Ea h Sys . Sci. Da a, 10, 173–184, 2018
h ps://doi.o g/10.5194/essd-10-173-2018
© Au ho (s) 2018. This wo k is dis ibu ed unde
he C ea i e Commons A ibu ion 4.0 License.
Webcam ne wo k and image da abase o s udies o
phenological changes o ege a ion and snow co e in
Finland, image ime se ies om 2014 o 2016
Mikko Pel oniemi1, Mika Au ela2, K is in Bö che 3, Pasi Kola i4, John Loeh 5, Jouni Ka hu6,
Maiju Linkosalmi2, Cemal Melih Tanis2, Juha-Pekka Tuo inen2, and Ali Nadi A slan2
1Na u al Resou ces Ins i u e Finland (Luke), La oka anonkaa i 9, 00790, Helsinki, Finland
2Finnish Me eo ological Ins i u e, E ik Palménin aukio 1, 00560 Helsinki, Finland
3Finnish En i onmen Ins i u e (SYKE), Mechelininka u 34a, 00251 Helsinki, Finland
4Depa men o Physics, Uni e si y o Helsinki, P.O. Box 68, 00014 Helsinki, Finland
5Lammi Biological S a ion, Uni e si y o Helsinki, Pääjä en ie 320, 16900 Lammi, Finland
6Na u al Resou ces Ins i u e Finland (Luke), Paa o Ha aksen ie 3, 90014 Oulun yliopis o, Finland
Co espondence: Mikko Pel oniemi (mikk[email p o ec ed])
and Ali Nadi A slan (ali.nadi [email p o ec ed])
Recei ed: 30 June 2017 – Discussion s a ed: 10 Augus 2017
Re ised: 27 No embe 2017 – Accep ed: 29 No embe 2017 – Published: 25 Janua y 2018
Abs ac . In ecen yea s, moni o ing o he s a us o ecosys ems using low-cos web (IP) o ime lapse cam-
e as has ecei ed wide in e es . Wi h b oad spa ial co e age and high empo al esolu ion, ne wo ked cam-
e as can p o ide in o ma ion abou snow co e and ege a ion s a us, se e as g ound u hs o Ea h ob-
se a ions and be use ul o gap- illing o cloudy a eas in Ea h obse a ion ime se ies. Ne wo ked cam-
e as can also play an impo an ole in supplemen ing labo ious phenological ield su eys and ci izen sci-
ence p ojec s, which also su e om obse e -dependen obse a ion bias. We es ablished a ne wo k o dig-
i al su eillance came as o au oma ed moni o ing o phenological ac i i y o ege a ion and snow co e in
he bo eal ecosys ems o Finland. Came as we e moun ed a 14 si es, each si e ha ing 1–3 came as. He e,
we documen he ne wo k, basic came a in o ma ion and access o images in he pe manen da a eposi o y
(h p://www.zenodo.o g/communi ies/phenology_came a/). Indi idual DOI- e e enced image ime se ies consis
o hal -hou ly images collec ed be ween 2014 and 2016 (h ps://doi.o g/10.5281/zenodo.1066862). Addi ionally,
we p esen an example o a colou index ime se ies de i ed om images om wo con as ing si es.
1 In oduc ion
Yea - o-yea a ia ion in he phenological cycle o seasonal
ecosys ems is la ge. The snow co e and he iming o
snowmel a y conside ably om yea o yea . Fo exam-
ple, in no he n and high-ele a ion ecosys ems, ege a ion
ollows he empo al a ia ion o empe a u es, and budbu s
da es can a y by 40 days (Häkkinen, 1999), bu ou abili y o
p edic his a ia ion is limi ed (Basle , 2016). Consequen ly,
phenology emains among he key componen s causing un-
ce ain ies in ou es ima es o ege a ion ca bon balances
(Richa dson e al., 2013). Ob iously, mo e moni o ing da a
a e equi ed o imp o ing he unde s anding o he sou ces
o phenological a ia ion and o di ec ly suppo i s moni o -
ing, as p esen me hods o moni o ing plan phenology and
snow co e in he ield a e labo ious and cos ly. Consume -
g ade came as p o ide an in e es ing oppo uni y o moni o
he seasonal cycle o ecosys ems because hey eco d s uc-
u al in o ma ion a high spa ial and empo al esolu ion, and
in doing so, ime se ies o images can p o ide use ul in o -
ma ion abou empo al changes in ecosys ems a a low cos .
In ecen yea s, he ecological communi y has s a ed o
collec image ime se ies o he pu poses o phenological
analysis (Richa dson e al., 2007). In hese analyses, ypi-
Published by Cope nicus Publica ions.
174 M. Pel oniemi e al.: Webcam image da abase o phenological s udies in Finland
cally some sub- egions a e used o ack colou changes in
ime, using indices ep esen ing ch oma ic changes in he
a ge scene, which a e ela ed o phenological changes o
lea colou , quan i y (eme gence and abscission), mo phol-
ogy and p e ailing condi ions a he si e. Analyses o ch o-
ma ic changes and associa ed phenological ansi ion da es
can be easily au oma ed by using s a is ical me hods ha
iden i y b eak poin s in ime se ies (Elmo e e al., 2012;
Klos e man e al., 2014; Winga e e al., 2015). Al hough
colou s o he images in he ime se ies a e a ec ed by il-
lumina ion angle, cloud co e and came a ype, much o he
po en ial p oblems in phenological ime se ies can be la gely
a oided by using images om simila en i onmen al condi-
ions, and by selec ing a subse o images o analyses and
applying colou index app op ia ely. G een ch oma ic coo -
dina e, GCC =G/(R +B+G), on he o he hand, is a ai ly
obus index o colou s, because he opposing changes in ed
and blue pa ially balance each o he ou o e he cou se o
he day (Winga e e al., 2015). GCC has been used in pheno-
logical analyses by es ima ing a mean using midday images
and hen a e aging he es ima es o e 1–3 days (e.g. Sonnen-
ag e al., 2012; Linkosalmi e al., 2016).
Came as ha e been deployed on a ange o ecosys ems, in-
cluding deciduous and mixed species ecosys ems (Richa d-
son e al., 2007; Ah ends e al., 2009; Sonnen ag e al., 2012;
Mizunuma e al., 2013), g asslands (e.g. Miglia acca e al.,
2011), pea lands (Wes e gaa d-Nielsen e al., 2013; Peichl
e al., 2015; Linkosalmi e al., 2016) and coni e ous o es s
(Nagai e al., 2012; Linkosalmi e al., 2016). Budbu s and
lea senescence o deciduous species and hei ela ionship
wi h CO2exchange ha e been a ocus in a numbe o s udies
(Richa dson e al., 2007; Ah ends e al., 2009; Sonnen ag e
al., 2012; Mizunuma e al., 2013; Winga e e al., 2015). Rea-
sons o plan colou changes ha e been ela ed o he a i-
a ion o g een biomass, and also o changes in he biochem-
ical p ope ies o lea es occu ing o e he season (Keenan
e al., 2014; Yang e al., 2014). While mo e de ailed adia-
i e ans e analyses may be equi ed o con i m he conse-
quences o pigmen and o he changing p ope ies o lea es
on he colou s (Winga e e al., 2015), he clea es phenomena
such as budbu s and au umn colou peaks can al eady be ex-
ac ed wi h simple ime se ies and b eak poin analyses wi h
ai ly good accu acy (e.g. Sonnen ag e al., 2012; Klos e man
e al., 2014; Winga e e al., 2015; Pel oniemi e al., 2018).
Besides ege a ion moni o ing, ime lapse images ha e been
used in snow co e moni o ing. Sal a o i e al. (2011) de el-
oped a me hod o analysing snow co e changes based on
he h esholding o blue channel his og ams. The me hod has
been ecen ly es ed also in no he n bo eal Finland (A slan
e al., 2017). Pe haps he mos ob ious use o came a ime
se ies s ems om he possibili y o using image a chi es o
documen ing he si e his o y. As such, hey p o ide suppo -
i e in o ma ion on he ambien condi ions, i.e. in o ma ion
abou wea he , ege a ion and unexpec ed dis u bances.
Gi en he weal h o ea u es ha can be ex ac ed om im-
age ime se ies, i is no su p ising he e is a la ge in e es in
using hem as g ound e e ences o Ea h obse a ion (EO)
p oduc s. A pa icula bene i o came as is ha hey can
p o ide a con inuous ime se ies o e e ence obse a ions,
as hey a e no o a e a ely occluded by clouds. P e ious
esea ch has al eady shown good co espondence be ween
he came a-based and EO phenology p oduc s (Zhang e al.,
2006; Mo ise e e al., 2009; Elmo e e al., 2012; Hu kens a
al., 2012; Klos e man e al., 2014).
The ull bene i s o using came as in ecosys em moni-
o ing a e eached when in o ma ion om mul iple came as
a e analysed oge he , and hus many o he came as ha e
been se up in smalle o la ge ne wo ks o si es. In he
US, PhenoCam ne wo k has collec ed ime lapse images o
se e al yea s (Richa dson e al., 2009; h p://phenocam.s .
unh.edu). In Japan, esea che s ha e collec ed images om
a mosphe e–ecosys em ca bon lux si es wi hin he ame-
wo k o he Japan Phenological Eyes Ne wo k (PEN; Nasa-
ha a and Nagai, 2015; h p://pen.agbi. sukuba.ac.jp) wi h he
aim o linking hem o o he emo e sensing and ield obse -
a ions. The Eu opean ne wo k o ca bon lux si es ope a e
came as a leas in 50 si es (Winga e e al., 2015). Phenolog-
ical came a moni o ing is being mo e widely implemen ed
wi hin he amewo k o he US Na ional Ecological Ob-
se a o y Ne wo k (NEON) and he Eu opean Union’s In e-
g a ed Ca bon Obse a ion Sys em (ICOS). Many ne wo ks
(e.g. Aus alian Phenocam Ne wo k, h p://www.phenocam.
o g.au) and ICOS a e p o iding da a online by eques and/o
ha e plans o open dis ibu ion o images (B own e al.,
2016).
Keeping he bene i s o using came as o ecosys em close
ange emo e sensing in mind, and in o de o con ibu e
o he g owing body o came as, which a e moni o ing he
ecosys ems o he Ea h, we es ablished a ne wo k o cam-
e as o moni o ing bo eal ecosys ems in Finland. He e, we
documen he si es o his moni o ing ne wo k including he
equipmen used and he a ailabili y o da a in open eposi-
o ies, as well as epo ing ou i s -hand expe iences while
se ing up he ne wo k and en isage u he de elopmen
needs o he ne wo k. We ha e made image da a public
om 27 came as a 14 si es in he Zenodo da a a chi e
(h ps://www.zenodo.o g/), which is appended wi h me ada a
shee s ul illing he c i e ia de eloped by B own e al. (2016)
o in eg a ed came a ne wo ks. Addi ionally, we p epa ed
analyses o Sun angle e ec s on he annual cou se o GCC
a wo con as ing si es, one loca ed in he sou he n bo eal
egion and one in he suba c ic egion whe e days a e sho .
We analysed wo ypes o a ge s in he sou he n si e, namely
a b oad-lea ed and a coni e ee which undamen ally di e
in canopy and lea mo phology and annual lea pe iod (sea-
sonal s. e e g een). In he suba c ic si e, we also analysed
colou changes o bi ch and we land g asses.
Ea h Sys . Sci. Da a, 10, 173–184, 2018 www.ea h-sys -sci-da a.ne /10/173/2018/
M. Pel oniemi e al.: Webcam image da abase o phenological s udies in Finland 175
2 Ne wo k o came as
2.1 Came as and came a con igu a ions
All came as a e se o a ixed whi e balance and au o-
ma ic exposu e, a ge ed no hwa ds whe e easible depend-
ing on he iew and moun ing op ions, and igge ed o hal -
hou ly submission o snapsho s o an FTP se e , excluding
he nigh hou s. Images a e aken a maximum esolu ion
allowed by he came a (2594 ×1944), excep in Hyy iälä
whe e hey a e aken a qua e o he maximum esolu-
ion allowed by he came a (1024 ×768). Image JPEGs a e
o highes possible quali y. Mos o he si es and analyses
o his s udy used image ime se ies acqui ed wi h S a Do
Ne Cam SC5 came as, while mo e ecen ly also an AXIS
P1357E came a was ins alled a one o he si es (Paljakka).
We did no expec colou ep oduc ion o came as o be s an-
da dized and calib a ed, i.e. compa able ac oss came as, so
we allowed some a ia ion in he came a se ings by lo-
cal condi ions. In S a Do came as, we modi ied he R, G
and B channel ampli ica ion se ings om he de aul se -
ings o local scenes so ha came as p oduced isually ap-
p oxima ely ealis ic colou s. These se ings we e kep ixed
he ea e , and hey a e epo ed in he came a da ashee
(h ps://doi.o g/10.5281/zenodo.1066862; Pel oniemi e al.,
2017). In AXIS P1357E, whi e balance was se o ixed
(“Fixed ou doo 1” in he came a menu). Pixel digi al num-
be s a e no hus di ec ly compa able ac oss came as, bu al-
low came a-speci ic analyses o , o example, he empo al
de elopmen o digi al numbe s and indices based on hem.
I is and zoom we e ixed o came a speci ic alues depend-
ing on he iew o he came a. Came as adjus o b igh ness
o he images au oma ically. Selec ed came as also ha e g ey
e e ence pla es o moni o he s abili y o colou ep oduc-
ion o he came as.
2.2 Si es
The loca ions o came a moni o ing si es anged om he
hemibo eal (T ä minne) o he no he n bo eal (Kaamanen)
phy ogeog aphical zone (Fig. 1, Table 1). All si es belong o
he suba c ic clima e zone (D c) acco ding o Köppen’s cli-
ma e classi ica ion, excep he T ä minne si e, which is si -
ua ed in he wa m-summe humid con inen al clima e zone
(D b). The s a and du a ion o con inuous snow co e a ies
along he clima e g adien om sou hwes o no h in Fin-
land. Pe manen snow co e a i es on a e age (no mal pe-
iod 1981–2010) in Oc obe and Janua y and mel s in May
and Ma ch o he Kaamanen and T ä minne si es, espec-
i ely. The he mal g owing season begins when mean daily
empe a u es exceed 5 ◦C a he end o Ap il in sou he n Fin-
land and a he end o May in no he n Lapland. Tempe a-
u es dec ease below 5 ◦C ending he he mal g owing season
a he end o Sep embe in he no h and in la e Oc obe o
he beginning o No embe in sou hwes e n Finland.
Figu e 1. Si es in he phenological came a ne wo k, s a us end o
2016. Numbe s e e o si es in Table 1. Da a sou ces: Coun y bo -
de © ESRI, Vege a ion zones © SYKE, 2015.
The came a moni o ing si es include ICOS and o he eddy-
co a iance si es, LTER (Long-Te m Ecosys em Resea ch)
and ICP Fo es s Le el II si es, and o he si es wi h p e i-
ous phenological moni o ing. Some si es a e highly ins u-
men ed and in ensi ely moni o ed ecosys em esea ch si es
such as Hyy iälä SMEAR II, Vä iö SMEAR I and So-
dankylä eddy co a iance si es, while o he s a e less in en-
si ely moni o ed. Fo mos o he si es he e is a leas a me-
eo ological s a ion, and a ying numbe and ypes o ancil-
la y moni o ing measu emen s. Da a om he Finnish Me-
eo ological Ins i u e (FMI) me eo ological s a ions can be
eely downloaded om he FMI open da a po al (h ps:
//en.ilma ie eenlai os. i/open-da a), e.g. by using a use in-
e ace “FMI wea he da a downloade ” (h ps://gi hub.com/
Tume su/FMI-wea he -downloade ).
We ins alled 1–3 came as o he moni o ing si es. The
numbe o came as depended on he canopy condi ions a he
si e (Table 1; see also h p://monime . mi. i/?page=Came as).
The came as iew he canopy, g ound and/o a wide land-
scape (Fig. 2). Some came as we e moun ed close o in-
di idual ee c owns, hus na owing he egion o in e es
(ROI) o image analysis o he a ge ees wi hou con ound-
ing con ibu ion om he g ound. The came as moun ed be-
low he canopy moni o he phenology o unde s o ey ege-
a ion and snow co e . A he we land si es, a single came a
ul ils all hese pu poses. Many o he came as allow de ining
mul iple ROIs o moni o a ious species and phenomena a
he same ime.
www.ea h-sys -sci-da a.ne /10/173/2018/ Ea h Sys . Sci. Da a, 10, 173–184, 2018
176 M. Pel oniemi e al.: Webcam image da abase o phenological s udies in Finland
Table 1. Basic in o ma ion abou he came a si es, a ge s and he a ailabili y o images. Addi ional in o ma ion o came a ne wo k and
images om 2014 o 2016 can be ound in Zenodo collec ion (h ps://doi.o g/10.5281/zenodo.1066862). Coo dina es a e in decimal deg ees
WGS84.
No. Si e (and image
ime pe iod)
La ., long. Al i ude
abo e sea
(m)
Mean annual ai
empe a u e (◦C)
and p ecipi a ion
(mm)
Ecosys em/dominan
species
Came a ID/posi ion/
colou space
DOI
1 Hyy iälä
(2014–2016)
61.85, 24.30 180 3.5/711 Sco s pine MC106/C own/RGB
MC107/G ound/RGB
h ps://doi.o g/10.5281/zenodo.815559
h ps://doi.o g/10.5281/zenodo.815557
2 Kaamanen
(2015–2016)
69.14, 27.27 155b−0.4/472 We land MC128/G ound/RGB h ps://doi.o g/10.5281/zenodo.815553
3 Ken ä o a
(2015–2016)
67.99, 24.24 347b−1.0/521 No way sp uce MC114/Canopy/RGB
MC115/C own/RGB
MC116/G ound/RGB
h ps://doi.o g/10.5281/zenodo.815519
h ps://doi.o g/10.5281/zenodo.815523
h ps://doi.o g/10.5281/zenodo.815521
4 Lammi (2016) 61.05, 25.04 119a4.3/644 Canopy: mixed;
O he came as:
Downy bi ch
MC122/Canopy/RGB
MC123/Canopy/IR
MC124/Landscape/RGB
MC125/Landscape/IR
MC126/G ound/RGB
MC127/G ound/IR
h ps://doi.o g/10.5281/zenodo.815538
h ps://doi.o g/10.5281/zenodo.815540
h ps://doi.o g/10.5281/zenodo.815542
h ps://doi.o g/10.5281/zenodo.815544
h ps://doi.o g/10.5281/zenodo.815546
h ps://doi.o g/10.5281/zenodo.815548
5 Lompolo-
jänkkä (2015–
2016)
69.80, 24.21 274 −1.0/521bWe land MC129/Lompolojänkkä/
RGB
h ps://doi.o g/10.5281/zenodo.815555
6 Paljakka
(2015–2016)
64.68, 28.11 257a1.6/918cSp uce s and MC117/Canopy/RGB
MC118/Canopy/IR
MC117-1/ Canopy/RGB
h ps://doi.o g/10.5281/zenodo.815529
h ps://doi.o g/10.5281/zenodo.815527
h ps://doi.o g/10.5281/zenodo.815525
7 Pa kano (2015–
2016)
62.03, 23.04 96a3.5/667bMixed MC112/Landscape/RGB h ps://doi.o g/10.5281/zenodo.815487
8 Punkaha ju
(2014–2016)
61.81, 29.32 88a3.8/604bNo way sp uce gMC103/G ound/RGB
gMC104/C own/RGB
gMC105/Landscape/RGB
h ps://doi.o g/10.5281/zenodo.815460
h ps://doi.o g/10.5281/zenodo.815462
h ps://doi.o g/10.5281/zenodo.815464
9 Sodankylä,
pine (2014–
2016)
67.36, 26.64 179 −0.4/527bSco s pine MC108/Canopy/RGB
MC109/C own/RGB
MC110/G ound/RGB
h ps://doi.o g/10.5281/zenodo.815479
h ps://doi.o g/10.5281/zenodo.815481
h ps://doi.o g/10.5281/zenodo.815483
10 Sodankylä,
we land (2014–
2016)
67.37 180 −0.4/527bWe land MC111/G ound/Pea land/
RGB
h ps://doi.o g/10.5281/zenodo.815485
11 Suonenjoki
(2015–2016)
62.64, 27.05 135a3.9/566dSco s pine MC113/Canopy/RGB h ps://doi.o g/10.5281/zenodo.815489
12 Tammela
(2014–2016)
60.65, 23.81 144a5.4/601eNo way sp uce MC102/G ound/RGB
hMC101/C own/RGB
MC100/Canopy/RGB /
h ps://doi.o g/10.5281/zenodo.815456
h ps://doi.o g/10.5281/zenodo.815454
h ps://doi.o g/10.5281/zenodo.815450
13 T ä minne
(2016)
59.84, 23.25 3 6/634bMixed MC130/Landscape/RGB h ps://doi.o g/10.5281/zenodo.815550
14 Vä iö (2015–
2016)
67.75, 29.61 400 −0.5/601 Sco s pine MC119/Canopy/RGB
MC120/C own/RGB
MC121/G ound/RGB
h ps://doi.o g/10.5281/zenodo.815532
h ps://doi.o g/10.5281/zenodo.815534
h ps://doi.o g/10.5281/zenodo.815536
aAl i ude was de e mined om digi al ele a ion model wi h a g id size o 10 m ×10 m p o ided by he Na ional Land Su ey o Finland.
bPi inen e al. (2012).
cWea he s a ion “Puolanka Paljakka” da a 2009–2017.
dFMI P ecipi a ion s a ion “Suonenjoki Iis esi” da a 2015–2016, empe a u e s a ion measu ed a he si e 1999–2010.
eMean ai empe a u e a wea he s a ion “Some o Salkola”, 2011–2017; p ecipi a ion sum 1981–2010
(h p://www.me la. i/me in o/ o es -condi ion/p og ammes/in ensi e-moni o ing.h m).
Replaces MC117 11 Feb ua y 2016 onwa ds.
gRe-posi ioned and e ocused Augus 2016 a e damage in June 2016.
hSho pe iod a ailable om 2014 be o e came a damage.
2.2.1 Hyy iälä
The Hyy iälä si e is si ua ed a he Uni e si y o Helsinki’s
Hyy iälä Fo es y Field S a ion in an e en-aged Sco s pine
(Pinus syl es is) s and wi h sca e ed sp uce and deciduous
ees. One came a is ocused on he pine canopy wi h one
bi ch c own in iew. The second came a moni o s he o -
es loo which is domina ed by dwa sh ubs and ea he
mosses. The came a obse a ions a e suppo ed by a wide
selec ion o con inuous measu emen s o me eo ology, gas
Ea h Sys . Sci. Da a, 10, 173–184, 2018 www.ea h-sys -sci-da a.ne /10/173/2018/
M. Pel oniemi e al.: Webcam image da abase o phenological s udies in Finland 177
Figu e 2. Came a iews: MC100 (a), MC101 (b), MC102 (c),
MC103 (d), MC104 (e), MC105 ( ), MC106 (g), MC107 (h),
MC108 (i), MC109 (j), MC110 (k), MC111 (l), MC112 (m),
MC113 (n), MC114 (o), MC115 (p), MC116 (q), MC117 ( ),
MC118 (s), MC119 ( ), MC120 (u), MC121 ( ), MC122 (w),
MC123 (x), MC124 (y), MC125 (z), MC126 (aa), MC127 (ab),
MC128 (ac), MC129 (ad), MC130 (ae). Abb e ia ions e e o
came a ID in Table 1.
exchange, ee ecophysiology and soil a he SMEAR II
esea ch s a ion (Ha i and Kulmala, 2005; h p://www.a m.
helsinki. i/SMEAR). These da a a e a ailable a AVAA po -
al (h ps://a aa. da a. i/web/sma /smea /).
2.2.2 Kaamanen
The Kaamanen we land is an open meso ophic en in no h-
e n Finland, wi h su ace pa e n consis ing o we la ks and
d ie s ings. Mos o he ime he la ks a e inunda ed, bu he
s ings wi h a heigh o app oxima ely 0.8 m a e cons an ly
abo e he wa e able co e ing abou 40 % o he en su ace.
The si e has no pe ma os , bu hin lenses o ice may emain
in he well-insula ed s ings un il la e summe . The la ks a e
co e ed by di e en sedges (Ca ex spp.) and moss species,
while he highe s ings a e domina ed by a ious sh ubs,
such as Ledum palus e,Empe um nig um,Rubus chamae-
mo us and Be ula nana. The CO2and CH4 luxes be ween
he en and he a mosphe e, as well as a a ie y o en i on-
men al pa ame e s (ai and soil me eo ology oge he wi h
di e en adia ion componen s), a e con inuously measu ed
a he si e (Au ela e al., 2009).
2.2.3 Ken ä o a
The Ken ä o a si e is si ua ed in a No way-sp uce-
domina ed o es a Pallas wi h a small numbe o Be-
ula pubescen also p esen , and is ope a ed in conjunc ion
wi h he Pallas–Sodankylä Global A mosphe e Wa ch s a-
ion. The main species o he g ound loo a e Vaccinium my -
illus, Empe um nig um,Vaccinium i is-idaea and he o -
es mosses Pleu ozium sch ebe i,Hylocomium splendens and
Dic anum polyse um. Ca bon dioxide luxes oge he wi h an
ex ensi e se o me eo ological pa ame e s a e measu ed a
he si e. Ken ä o a is an associa ed si e wi hin he ICOS lux
ne wo k.
2.2.4 Lammi
The Lammi si e is si ua ed a Uni e si y o Helsinki’s Lammi
Biological S a ion (LBS) which belongs o he Finnish Long-
Te m Socio-Ecological Resea ch ne wo k (FinLTSER). The
s a ion is loca ed in a egion cha ac e ized by bo eal o es ,
lake and ag icul u al landscapes. Th ee came as a e p esen
(g ound, c own and landscape) including deciduous species
Be ula pubescens,Ace pla anoides,Ulmus glab a,P unus
padus,Tilia co da a and Ribes alpinum. The FMI wea he
s a ion “Hämeenlinna Lammi Pappila” is loca ed a LBS.
LBS also main ains a pho osyn he ically ac i e adia ion sen-
so (da a a ailable upon eques om LBS).
2.2.5 Lompolojänkkä
The Lompolojänkkä si e is loca ed on an open, meso ophic
sedge en. The ield laye ege a ion in he we e pa s o he
en is domina ed by sedges, while d ie pa s a e cha ac e -
ized by ai ly dense s ands o Be ula nana. Low sh ubs can
be ound ac oss he en wi h ela i ely low co e ages. Fo a
mo e de ailed desc ip ion o Lompolojänkkä, see Au ela e
al. (2009) and Lohila e al. (2010). The came a has a gene al
iew on he d ie pa o he en. The CO2and CH4 luxes
oge he wi h a se o en i onmen al pa ame e s (ai and soil
me eo ology oge he wi h di e en adia ion componen s)
a e con inuously measu ed a he si e. Ken ä o a is a Class 2
si e wi hin he ICOS lux ne wo k.
2.2.6 Paljakka
Paljakka is a phenology moni o ing si e in cen al Finland
belonging o he phenological moni o ing ne wo k ope a -
ing in Finland (Poikolainen e al., 1996; Pudas e al., 2008),
www.ea h-sys -sci-da a.ne /10/173/2018/ Ea h Sys . Sci. Da a, 10, 173–184, 2018

178 M. Pel oniemi e al.: Webcam image da abase o phenological s udies in Finland
whe e seasonal de elopmen o main ee species in Finland
is moni o ed. The came a is loca ed on he s a ion oo , and
is ocused on a No way sp uce s and. Image iew includes
small bi ch ees ha a e moni o ed o phenology. The FMI
wea he s a ion “Puolanka Paljakka” is loca ed in he imme-
dia e icini y o he came a.
2.2.7 Pa kano
Paljakka is a phenology moni o ing si e in sou he n Finland,
belonging o he same ne wo k o si es as Paljakka si e. The
came a is loca ed a he s a ion oo , and is ocused on a
mixed o es landscape including ees ha a e moni o ed o
phenology, and a lake behind he ees. The FMI wea he s a-
ion (“Ka ia Alkkia”) is wi hin app oxima ely 20 km o he
came a.
2.2.8 Punkaha ju
The Punkaha ju si e is a ma u e sp uce (Picea abies)
s and on mesic soil. I hos s h ee came as a di e en
heigh s, one aking images o he g ound, one a he c own
le el, and one aking images o he su ounding landscape.
The nea es FMI wea he s a ion “Sa onlinna Punkaha ju
Laukansaa i” is wi hin 3 km. The Punkaha ju si e belongs
o he ICP Fo es s (In e na ional Coope a i e P og amme
on he Assessmen and Moni o ing o Ai Pollu ion E ec s
on Fo es s) le el II moni o ing si e o long- e m ecosys-
em moni o ing. A di e se se o ecosys em moni o ing is
conduc ed a he si e; see ICP Fo es s manual o de ails
and moni o ing in o ma ion collec ed (h p://icp- o es s.ne /
page/icp- o es s-manual; Me ilä e al., 2013).
2.2.9 Sodankylä o es
The Sco s pine (Pinus syl es is) o es si e in Sodankylä is
si ua ed wi hin he A c ic Resea ch Cen e o he FMI. The
spa se g ound ege a ion consis s o lichens (73 %), mosses
(12 %) and e icaceous sh ubs (15 %). The e a e h ee cam-
e as a his si e, wi h iews on he o es canopy, c own and
he g ound. Ca bon dioxide luxes oge he wi h an ex ensi e
se o me eo ological pa ame e s a e measu ed a he si e.
The Sodankylä o es is a Class 1 si e wi hin he ICOS lux
ne wo k.
2.2.10 Sodankylä we land
The Sodankylä we land (Halssiaapa) is an open en in he
icini y o he Sodankylä o es lux si e. I is domina ed by
la ge, eeless la ks wi h abundan sedge ege a ion, com-
plemen ed by ai ly low and na ow idges wi h bi ch ees
(o 5–7 m in heigh ). CO2and CH4 luxes as well as basic
me eo ological pa ame e s a e measu ed a he si e.
2.2.11 Suonenjoki
The Suonenjoki si e has a Sco s pine (P. syl es is) s and
on sub-xe ic soil. The e is a wea he s a ion main ained by
he Na u al Resou ces Ins i u e Finland a he si e (da a on
eques om he co esponding au ho ). The nea es FMI
wea he s a ion “Suonenjoki Iis esi” is wi hin 3 km dis ance.
No o he moni o ing ac ions a e conduc ed a his si e.
2.2.12 Tammela
The si e p esen ly hos s wo came as, one aking images
o he g ound and ano he aking images o he su ound-
ing landscape. The Tammela si e has a ma u e sp uce (P.
abies) s and on mesic soil. I belongs o he ne wo k o
ICP Fo es s le el II si es. Consul ICP II Fo es s manual o
pe manen moni o ing da a collec ed (h p://icp- o es s.ne /
page/icp- o es s-manual). The FMI wea he s a ion “Some o
Salkola” is also a he si e.
2.2.13 T ä minne
The T ä minne si e belongs o he Finnish Long-Te m Socio-
Ecological Resea ch ne wo k (FinLTSER) and is loca ed a
he wes e n Gul o Finland a he Uni e si y o Helsinki’s
T ä minne Zoological S a ion. The a ea was pos -glacially
upli ed abo e he sea le el and is cha ac e ized by a mix o
bed ock and clay soil ma e ial and he b- ich o es (nemo al)
o es . One came a is ins alled on he oo o he s a ion’s
main building iewing he landscape a he sho eline o he
a chipelago (Fig. 2ae). The came a allows he moni o ing o
he ee phenology, he seasonal de elopmen o eeds and
he e olu ion o sea ice. The ee species in he came a’s ield
o iew include elm (Ulmus glab a), maple (A. pla anoides),
downy bi ch (B. pubescens), ash (F axinus excelsio ), com-
mon alde (Alnus glu inosa) and Sco s pine (P. syl es is L.).
The FMI wea he s a ion “Hanko-T ä minne” is loca ed a
he si e and p o ides measu emen s o ai empe a u e, p e-
cipi a ion, humidi y and snow dep h.
2.2.14 Vä iö
The Vä iö si e is loca ed in Salla, eas e n Lapland, a he
no he n alpine imbe line on he summi pla eau o a hill.
The ela i ely open o es s and is domina ed by Sco s pine
(P. syl es is) o a ious age wi h occasional moun ain bi ch
(Be ula pubescens a . punilla) in he unde s o ey. The o -
es loo ege a ion consis s o a a ie y o mosses, lichens
and dwa sh ubs. The si e accommoda es h ee came as o
aking images o he landscape, he pine c owns and he o -
es loo . The came as a e pa o SMEAR I esea ch s a-
ion (Ha i e al., 1994; h p://www.a m.helsinki. i/SMEAR)
whe e con inuous measu emen s o me eo ology, gas ex-
change, ee ecophysiology and soil a e pe o med. The da a
a e a ailable a AVAA po al (h ps://a aa. da a. i/web/sma /
smea /).
Ea h Sys . Sci. Da a, 10, 173–184, 2018 www.ea h-sys -sci-da a.ne /10/173/2018/
M. Pel oniemi e al.: Webcam image da abase o phenological s udies in Finland 179
3 Examples o p ocessed image da a
He e we show examples o he da a collec ed wi hin ou ne -
wo k, ocusing on he e ec o sys ema ic change in illu-
mina ion condi ions du ing he annual cycle o sola ele a-
ion known o in luence colou spec a o incoming adia-
ion, which po en ially a ec s he calcula ed colou a ios in
images. The analyses we e done o i e egions o in e es
(ROI) o came a iews a wo con as ing si es sepa a ed by
10◦la i ude (Fig. 3).
Mean g een ac ion o egions o in e es ( ege a ion) in
images has been ound o a y by illumina ion condi ions
(Ah ends e al., 2009). I has been epo ed ha he sys em-
a ic changes o image-based colou indices du ing he season
can be la gely omi ed by ocusing he analyses on midday
images (Ah ends e al., 2009), possibly by agg ega ing in-
o ma ion om nea by days (Sonnen ag e al., 2012), and by
selec ing a obus es ima o ha is agg ega ed, e.g. median.
Colou changes in deciduous ege a ion and snow co e o e
he season a e dis inc i e enough o be clea ly depic ed be-
hind he day- o-day a ia ion due o clouds and i adia ion
(Pel oniemi e al., 2018; A slan e al., 2017), bu in we lands
and coni e ous ees hese di e ences can be mo e obscu e.
We calcula ed mean g een ch oma ic coo dina es
(GCC =G/(R +G+B), o pixels in he egion o in e -
es (ROI). Indi idual pixels we e accep ed o he mean
calcula ion i exposu e was su icien – i.e. pixels we e no
poo ly exposed o o e exposed (Sonnen ag e al., 2012).
We equi ed ha pixel digi al numbe s o R, G and B we e
all in he ange [30, 254]. Daily images we e classi ied o
classes 1, 2 and 3, based on Sun ele a ion angles alling
in o ca ego ies > 30◦[20, 30◦), (0, 20◦), espec i ely. Daily
medians o GCC in hese g oups we e calcula ed. Sun
ele a ion angles we e calcula ed using coo dina es and ime
o day. In addi ion o hese Sun ele a ion class medians, we
also calcula ed daily median om midday images be ween
10:00 and 14:00 UTC +2. Daily ime se ies o GCC we e
subsequen ly smoo hed wi h loess eg ession so as o p esen
in e pola ed daily es ima es because he numbe o he
images o median calcula ion was small.
As Kaamanen is a no he n si e wi h no dayligh du ing
he win e ime, he e is a long gap in image da a (Fig. 4).
The o e all end in he phenological de elopmen o he
bi ch (B. pubescens) and o he we land ege a ion (mosses,
sedges, sh ubs) is e y simila among he Sun ele a ion ca -
ego ies. The di e ences be ween hese classes a e clea ly
smalle han any changes ela ed o phenological g eening
up o mosses and o he we land ege a ion and e en mo e
limi ed han he inc ease o GCC due o budbu s o dwa
bi ches (mid-May o ea ly July). The e ec o snowmel is
seen as a quick decline o GCC in Bi ch and We land GCC
in ea ly May o 2015 and 2016. A e e ence panel shows a i-
able esponse du ing he ime pe iod o No embe o Ap il.
The panel was no snow co e ed in Ap il (inspec ed isually
om images), which implies ha he a iable esponse o he
Figu e 3. Regions o in e es used in GCC ime se ies cons uc-
ion. (a) Kaamanen we land si e (came a MC128), wi h Sphagnum
spp and g asses (We land ROI), sh ubs like pubescen bi ch (Be ula
pubescens subsp. Cze epano ii) (bi ch), and e e ence g ey pla e.
(b) Pa kano si e (came a MC112), wi h sil e bi ches (Be ula pen-
dula) ROI dis ibu ed in h ee sepa a ed polygons (bi ch), and No -
way sp uce (Picea abies L. Ka s ) ROI.
panel’s colou index may e lec he p o ound change o he
ligh en i onmen due o snow mel o he su oundings and
mul iple e lec ions.
Loca ed in sou he n Finland, Pa kano has gaps in he da a
du ing he win e only in he classes 1 and 2 (Fig. 5) be-
cause o highe Sun ele a ions du ing he win e . Again all
ca ego ies o Sun ele a ion gi e simila o e all de elopmen
o GCC, including midday median. The g eenness change
om win e o summe is nea ly as much o he plan ed
non-na i e Se bian sp uce (P. omo ika) as i is o bi ch (B.
pendula). Howe e , GCC es ima ed o sp uce by Sun ele a-
ion class shows ac oss class di e ences and lowe GCC o
highe Sun ele a ion angles, which could be ela ed o he
da ke su ace o sp uces oliage and how colou a ios a e
ep oduced a low digi al numbe s in he shaded pa s o he
canopy. Canopy geome y and came a o ien a ion a ec how
he shaded pa s a e de ec ed by he came a, which sugges s
www.ea h-sys -sci-da a.ne /10/173/2018/ Ea h Sys . Sci. Da a, 10, 173–184, 2018
180 M. Pel oniemi e al.: Webcam image da abase o phenological s udies in Finland
Bi ch
We land
Re e ence
03 04 05 06 07 08 09 10 11 12 01 02 03 04 05 06 07 08 09 10 11 12
0.31
0.33
0.35
0.37
0.39
0.31
0.33
0.35
0.37
0.39
0.31
0.33
0.35
0.37
0.39
Mon h, i s day (yea s 2015–2016)
GCC
Sun angle ●● ●● ●● ●●
Class 1 Class 2 Class 3 Midday Sun angle Class 1 Class 2 Class 3
Figu e 4. Daily GCC ime se ies o ROI in Kaamanen si e by sola ele a ion ca ego y – class 1: > 30◦; class 2: (20, 30◦); class 3: (0,
20◦) – and as day ime a e ages (midday: 10:00–14:00 UTC +2) and smoo hed de elopmen o GCC o e he yea s. Ve ical line shows he
snowmel da e, i.e. he i s day wi hou any snow.
ha di e ences in GCC due o Sun angle could be sensi i e
o hese con igu a ions. O e all, ou esul s con i m he ea -
lie conclusions ecommending he use o midday images o
daily agg ega ion and il e ing o phenological ime se ies by
se e al au ho s (Sonnen ag e al., 2012; Winga e e al., 2015;
Filippa e al., 2016).
4 Da a a ailabili y
To dis ibu e images, we es ablished a communi y (Pheno-
logical ime lapse images and da a om Monime EU Li e+
p ojec (LIFE12 ENV/FI/000409), h ps://www.zenodo.o g/
communi ies/phenology_came a/) in Zenodo eposi o y se -
ice (h ps://www.zenodo.o g/) main ained by CERN. The
se ice was es ablished in he EU OpenAIRE p ojec (h ps:
//www.openai e.eu/) and is mean o pe manen a chi ing
and dis ibu ion o esea ch ma e ials.
The images om each came a co e s he pe iod be-
ween 2014 and 2016 o pa s o i (Fig. 6). Images
a e in hei o iginal quali y, and ha e no been qual-
i y checked. Only known majo s a us issues in luencing
analyses a e epo ed in he associa ed came a da ashee
(h ps://doi.o g/10.5281/zenodo.1066862; Pel oniemi e al.,
2017). Use s a e welcomed o epo new s a us issues wi h
he images o hei con ac pe sons (see came a da ashee ).
Tempo al co e age o images a ies by si e and came a, de-
pending on he moun ing da e, and da a connec ion a iabil-
i y a he si es, which occasionally e mina es hal -hou ly
submissions o he images o he FTP se e in emo e lo-
ca ions when came as a e on a mobile phone ne wo k.
Images a e o ganized in came a-speci ic collec ions a ail-
able in ou Zenodo communi y, each ha ing hei unique
DOI-numbe . Collec ions include images om each cam-
e a om 2014 o 2016 packed in o a zip ile, named
a e came a ID and yea s o co e age. DOI o image
zips a e lis ed in Table 1. Image iles names include
he in o ma ion abou he loca ion, a ge , da e and cap-
u e ime o he image and ollow he o ma <loca-
ion>_< a ge >_<YYYYMMDD>_<HHMMSS>.jpg. I he
Ea h Sys . Sci. Da a, 10, 173–184, 2018 www.ea h-sys -sci-da a.ne /10/173/2018/
M. Pel oniemi e al.: Webcam image da abase o phenological s udies in Finland 181
Sp uce
Bi ch
03 04 05 06 07 08 09 10 11 12 01 02 03 04 05 06 07 08 09 10 11 12
0.32
0.36
0.40
0.32
0.36
0.40
Mon h, i s day (yea s 2015–2016)
GCC
Sun angle ●● ●● ●● ●●
Class 1 Class 2 Class 3 Midday Sun angle Class 1 Class 2 Class 3
Figu e 5. Daily GCC ime se ies o ROI in Pa kano si e by sola ele a ion ca ego y – class 1: > 30◦; class 2: (20, 30◦); class 3: (0, 20◦) – and
as day ime a e ages (midday: 10:00–14:00 UTC +2) and smoo hed de elopmen o GCC o e he yea s. Ve ical line shows he snowmel
da e, i.e. he i s day wi hou any snow. In 2015 snow had mel ed be o e he came a was ins alled.
image is cap u ed in IR o ma i is indica ed by an addi ional
ag <_IR> be o e he da e and ime.
Came a-speci ic in o ma ion is a ailable in he came a
da ashee also s o ed in Zenodo. Fo he p esen submission
o 2014–2016 images o came as in he Table 1, see he
came a da ashee (h ps://doi.o g/10.5281/zenodo.1066862).
We plan o upda e he collec ion o images in he u u e, by
adding new e sions o he image da ase s. Fo newe images
no ye uploaded o Zenodo, please con ac came a con ac
pe sons in he da ashee , and me ada a o p esen came a im-
age uploads.
Ma e ial in he da abase is published unde C ea i e Com-
mons A ibu e 4.0 license.
5 Conclusions
Digi al came as ha e shown hei alue in a ious phenolog-
ical applica ions. They ga he aluable in o ma ion, which
can be analysed o scien i ic pu poses, pa icula ly o hose
add essing changes in ime. He e, we ha e p esen ed a cam-
e a ne wo k ope a ing in Finland, wi h came as moun ed in
si es anging om hemibo eal o suba c ic egions. So a , we
ha e used he came as in he analyses o la i udinal ends o
bi ch phenology (Pel oniemi e al., 2018), snow co e (A -
slan e al., 2017), and analyses o Sco s pine and we land
CO2 luxes (Linkosalmi e al., 2016) wi h encou aging e-
sul s. Se e al possibili ies o u he esea ch exis , pa icu-
la ly associa ed wi h empo al changes o colou s in coni e
canopies, and changes o snow co e .
So a , mos ly simple analyses based on colou s ha e
been used in phenological and snow analyses, and we ind
hem use ul in acking seasonal phenomena, pa icula ly
as hey ha e been shown o be obus o ligh condi ions
he e and elsewhe e (Ah ends e al., 2009). Some aspec s
o colou analysis equi e u he esea ch, including he
ole o ligh sca e ing om snow and associa ed ligh en-
i onmen change, which seems o complica e analyses o
colou a io changes o ege a ion a ge s. Some o hese
challenges could be o e come by combining adia ion ans-
e app oaches o image analyses (Winga e e al., 2015).
www.ea h-sys -sci-da a.ne /10/173/2018/ Ea h Sys . Sci. Da a, 10, 173–184, 2018