall- o-wall o es in en o y da a. Fo Ecol Manag
389:272–284
Vauhkonen J, Ruo salainen R (2017b) Recons uc ing o es canopy om he 3D
iangula ions o ai bo ne lase scanning poin da a o he isualiza ion and
planning o o es ed landscapes. Ann Fo Sci 74:9. h ps://doi.o g/10.1007/
s13595-016-0598-6
Vihe aa a P, Au inen AP, Mononen L, To ma M, Ahl o h P, An ila S, Bo che K,
Fo sius M, Heino J, Heliola J, Koskelainen M, Kuussaa i M, Meissne K, Ojala O,
Tuominen S, Vii asalo M, Vi kkala R (2017) How essen ial biodi e si y a iables
and emo e sensing can help na ional biodi e si y moni o ing. Glob Ecol
Conse 10:43–59
Zimble DA, E ans DL, Ca lson GC, Pa ke RC, G ado SC, Ge a d PD (2003)
Cha ac e izing e ical o es s uc u e using small- oo p in ai bo ne LiDAR.
Remo e Sens En i on 87:171–182
Zolkos SG, Goe z SJ, Dubayah R (2013) A me a-analysis o e es ial abo eg ound
biomass es ima ion using lida emo e sensing. Remo e Sens En i on 128:
289–298
Vauhkonen Fo es Ecosys ems (2018) 5:24 Page 19 o 19