Multiscale mapping of plant functional groups and plant traits in the High Arctic using field spectroscopy, UAV imagery and Sentinel-2A data

Abstract

The Arctic is warming twice as fast as the rest of the planet, leading to rapid changes in species composition and plant functional trait variation. Landscape-level maps of vegetation composition and trait distributions are required to expand spatially-limited plot studies, overcome sampling biases associated with the most accessible research areas, and create baselines from which to monitor environmental change. Unmanned aerial vehicles (UAVs) have emerged as a low-cost method to generate high-resolution imagery and bridge the gap between fine-scale field studies and lower resolution satellite analyses. Here we used field spectroscopy data (400-2500 nm) and UAV multispectral imagery to test spectral methods of species identification and plant water and chemistry retrieval near Longyearbyen, Svalbard. Using the field spectroscopy data and Random Forest analysis, we were able to distinguish eight common High Arctic plant tundra species with 74% accuracy. Using partial least squares regression (PLSR), we were able to predict corresponding water, nitrogen, phosphorus and C:N values (r 2 = 0.61-0.88, RMSEmean = 12%-64%). We developed analogous models using UAV imagery (five bands: Blue, Green, Red, Red Edge and Near-Infrared) and scaled up the results across a 450 m long nutrient gradient located underneath a seabird colony. At the UAV level, we were able to map three plant functional groups (mosses, graminoids and dwarf shrubs) at 72% accuracy and generate maps of plant chemistry. Our maps show a clear marine-derived fertility gradient, mediated by geomorphology. We used the UAV results to explore two methods of upscaling plant water content to the wider landscape using Sentinel-2A imagery. Our results are pertinent for high resolution, low-cost mapping of the Arctic.

Full text

Environ. Res. Lett. 16 (2021) 055006 https://doi.org/10.1088/1748-9326/abf464
OPEN ACCESS
RECEIVED
28 October 2020
REVISED
25 March 2021
ACC EPT ED FOR PUB LICATI ON
1 April 2021
PUBLISHED
21 April 2021
Original content from
this work may be used
under the terms of the
Creative Commons
Attribution 4.0 licence.
Any further distribution
of this work must
maintain attribution to
the author(s) and the title
of the work, journal
citation and DOI.
LETTER
Multiscale mapping of plant functional groups and plant traits in
the High Arctic using eld spectroscopy, UAV imagery and
Sentinel-2A data
Eleanor R Thomson1,∗, Marcus P Spiegel1, Inge H J Althuizen2, Polly Bass3, Shuli Chen4,
Adam Chmurzynski4, Aud H Halbritter2, Jonathan J Henn5, Ingibjörg S Jónsdóttir6,11,
Kari Klanderud7, Yaoqi Li4,8, Brian S Maitner4, Sean T Michaletz9, Pekka Niittynen10,
Ruben E Roos7, Richard J Telford2, Brian J Enquist4, Vigdis Vandvik2, Marc Macias-Fauria1
and Yadvinder Malhi1
1School of Geography and the Environment, University of Oxford, Oxford, United Kingdom
2Department of Biological Sciences and Bjerknes Centre for Climate Research, University of Bergen, Bergen, Norway
3Matanuska-Susitna College, University of Alaska, Anchorage, AK, United States of America
4Department of Ecology and Evolutionary Biology, University of Arizona, Tucson, AZ, United States of America
5Department of Integrative Biology, University of Wisconsin-Madison, Madison, WI, United States of America
6Faculty of Life and Environmental Sciences, University of Iceland, Reykjavik, Iceland
7Faculty of Environmental Sciences and Natural Resource Management, Norwegian University of Life Sciences, Aas Norway
8Institute of Ecology and Key Laboratory for Earth Surface Processes of the Ministry of Education, College of Urban and
Environmental Sciences, Peking University, Beijing, People’s Republic of China
9Department of Botany and Biodiversity Research Centre, University of British Columbia, Vancouver, Canada
10 Department of Geosciences and Geography, University of Helsinki, Helsinki, Finland
11 Arctic Biology, The University Centre in Svalbard, Longyearbyen, Norway
∗Author to whom any correspondence should be addressed.
E-mail: eleanor.thomson@oriel.ox.ac.uk
Keywords: Svalbard, functional trait, tundra, moss, shrubs, remote sensing, bird cliff
Supplementary material for this article is available online
Abstract
The Arctic is warming twice as fast as the rest of the planet, leading to rapid changes in species
composition and plant functional trait variation. Landscape-level maps of vegetation composition
and trait distributions are required to expand spatially-limited plot studies, overcome sampling
biases associated with the most accessible research areas, and create baselines from which to
monitor environmental change. Unmanned aerial vehicles (UAVs) have emerged as a low-cost
method to generate high-resolution imagery and bridge the gap between fine-scale field studies
and lower resolution satellite analyses. Here we used field spectroscopy data (400–2500 nm) and
UAV multispectral imagery to test spectral methods of species identification and plant water and
chemistry retrieval near Longyearbyen, Svalbard. Using the field spectroscopy data and Random
Forest analysis, we were able to distinguish eight common High Arctic plant tundra species with
74% accuracy. Using partial least squares regression (PLSR), we were able to predict corresponding
water, nitrogen, phosphorus and C:N values (r2=0.61–0.88, RMSEmean =12%–64%). We
developed analogous models using UAV imagery (five bands: Blue, Green, Red, Red Edge and
Near-Infrared) and scaled up the results across a 450 m long nutrient gradient located underneath
a seabird colony. At the UAV level, we were able to map three plant functional groups (mosses,
graminoids and dwarf shrubs) at 72% accuracy and generate maps of plant chemistry. Our maps
show a clear marine-derived fertility gradient, mediated by geomorphology. We used the UAV
results to explore two methods of upscaling plant water content to the wider landscape using
Sentinel-2A imagery. Our results are pertinent for high resolution, low-cost mapping of the Arctic.
© 2021 The Author(s). Published by IOP Publishing Ltd

Environ. Res. Lett. 16 (2021) 055006 E R Thomson et al
1. Introduction
The Arctic is the fastest warming region on earth
[1]. Air temperatures have risen at twice the global
rate [2], driving changes to the structure and func-
tioning of tundra ecosystems [3–7]. Increased tem-
peratures lead to shifts in vegetation cover and com-
position and accelerate belowground nutrient cycling
and mineralization rates, with potentially important
effects on global climate [8–14].
Plant traits are a primary control on the dis-
tribution and functioning of plants and underpin
vegetation–climate responses (e.g. [15–17]). Traits
related to the uptake and allocation of resources,
such as leaf mass per area, leaf water content, leaf
nitrogen content (N) and leaf phosphorus content
(P) can affect growth rates, plant longevity, primary
productivity, decomposition rates and biogeochem-
ical cycling [18–22]. Simultaneously, morphology-
related traits, such as leaf area and plant height,
can influence aboveground biomass, surface albedo
and snow dynamics [14,23,24]. Establishing spa-
tial environmental–trait relationships could provide
a quantitative basis to forecast the effects of cli-
mate change in the Arctic [25,26]. Yet obser-
vational plot-studies often fail to isolate environ-
mental drivers or take place at a smaller scale than
the community/landscape they represent. Hence, the
biotic consequences of warming in the coldest envir-
onments remain poorly predicted [27].
When variations in vegetation characteristics
affect spectral reflectance, these changes can be meas-
ured using optical remote sensing techniques. In
comparison to traditional field-based methods, using
remote sensing to generate vegetation data is less
laborious, less invasive, more cost-effective and spa-
tially continuous. Using multispectral satellite-based
sensors, the distribution of Arctic vegetation has been
well-established at broad scales through datasets such
as the Circumpolar Arctic Vegetation Map (CAVM
[28]), which classifies 16 Arctic vegetation types at
1 km resolution. Using coarse-resolution satellite
imagery, widespread vegetation change at high latit-
udes has also been well-documented, including pan-
Arctic ‘greening’ and ‘browning’ trends [29–32], with
important implications for albedo, active layer depth,
permafrost dynamics, carbon cycling and wildlife
[10,33–41]. However, vegetation cover in the Arctic
is extremely heterogeneous, varying at very fine spa-
tial scales ([42–44]), and observations do not always
agree at the satellite and plot level [32,45,46]. Thus,
while remote sensing offers enormous potential to
generate large-scale vegetation data across the Arctic,
such coarse spatial and spectral resolution is limited
in its ability to capture the fine-scale dynamics of tun-
dra plants or resolve the key drivers of observed veget-
ation trends [32,47,48].
In lower-latitude environments, high-
resolution field spectroscopy (visible-near infrared
(VNIR: 400–1100)) or visible-short-wave infrared
(VSWIR: 400–2500 nm) is a well-established tech-
nique for gathering plant taxonomic and trait data
with a high degree of accuracy [49–52]. Specific
spectral reflectance and absorption features have
been directly linked to concentrations of cellulose,
lignin, chlorophyll, nutrients and water (see [53]
for a comprehensive synthesis). These plant traits
can be combined to create unique chemical ‘finger-
prints’ to distinguish individual genera or species
[54–57]. Compared to multispectral remote sensing,
field spectroscopy provides numerous spectral bands
that can capture fine-scale spectral differences and
improve discrimination of vegetation features. Spec-
troscopy studies are rarely extended to the tundra
however, likely due to the challenging logistics and
the fact that field spectrometers are not designed to
accommodate the small size of tundra plant leaves.
Spectroscopy studies carried out in the Arctic have
mainly focused on classifying tundra communities
[58–62], vegetation cover fraction [63], biophysical
plant traits, such as vegetation height and biomass
[61,64], and leaf chlorophyll content or ‘greenness’
[61,65], but not individual species or plant chem-
ical traits. Remote sensing of high-latitude vegetation
would benefit from more spectroscopy studies, which
can link plant characteristics to spectral features at
specific wavelengths, as well as establish remote sens-
ing baselines (i.e. if a vegetation characteristic cannot
be determined using field spectroscopy, it is unlikely
to be distinguished using coarser spatial and spectral
resolution imagery).
By itself however, spectroscopy cannot gener-
ate the spatial and temporal data required to map
and monitor biotic change in the Arctic. In order
to bridge this ‘scale-gap’, unmanned aerial vehicles
(UAVs) have emerged as platforms to upscale field
data and provide nuance to satellite observations
[47,48,66]. Due to the miniaturization of tech-
nology and decreasing costs, UAV-mounted sensors
can facilitate the mapping and monitoring of vegeta-
tion at previously unachievable spatial, spectral and
temporal resolutions. Using hyper and multispec-
tral imagery, UAVs have been used to map plant
communities [67,68], individual species [69–71],
plant traits [72–74] and plant health [69,75–77].
While some UAV studies have applied structure-
from-motion methods to analyse biophysical plant
characteristics in the Arctic (e.g. [78,79]), or applied
spectral techniques to classify high latitude plant
communities (e.g. [59,67,68,80]), to the best of our
knowledge, this is the first study to use field spectro-
scopy and UAVs to identify high Arctic plant species
and map vegetation chemistry across the tundra.
In this study, we investigate the use of field spec-
troscopy for predicting species identity and retriev-
ing plant water and chemistry concentrations across
a range of common species at three sites found near
Longyearbyen, Svalbard. Based on the demonstrated
2

Environ. Res. Lett. 16 (2021) 055006 E R Thomson et al
feasibility of generating plant taxonomic and trait
data using field spectra, we develop analogous mod-
els using UAV five-band multispectral imagery. Addi-
tionally, we explore methods of upscaling our UAV
results to the wider landscape using Sentinel-2A
imagery. Specifically, we aim to:
(a) develop predictive models of species identity
and biochemical plant traits (water, nitrogen,
phosphorus and C:N concentrations) using field
spectroscopy,
(b) develop predictive models of species identity and
the same plant traits using UAV spectra,
(c) apply the UAV models to UAV imagery to derive
spatially continuous maps of plant species and
their corresponding trait values,
(d) explore two methods of upscaling the UAV mod-
els to the wider landscape using Sentinel-2A
imagery.
2. Methods
2.1. Study sites
The data were collected at three High Arctic
study sites around Longyearbyen on the Svalbard
Archipelago, Norway (figure 1). The sites comprised
two contrasting environmental gradients and an
experimental warming site. The sites are charac-
terized by a dry Arctic climate with a mean aver-
age temperature of −2.6 ◦C and an average max-
imum and minimum monthly temperature of 7.6 ◦C
and −11.4 ◦C (2005–2018). Precipitation is around
190 mm per year (all data obtained from the National
Center of Environmental Information [81]). In all
cases, underlying soils were typical cryosols with a
thin organic layer on top of inorganic sediments [82].
2.1.1. The Birdcliffs (78.14◦N, 15.2◦E)
Located near Bjørndalen, up the slope of Platåfjellet,
the Birdcliffs is a 175 m high, 450 m long nutrient and
elevation gradient, stretching from the base of a Little
Auk (Alle alle) and Black-legged Kittiwake (Rissa tri-
dactyla) colony to sea level at Isfjorden. The top of the
gradient receives high nutrient input in the form of
bird guano and other associated biological material
[83]. The majority of the site is characterised by steep
slopes and large alluvial fans made up of a mix uncon-
solidated material of various sizes. Vegetation is dom-
inated by dry dwarf shrub tundra in topographic-
ally elevated areas, particularly Salix polaris and Dryas
octopetala. Mosses and graminoids dominate in moist
areas, especially around the edge of the alluvial fans
and towards the flatter end of the gradient near the
shore.
2.1.2. The control gradient (78.12◦N, 15.4◦E)
The Control Gradient is an 850 m long colluvial slope
located up the north face of Lindholmhøgda, a spur
at the mouth of the Adventdalen valley. The gradient
runs from 200 m.a.s.l to a reservoir (Isdammen)
at sea level. The area is characterised by dry dwarf
shrub tundra that alternates with ridge communities
of scarce vegetation. Habitats with thin winter snow
cover are dominated by Dryas octopetala and Salix
polaris (Dryas heath), while habitats of intermedi-
ate snow depth are dominated by Cassiope tetragona
(Cassiope heath).
2.1.3. The international tundra experiment (78.18◦N,
15.77◦E)
The International Tundra Experiment (ITEX [84])
is an experimental warming site located in End-
alen. Data were collected outside the warming cham-
bers from three predominant habitats: an exposed
and relatively dry Dryas heath with shallow snow
cover in winter, a mesic Cassiope heath with inter-
mediate winter snow depth and a moist snowbed
community dominated by bryophytes, Salix polaris,
Bistortavivipara and graminoids.
2.2. UAV imagery acquisition
All fieldwork was conducted as part of the
Plant Functional Trait Course 4 (https://
plantfunctionaltraitscourses.w.uib.no/), from 16 July
2018 to 27 July 2018. See figure 2for a diagrammatic
representation of the data collection and analysis.
UAV imaging data were acquired from all three
sites. The data were acquired using a 3DR Solo drone
equipped with a MicaSense RedEdge-MX multispec-
tral camera and MicaSense RedEdge Downwelling
Light Sensor (DLS). The MicaSense RedEdge–MX
camera captures surface reflectance at five narrow
spectral bands: blue (475 nm), green (560 nm), red
(668 nm), Red Edge (717 nm) and NIR (840 nm). The
DLS points upwards and captures illumination con-
ditions, which are embedded in the metadata of each
image. The drone was flown at an altitude of 40–60 m,
resulting in <10 cm resolution imagery. To map the
entirety of the study sites, multiple overlapping flights
were required, which were taken over the course of
one day at each site. Radiometric calibration images
were recorded using a MicaSense reflectance panel
as the calibration target. Ground control points for
georeferencing were taken using the Emlid Reach+
differential GNSS system (Emlid, Hong Kong).
The imagery was processed in Pix4Dmapper
(v.4.3.31, Pix4D, Lausanne, Switzerland) using a
standard structure-from-motion (SfM) workflow.
For each site, images from all flights were processed
in the same project to form a single orthomosaic.
The GCPs were manually identified in the images and
georeferenced using their field-collected RTK–GNSS
coordinates. Across the three sites, this resulted in
RMS errors of 0.10–0.14 m. Radiometric calibration
was performed in Pix4D using the MicaSense reflect-
ance target images and the metadata from the DLS.
3

Environ. Res. Lett. 16 (2021) 055006 E R Thomson et al
Figure 1. (a) Map of Svalbard archipelago. (b) Map of study sites near Longyearbyen, Svalbard. Basemaps are from the Norwegian
Polar Institute. Coordinates in UTM Zone 33 N, WGS84 ellipsoid.
2.3. Vegetation sampling
After the UAV imagery acquisition, 68 20 ×20 cm
single-species turfs were collected from across the
Birdcliffs and ITEX sites. The turfs were selected
to represent the most common plant functional
types identified across all sites: mosses, gramin-
oids and dwarf shrubs (table 1, figure 2). High-
accuracy GNSS coordinates were taken from the loc-
ations of the extracted turfs with an Emlid Reach+
differential GNSS system. Additional vegetation
ground-truthing points were taken from all three sites
(table 2, figure 2). The turfs were cut to a substrate
depth of approximately 5 cm, sealed inside plastic
bags and transported back to the University Centre
in Svalbard (UNIS) for species identification and
analysis.
Some shrub samples collected from the dry and
mesic heaths at ITEX showed signs of tissue degrad-
ation, probably due to drought or frost damage
[85]. These samples were included in the plant trait
analyses, as part of a continuum of trait values,
but excluded from the species classification ana-
lyses. Mapping plant health falls outside the scope of
this study and only healthy shrub communities were
observed over the UAV upscaling areas.
2.3.1. Field spectroscopy measurements
Field spectroscopy measurements of the turf samples
(350–2500 nm) were taken at UNIS using an ASD
Fieldspec Pro with fibre optic cable and contact
probe (Analytical Spectral Devices, Boulder, CO,
USA). Turfs were stored outside and reflectance
4

Environ. Res. Lett. 16 (2021) 055006 E R Thomson et al
Figure 2. Diagrammatic representation of (a) the data collection and (b) the data analysis. Only the Birdcliffs site is shown here.
Further turfs, ground-truth points and UAV imagery were collected at the Control Gradient and ITEX site.
measurements were taken within 24 h of turf cut-
ting. Spectroscopy measurements were not taken ‘in
situ’ at study sites, due to the non-portable set-up
of the spectrometer. Turfs were selected to be as
homogenous as possible. If multiple plant species
were present across the turf, measurements were only
taken from areas where the main species dominated.
The contact probe was pushed firmly down onto the
turf so all extraneous light was excluded from the
measurement. Five measurements were taken at dif-
ferent locations across each turf. Each measurement
consisted of 40 internally averaged reflectance read-
ings to increase the signal-to-noise ratio. The spec-
trometer was optimised and calibrated for dark cur-
rent and white light after every turf. For all statistical
analyses, the spectral data were trimmed to the 400–
2500 nm range. For some analyses, the five measure-
ments were averaged to form one spectrum per turf.
2.3.2. Plant trait sampling
After the field spectroscopy measurements had been
taken, three 5 ×5 cm vegetation samples were
cut from each turf. For each sample, all vegetation
above the substrate was harvested. The samples were
weighed for fresh mass, dried at 60 ◦C for 48 h
and re-weighed for dry mass. Vegetation water con-
tent was calculated as ((fresh mass −dry mass)/fresh
mass) ×100.
Nutrient analysis of the dried samples was car-
ried out at the University of Arizona. The protocol to
obtain total phosphorus concentration involved using
persulfate oxidation followed by the acid molyb-
date method (APHA, 1992), after which P con-
centration was determined via colorimetric ana-
lysis with a spectrophotometer (ThermoScientific
Genesys20, United States). Carbon, nitrogen, and
their stable isotope ratios were measured at the
Department of Geosciences Environmental Isotope
Laboratory on a continuous-flow gas-ratio mass
spectrometer.
The three samples were averaged to form one
trait value per turf. See table S1 (available online
at stacks.iop.org/ERL/16/055006/mmedia) for aver-
age trait values.
5

Environ. Res. Lett. 16 (2021) 055006 E R Thomson et al
Table 1. Information on turf samples, including site location, plant functional type, genus and species. Turfs measured 20 ×20 cm. Five
independent spectrum readings were taken per turf.
Turf samples
Site Plant functional type Genus Species N. of turf samples N. of Spectra
Birdcliffs Moss Aulacomnium Palustre 3 15
Aulacomnium Turgidum 4 20
Polytrichastrum Alpinum 2 10
Polytrichum Hyperboreum 1 5
Polytrichum Strictum 1 5
Racomitrium Canescens 2 10
Racomitrium Lanuginosum 2 10
Birdcliffs Graminoid Alopecurus Ovatus 7 35
Luzula Confusa 6 30
Luzula Nivalis 1 5
Birdcliffs Dwarf shrub Cassiope Tetragona 2 10
Dryas Octopetala 5 25
Salix Polaris 9 45
ITEX Moss Identity not recorded 5 25
ITEX Graminoid Identity not recorded 15
ITEX Dwarf shrub Cassiope Tetragona 2 10
Dwarf Shrub Dryas Octopetala 3 15
Dwarf Shrub Salix Polaris 3 15
Senesced Shrub Cassiope Tetragona 6 30
Senesced Shrub Dryas Octopetala 3 15
Total: 68 Total: 340
Table 2. Information on ground-truthing vegetation survey,
including site, plant functional type and number of data points.
GNSS locations were taken with an Emlid Reach+differential
GNSS.
Ground-truthing vegetation survey
Site
Plant functional
type N. data points
Birdcliffs Moss 15
Graminoid 23
Dwarf Shrub 24
Control Gradient Moss 9
Graminoid 0
Dwarf shrub 9
ITEX Moss 6
Graminoid 1
Dwarf shrub 30
Total: 117
2.4. Field spectroscopy analyses
2.4.1. Field spectroscopy-species analysis
The turf spectra were divided into eight vegeta-
tion classes (below) representing a mixture of fam-
ilies, genera and species. Some of the moss and
graminoid species were combined into their famil-
ies or genera to increase the class sample size and
because they were assumed to be spectrally sim-
ilar. Mosses were classified into Aulacomnium spp.
(containing Aulacomnium palustre and Aulacom-
nium turgidum); Polytrichaceae spp. (containing
Polytrichum hyperboreum, Polytrichastrum alpinum
and Polytrichastrum strictum); and Racomitrium spp.
(containing Racomitrium canescens and Racomitrium
lanuginosum). The graminoids were classified into
Alopecurus ovatus (a grass) and Luzula spp. (a rush,
containing Luzula confusa and Luzula nivalis). The
dwarf shrubs were separated into Cassiope tetragona,
Dryas octopetala and Salix polaris.
To quantify similarity or differences between the
spectra, agglomerative hierarchical cluster analysis
was performed using the eight classes described above
and one averaged spectrum per turf. Cluster ana-
lysis works by calculating the distance (difference)
between each possible pair of spectra. The two spec-
tra closest to each other are merged and the process
repeats, with the number of clusters reduced by one
each cycle. Spectral distances were calculated using
Euclidean Distance in MATLAB (MathWorks, Natick,
MA, USA).
The eight vegetation classes were predicted using
five spectra per turf and Random Forest classifica-
tion [86]. The five spectra per turf were used as inde-
pendent samples, as the spectral curves were highly
diverse due to the heterogeneous nature of the turfs.
Random Forest is a machine learning classifier that
grows as an ensemble of decision trees [87,88]. It
is useful for classifying hyperspectral data as it is
robust against overfitting and can be used when the
number of predictor variables (e.g. 2150 wavelength
bands) is greater than the number of samples
(e.g. 470 spectra). All the Random Forest analyses
were carried out in MATLAB using the Treebagger
function.
6

Environ. Res. Lett. 16 (2021) 055006 E R Thomson et al
2.4.2. Field spectroscopy-trait analysis
Plant water, nitrogen, phosphorous and C:N values
were predicted using one averaged spectrum per turf
and partial least squares regression (PLSR [89]). The
PLSR method is effective, as it uses the continuous
spectrum as a single measurement, rather than carry-
ing out a band-by-band analysis and reduces a large
predictor matrix (2150 spectral bands) down to a
few relatively uncorrelated factors (known as latent
variables). For the trait predictions, we chose to use
PLSR instead of Random Forest Regression because,
unlike Random Forest Regression, PLSR is able to
extrapolate beyond its calibration data. While Ran-
dom Forest is a strong tool for classification, it cannot
predict values outside of its training range, limiting its
accuracy when scaling up continuous data over wider
areas that may contain trait values outside the range
of the sample data.
PLSR analyses were carried out using the PLSre-
gress function in MATLAB. The PLSR models were
validated using an unseen 30% of the dataset. Due to
the random nature of the 70:30 split, 1000 PLSR iter-
ations were made for each spectra-trait analysis. The
PLSR equations resulting in robust models (r2> the
mean value of the 1000 iterations) were evaluated
using r2for the independent testing (val) dataset and
RMSE as a percentage of the sample mean (as in
[90,91]).
All Random Forest and PLSR analyses were car-
ried out using (a) the full hyperspectral range from
400 to 2500 nm, (b) just the VIS–NIR region from
400 to 1000 nm and (c) downscaled spectra to match
the five bands represented by the MicaSense camera
(blue, green, red, red edge and near-infrared). The
three analyses were carried out in order to act as a
baseline from which the effects of decreased spectral
information and the effects of using an airborne cam-
era could be accurately separated.
2.5. UAV spectral analysis
2.5.1. UAV spectra-species analysis
UAV spectra were extracted from the orthomosaic at
all the points where the turfs and ground-truthed spe-
cies coordinates were taken. The spectra were extrac-
ted using the ‘extract’ function in the Raster pack-
age [92] in R (v.3.6.0, R Core Team, 2019). First, an
NDVI mask was applied to the imagery to mask out
any pixels with an NDVI of <0.5. The threshold of 0.5
was chosen as it represents the lower limit of NDVI
values measured for vegetation in the Arctic [93].
The UAV spectra were extracted using three methods:
simple (only the pixel value the coordinate falls in is
returned), bilinear (the returned value is interpolated
from the values of the four nearest pixels) and buf-
fer (the returned value is averaged from all the pixels
that fall within a specified 3 ×3 pixel buffer area).
Across all analyses, the buffer method demonstrated
the most robust results, hence only the results from
these analyses are shown.
Using the UAV spectra, Random Forest classifica-
tion was used to predict three broad plant functional
groups: mosses, graminoids and dwarf shrubs. These
plant functional groups represented the finest classi-
fication of vegetation that could be achieved using the
UAV data and the Random Forest method of predic-
tion. The Random Forest model that demonstrated
the best result on the training/validation data was
scaled up and applied to each pixel of the Birdcliff
imagery to generate a continuous ecosystem map of
plant functional type. The models were only applied
to the Birdcliff site, as this was where the majority of
the turfs were collected. As such, we only had confid-
ence in the model validation for the species and envir-
onmental conditions found at this site.
2.5.2. UAV spectra-trait analysis
Trait values were predicted using the same UAV spec-
tra and PLSR method described above. The PLSR
model that maximised the r2and minimised the
%RMSE was chosen and applied to each pixel of
the UAV imagery to generate continuous maps of
trait information. Trait values were also compared to
an NDVI map generated from the UAV imagery to
quantify the benefits of using a multispectral machine
learning approach to trait prediction, compared to an
NDVI-based prediction (figure S3).
2.6. Sentinel-2A scaling analysis
Sentinel-2A imagery was used to explore methods of
upscaling the UAV maps to the wider region. Plant
water content was chosen as an example trait, as water
has a strong direct expression in spectra and the Bird-
cliff site was assumed to be a nutrient anomaly in the
landscape.
The Sentinel image used in the analysis was
a Level-2A cloud-free, atmospherically-corrected
image taken on 30th July 2018, one week after the
field-collection. Only vegetation pixels were included
in the analysis, identified using the Sentinel-2 scene
classification layer (SCL). Analyses were carried out
in R (v.3.6.0, R Core Team, 2019) and Google Earth
Engine (https://earthengine.google.org/).
Two methods of upscaling from UAV to satellite
data were tested:
(a) the UAV-generated map of plant water content
(figure 8) was resampled to match Sentinel-2A
resolution (10 m) and gridlines using bilinear
interpolation. Lower-resolution Sentinel bands
were also resampled to 10 m using bilinear inter-
polation. The resampled UAV water values were
used as a calibration and validation dataset for
the Sentinel imagery. Thus, each Sentinel pixel
(n=329) had a single water value as well as a
reflectance value for each relevant Sentinel band
(Band 2 (492 nm), 3 (560 nm), 4 (665 nm),
5 (704 nm), 6 (741 nm), 7 (783 nm), 8 (833 nm),
8A (865 nm), 11 (1614 nm) and 12 (2202 nm)).
7

Environ. Res. Lett. 16 (2021) 055006 E R Thomson et al
Figure 3. Mean (±standard deviation) of turf spectral reflectance (400–2500 nm). Reflectance was measured using an ASD
Fieldspec Pro with fibre optic cable and contact probe. (a) Spectra were grouped into three plant functional types; mosses,
graminoids and dwarf shrubs. The MicaSense RedEdge-MX five spectral bands (blue, green, red, red edge and near–infrared) are
shown in grey for comparison purposes. (b) Mosses were classified into Aulacomnium spp. (A. palustre and A turgidum),
Polytrichaceae spp. (P. hyperboreum, P. alpinum and P. strictum), and Racomitrium spp. (R. canescens and R. lanuginosum).
(c) Graminoids were classified into Alopecurus ovatus and Luzula spp. (L. confusa and L. nivalis). (d) Shrubs were separated into
C. tetragona, D. octopetala and S. polaris.
Using this dataset and the same PLSR method
described above, model coefficients were gener-
ated for each Sentinel band and applied to each
pixel of the Sentinel image.
(b) the coefficients from the PLSR model used to cre-
ate the UAV map of water content were applied to
the Sentinel bands that best matched the MicaSe-
nse bands (Sentinel bands 2,3,4,5 and 8) for each
pixel of the Sentinel image.
3. Results
3.1. Field spectroscopy identification of plant
functional types and species
Average reflectance spectra for three plant functional
types (moss, graminoids and dwarf shrubs) showed
distinct spectral separability, especially in the NIR
(750–1300 nm) and SWIR regions (1500–2500 nm,
figure 3(a)). Moss displayed the lowest reflectance
across the NIR–SWIR region, especially in the water
absorption bands near 1450, 1940 and 2500 nm
[53]. Within each plant functional type, unique spec-
tral signatures were displayed by genera and species
(figures 3(b)–(d)). Shrubs were more spectrally sim-
ilar to each other, but showed some divergence in the
NIR–SWIR. Intraspecific spectral variability was low
across most of the mosses, while Alopecurus ovatus
and Cassiope tetragona displayed the highest spectral
variability, especially across the NIR.
Using cluster analysis to further quantify spec-
tral separability, mosses emerged as a distinct spec-
tral family (figure 4). All moss spectra clustered
together, with some stratification of genera, especially
Aulacomnium and Racomitrium. There was less guild
clustering of graminoids and shrubs, however Salix
polaris emerged as a tight cluster and most Dryas
octopetala spectra clustered together. Despite some
grouping of Luzula, the graminoids did not cluster
together as a functional group. The majority of the
spectral confusion was created by Alopecurus ovatus
and Cassiope tetragona, whose high spectral variab-
ility meant they were commonly paired with other
shrub and graminoid species.
Using Random Forest Classification and the full
hyperspectral range (400–2500 nm), our results show
that eight common Arctic tundra species can be clas-
sified with 74% accuracy (much above the expec-
ted statistical random accuracy value of 12.5%, given
eight classes, figure 5). Reflecting the results of the
cluster analysis, moss species displayed the highest
classification accuracy values (85, 80 and 80%), along
with Salix polaris (80%) and Luzula sp. (80%). Redu-
cing the amount of spectral information led to a
small decrease in accuracy to 69 and 66% for the
8

Environ. Res. Lett. 16 (2021) 055006 E R Thomson et al
Figure 4. Results of spectral agglomerative hierarchical cluster analysis. One average spectrum per turf was used (n=53). Spectra
are labelled according to the eight vegetative classes described in figure 3and coloured according to plant functional type (moss,
graminoid, dwarf shrub).
Figure 5. Percentage of turf spectra correctly classified using Random Forest classification. Five spectra per turf were used
(n=265). The spectra were classified three times: (1) using the full hyperspectral range from 400 to 2500 nm, (2) just the
VIS–NIR region from 400 to 1000 nm and (3) downscaled spectra to match the five bands represented by the MicaSense-MX
camera (blue, green, red, red edge and near-infrared).
400–1000 nm spectral range and downscaled Mica-
Sense bands, respectively. Across the eight classes, the
greatest source of uncertainty came from the classi-
fication of Cassiope tetragona and Dryas octopetala,
which were commonly confused with each other (see
figure S1 for confusion matrices).
3.2. Field spectroscopy identification of plant traits
Using the full reflectance spectra (400–2500 nm),
PLSR results show that plant water content was
estimated with the highest accuracy (r2=0.88,
RMSE =12%), followed by N (r2=0.77,
RMSE =28%), P (r2=0.67, RMSE =64%) and C:N
9

Environ. Res. Lett. 16 (2021) 055006 E R Thomson et al
Figure 6. (a)–(c) Results of predicted versus measured plant trait values (water (%), nitrogen (%), phosphorus (%) and C:N
ratios) using PLSR analysis. (d)–(f) PLSR spectral weightings for each plant trait. Deviation from 0 indicates the parts of the
spectrum that contribute most strongly to prediction. Spectra were analysed three times, using: (1) the full hyperspectral range
from 400 to 2500 nm, (2) the VIS–NIR region from 400 to 1000 nm and (3) downscaled spectra to simulate the five bands
represented by the MicaSense-MX camera (blue, green, red, red edge and near-infrared).
ratios (r2=0.61, RMSE =37%, figure 6(a)). Remov-
ing spectral information from the SWIR range had
the greatest impact on water prediction. Decreased
SWIR spectral information had almost no impact on
N prediction, and little impact on C:N ratio predic-
tion. P estimation showed little change with reduced
SWIR information but was strongly influenced by
fewer VIS–NIR spectral bands (figures 6(b) and
(c)). The spectral weightings showed that all regions
of the spectrum are important to trait prediction
when the full spectrum is used, but the red-edge and
NIR region are particularly important when SWIR
information is lacking (figures 6(d)–(f)).
3.3. UAV spectral mapping of plant functional
types
When using the UAV MicaSense camera, the eight
vegetation classes could not be classified with the
same accuracy as when using the downscaled hyper-
spectral data. Instead, vegetation cover was classified
into three plant functional types; mosses, gramin-
oids and dwarf shrubs (figure 7(b)). Overall classific-
ation accuracies were 72% across the validation data-
set, much above the statistical random accuracy of
33% expected for three classes. Dwarf shrubs were
classified with the highest accuracy (83%), followed
by graminoids (61%) and mosses (57%, figure 7(c)).
Vegetation cover across the Birdcliff site was dom-
inated by dwarf shrubs. Graminoids were shown as
present at the confluence of runoff channels and on
some of the flatter, wetter areas before the fjord.
Mosses showed speckling between the shrubs and
were found in dense communities around the bottom
of the colluvial fans.
3.4. UAV spectral mapping of plant traits
Using the UAV five-band data, average correlations
with water and N were significantly lower (p< 0.001)
than when using the downscaled hyperspectral data
(figure 8). Plant water trait correlations decreased
from r2=0.68 to r2=0.50 and N accuracies
decreased from r2=0.76 to r2=0.50. The excep-
tion was P, where the average correlations between the
UAV spectroscopy and field measurements increased
from r2=0.45 to r2=0.54 compared to the
downscaled hyperspectral data. Scaling up the most
robust models, there was a clear plant water and
nutrient gradient across the site. Plant water and
nutrient values were generally highest at the top
of the nutrient and elevation gradient and declined
downslope. Nutrient concentrations also appeared to
follow drainage channels across the site and were
tightly correlated to each other.
3.5. Sentinel-2A spectral mapping of plant
water content
Scaling from the UAV to the satellite level, calibrat-
ing a Sentinel-specific PLSR model on the UAV plant
water map values (figure 9(a)) and applying the UAV
plant water model directly to the Sentinel imagery
(figure 9(b)) produced similar patterns but a very dif-
ferent magnitude of result. The maps were correl-
ated with each other at r=0.86, with a y-intercept
of 41%. For method A, plant water content values
10

Environ. Res. Lett. 16 (2021) 055006 E R Thomson et al
Figure 7. (a) Annotated RGB image of the Birdcliff site. The image has been rotated to represent decreasing elevation from top to
bottom. (b) Map of moss, graminoid and dwarf shrub distribution across the Birdcliff site. Produced using UAV reflectance
spectra and Random Forest Classification. (c) User and producer accuracy of the model applied to the validation dataset.
ranged between 30% and 80%, while method B over-
estimated water content, with values ranging from
80% upwards.
4. Discussion
Our results show that UAV multispectral data can
be used to map fine-scale vegetation cover (moss,
graminoids and dwarf shrubs) across the High
Arctic and monitor changes in plant water and
nutrient traits. Using field spectroscopy data, we
were able to distinguish eight common tundra spe-
cies at 74% accuracy, suggesting good prospects
for near-future vegetation mapping at the spe-
cies level with the increasing commercialisation of
UAV–hyperspectral systems and next generation of
hyperspectral satellites (CHIME, EnMAP, HISUI,
HypXIM, HyspIRI, PRISMA etc [94–99]). Although
only CHIME, EnMAP and PRISMA cover the high
latitudes (extending to 84◦N, 80◦N and 70◦N
respectively), hyperspectral technology in this area is
expanding.
Mosses in particular displayed a unique spec-
tral signature and distinct spectral separability within
tundra vegetation. Average moss reflectance across
the NIR and SWIR was lower than for graminoids and
shrubs, reflecting their lack of vascular system and
high water content. All three families/genera of moss,
Aulacomnium, Polytrichaceae and Racomitrium, dis-
played high prediction accuracies in the Random
Forest analysis (85, 80% and 80% respectively), which
is promising for the automatic detection of moss taxa
in future hyperspectral studies. In northern environ-
ments, mosses can dominate aboveground primary
productivity and have been shown to display a wide
range of traits and life history strategies [100–103].
Yet, due to the difficulty of identifying species, mosses
are often ignored by ecologists in favour of vascu-
lar plants, or conglomerated into groups that may
or may not be meaningful [104]. Thus, the ability to
distinguish moss taxa at broad scales would facilitate
research focusing on this significant but understudied
plant group.
Graminoids showed similar spectral separab-
ility with a 73% and 80% prediction accuracy
for Alopecurus ovatus and Luzula, respectively. Salix
polaris also demonstrated a high prediction accuracy
of 80%, whereas some spectral confusion was created
between the two evergreen dwarf shrubs, Cassiope and
Dryas, likely due to Cassiope’s woody vertical struc-
ture that created high variation in its spectral signa-
ture (figure 3). Developing spectroscopic techniques
11

Environ. Res. Lett. 16 (2021) 055006 E R Thomson et al
Figure 8. Maps of plant water, nitrogen, phosphorus and C:N ratios across the Birdcliff site. The maps were produced using UAV
spectra and partial least squares regression PLSR). r2_map and RMSE_map represent the PLSR model that was chosen to scale up
the results of the calibration/validation dataset. r2_average and RMSE_average are the average of 1000 PLSR runs.
to improve the identification of dwarf shrub species
would be useful at broad scales as, along with snow
depth, the litter quality of Cassiope, Dryas and Salix
has been linked to differences in soil organic mat-
ter and respiration rates [105]. Cassiope, Dryas and
Salix also display contrasting evergreen and decidu-
ous strategies, which govern different relationships
between soil nutrient availability, photosynthesis and
growth rates [106]. Overall, these results are highly
encouraging with respect to the spectral separability
of tundra vegetation and application of fine-scale
remote sensing in the Arctic.
At the trait level, we found that four plant traits
(water, N, P and C:N values) could be estimated
with varying accuracy using field spectroscopy. Plant
water and N concentrations were strongly correl-
ated with the full range spectra and showed good
agreement with previous PLSR studies at the leaf
and canopy scale (e.g. [91,107,108]). The estim-
ation of P was slightly lower than previous studies
12

Environ. Res. Lett. 16 (2021) 055006 E R Thomson et al
Figure 9. Two methods of mapping vegetation water content (%) around Longyearbyen, Svalbard using the UAV-generated plant
water map in figure 8and Sentinel-2A imagery. Method (a): calibrating a Sentinel-specific PLSR model (ten bands) on the UAV
plant water map values and applying it across the imagery. Method (b): applying the UAV plant water PLSR model (five bands) to
the closest corresponding Sentinel-2 bands. The cloud-free Sentinel-2 image was taken on 30 July 2018.
(e.g. [91,106,109,110]), although the PLSR mod-
els still captured a substantial amount of variation
across the data. Unlike water and N, P has no direct
expression across the 400–2500 nm spectrum. Predic-
tions result from correlations with other traits, espe-
cially nitrogen, attributed to the stoichiometric link
between N and P—a link which has been used to map
leaf P across large areas of the tropics and identify
crop-level P deficiencies ([111–113]). In contrast to
forest or agricultural studies however, N:P ratios may
differ more widely among our samples, through dif-
ferent plant groups having fundamentally different
N:P ratios ([114]) and high nutrient deposition from
bird colonies leading to uneven N:P soil ratios.
When downscaling the hyperspectral data, estim-
ations of plant water content showed a correspond-
ing decrease in accuracy with the number of reduced
spectral bands. The nutrient traits were less sens-
itive to reduced SWIR information, reflecting that
water is expressed in a wide range of spectral bands
13

Environ. Res. Lett. 16 (2021) 055006 E R Thomson et al
across the VNIR–SWIR spectrum [53], whereas N is a
major component of chlorophyll with strong expres-
sion in blue and green VIS bandwidths (430–660 nm
[53,115]). P displayed similar spectral weightings
to N although, when downscaled to the five Mica-
Sense bands, P values were highly sensitive to a
reduction in VNIR information (as similarly found
by [110,116]). In contrast, N showed no signi-
ficant reduction in accuracy across the downscaled
analyses, further suggesting that, in Arctic envir-
onments, estimations of P arise from collinearities
with traits other than N. We have confidence that
the models are detecting trait variation, rather than
species-level differences, given the observed spec-
tral weightings and high intraspecific trait variation
within Arctic plant functional groups (see table S1
and [117]).
Using the multispectral UAV data, we were able
to map moss, graminoids and dwarf shrubs at 72%
accuracy. Forbs were not included in the model, as
the vegetation cover of any individual forb species was
not extensive enough to sample sufficiently. In future
studies, it would be useful to test whether forbs can be
spectrally distinguished from graminoids (as found
by [80] using multispectral satellite imagery, LiDAR
and phenology), or whether their similar functional-
ity render forbs and graminoids too related to be dif-
ferentiated by spectra alone.
Across the Birdcliff site, modelled vegetation dis-
tribution was dominated by shrub cover. High nutri-
ent input from nesting colonies generally leads to the
formation of moss tundra below bird cliffs on Sval-
bard [118,119], but along warmer coastlines, such as
Bjørndalen, vascular plants have been found to pro-
liferate [120]. Although moss only constituted 10%
of the vegetation cover across the site, the ability
to separate moss-dominated pixels from shrubs and
graminoids is a key strength of high-resolution map-
ping over satellite-scale analyses. Regular monitor-
ing of moss communities can act as an early warn-
ing system for climatic shifts and associated biotic
and abiotic effects in polar environments [121–123].
However, it is important to remember that this study
represented a snapshot in time, where moisture con-
ditions, illumination conditions and phenological
phase were unique. Time-series data is required to
investigate whether moss and other vegetation types
can be distinguished with the same accuracy at dif-
ferent phenological stages and under varying abiotic
conditions [124,125].
Using the UAV imagery, we were unable to dis-
tinguish vegetation at the species level. This is des-
pite the downscaled hyperspectral data displaying
species level accuracy of 64%. We attribute this
loss in accuracy to two main reasons. Firstly, UAV
imagery suffers from quality issues to a greater extent
than field-measured spectra. These include bidirec-
tional reflectance distribution function effects, vari-
ability in illumination conditions during a flight,
poor radiometric calibration and spectral mixing,
even at high resolutions [126,127]. These effects
are even more pronounced in the Arctic due to low
sun angle, variable cloud cover and heterogeneous
tundra. While the UAV imagery shows a correla-
tion with the field-collected spectra (figure S2), there
is substantial variation in the agreement, with the
UAV spectra likely exhibiting more statistical error.
This may explain why the highest accuracies in our
analyses were achieved by averaging spectra from a
square of nine pixels (18 ×18 cm) rather than extract-
ing single pixels, due to the increased signal-to-noise
ratio of the averaged spectra.
Secondly, accurate geolocation of the ground-
truthed vegetation points can be challenging. Across
the Arctic, species distribution is extremely hetero-
geneous and ground-truthing points may only define
species presence within a radius of <20 cm. Although
RMS errors for the orthomosaics were low (<0.14 m),
this could still have resulted in an offset between the
ground-truthed vegetation points and the imagery,
leading to some inaccuracies in the calibration and
validation points. Thus, in environments with high
spatial heterogeneity, extremely precise georeferen-
cing equipment is required to accurately calibrate
and validate the information collected from UAVs.
We recommend covering vegetation around ground-
truthing points to create an exclusion zone, or sim-
ilar methods to reduce the spectral effect of nearby
species. In addition, when dealing with mixed plant
communities, novel spectral methods are required
to capture the co-occurrence of plant types. Rather
than discrete vegetation categories, a more flexible
model may be required, that removes individual clas-
sifications and instead places a vegetation pixel on
a continuum of values between moss, graminoid
or shrub.
At the landscape level, plant trait distributions
displayed a steep water and fertility gradient from
the bird cliff colony to the sea, mediated by geo-
morphology. Plant water content was highest dir-
ectly underneath the bird colonies and at the con-
fluence of drainage channels. N, P and C:N values
indicated well-fertilised vegetation under nesting sites
and reflected the transport of nutrients downslope
through a network of drainage channels. For example,
plant water and nutrient levels were extremely high
in the depositional area below the snowmelt chan-
nel (figure 7), demonstrating the role that run-off
processes play in distributing marine-derived organic
matter across the otherwise nutrient-limited tun-
dra. While graminoids constituted just 4% of the
vegetation cover across the mapped area, they were
mostly found in moist, fertile areas, reflecting their
fast growth rates and ability to take swift advantage of
available nutrients [128]. This suggests that increased
mineralization rates under a warming climate may
preferentially benefit graminoid species (as found by
[129–131]).
14

Environ. Res. Lett. 16 (2021) 055006 E R Thomson et al
Adding to our confidence that the models were
predicting traits explicitly, rather than through
correlations with plant type, vegetation cover or
productivity, multispectral methods of plant trait
retrieval showed a clear independence from NDVI-
based predictions (figure S3). With the exception
of C:N ratios, neither measured nor predicted trait
values were significantly correlated with NDVI,
demonstrating the value of a multispectral mapping
approach and the lack of physical basis for using
NDVI to predict plant traits in the Arctic.
Scaling from the UAV to satellite level, we found
different methods of upscaling had a large effect on
the magnitude of results, but little influence on the
spatial pattern of trait values. Although we do not
present either map in figure 9as accurate, or have the
data to validate either map, we found that method A
(calibrating Sentinel imagery on UAV-derived plant
water values) produced more varying and realistic
trait values than method B (applying a UAV plant
water model to Sentinel imagery). The higher vari-
ation in trait values using method A is probably due
to the greater number of spectral bands used in the
analysis. In method A, all the relevant Sentinel bands
were included as predictors (ten in total), whereas
in method B, only the five Sentinel bands that best
matched the spectral bands of the MicaSense cam-
era could be used by the UAV plant water model. The
overestimation of trait values in method B demon-
strates the difference in reflectance captured by UAVs
and satellites (see figure S4). Cross-sensor differences
arise from atmospheric effects (e.g. haze), differences
in viewing and illumination angles, sensor specifica-
tions (e.g. bandwidths), and non-linear spectral mix-
ing at different spatial resolutions [32,127,132,133],
meaning predictive models generated from one type
of imagery cannot be directly applied to another with
the same accuracy.
To upscale from UAV to satellite imagery there-
fore, we recommend creating satellite-specific mod-
els calibrated on UAV maps (e.g. method A). Similar
methods have been used for a variety of ecological-
mapping tasks, such as tree species classification, per-
centage vegetation cover, agricultural assessments,
habitat mapping and fire severity quantification
[66,90,134–139]. To the best of our knowledge how-
ever, this is the first time that upscaling from UAV to
satellite imagery has been used to quantify plant traits.
With greater UAV coverage across the landscape and
more accurate UAV-models (r2> 0.9), using UAV
information to calibrate satellite-based models may
provide a valuable method for expanding spatially
limited plot studies across the Arctic.
5. Conclusion
Using field spectroscopy, we showed that eight com-
mon Arctic plants can be distinguished at the species
level at 74% accuracy. These results are relevant for
UAV-hyperspectral systems, the first Arctic passen-
ger aircraft equipped with hyperspectral instruments
[140] and upcoming hyperspectral satellites, which
should be able to use similar methods explored here
to advance these results and expand them over a wider
geographical area.
Using UAV multispectral information, we have
shown that High Arctic moss-, graminoid- and dwarf
shrub-dominated vegetation can be mapped at high
resolution (<10 cm) with 72% accuracy, alongside
corresponding water, nitrogen, phosphorus and C:N
values. We present two methods of upscaling these
results using Sentinel-2A imagery. The majority of
Arctic tundra regions remain under-investigated and
difficult to access for scientific research. We must
continue to develop methods to expand spatially-
limited plot studies, which inherently cannot capture
landscape-level vegetation patterns or functional trait
variation, in order to monitor environmental change
across the Arctic and understand the mechanisms
driving large-scale climate-vegetation feedbacks.
Data availability statement
The data that support the findings of this study are
openly available at the following URL/DOI: https://
osf.io/smbqh/.
Acknowledgments
We thank Plant Functional Trait Course 4 for the
funding and logistics that enabled the data collec-
tion. We thank the course instructors and course
participants for their help and collaboration during
the data collection and processing. We also thank
Christine Schirmer and the students at the Univer-
sity of Arizona for their assistance with the chemical
analyses.
Funding
The data collection was funded by a Norwegian
Research Council INTPART Grant (Project Num-
ber: 274831), two SIU-foundation projects (UTF-
2013/10074 and HNP-2015/10037) and a Research
Council of Norway Arctic Field Grant (Project Num-
ber: 282611, RiS: 10935). E R T is funded by NERC
DTP award (NE/L002612/1). Y M is supported by
the Jackson Foundation. M M-F was supported by a
NERC IRF (NE/L011859/1).
ORCID iDs
Eleanor R Thomson https://orcid.org/0000-0003-
1670-8970
Marcus P Spiegel https://orcid.org/0000-0001-
5879-5465
15

Environ. Res. Lett. 16 (2021) 055006 E R Thomson et al
Inge H J Althuizen https://orcid.org/0000-0003-
3485-9609
Aud H Halbritter https://orcid.org/0000-0003-
2597-6328
Jonathan J Henn https://orcid.org/0000-0003-
1551-9238
Ingibjörg S Jónsdóttir https://orcid.org/0000-
0003-3804-7077
Kari Klanderud https://orcid.org/0000-0003-
1049-7025
Yaoqi Li https://orcid.org/0000-0001-6540-395X
Brian S Maitner https://orcid.org/0000-0002-
2118-9880
Sean T Michaletz https://orcid.org/0000-0003-
2158-6525
Pekka Niittynen https://orcid.org/0000-0002-
7290-029X
Ruben E Roos https://orcid.org/0000-0002-1580-
6424
Brian J Enquist https://orcid.org/0000-0002-6124-
7096
Vigdis Vandvik https://orcid.org/0000-0003-4651-
4798
Marc Macias-Fauria https://orcid.org/0000-0002-
8438-2223
Yadvinder Malhi https://orcid.org/0000-0002-
3503-4783
References
[1] Box J E, Colgan W T, Christensen T R, Schmidt N M,
Lund M, Parmentier F-J W, Brown R, Bhatt U S,
Euskirchen E S and Romanovsky V E 2019 Key indicators
of arctic climate change: 1971–2017 Environ. Res. Lett.
14 045010
[2] Stocker T F, Qin D, Plattner G-K, Tignor M, Allen S K,
Boschung J, Nauels A, Xia Y, Bex V and Midgley P M 2013
Climate change 2013: the physical science basis
Contribution of Working Group I to the Fifth Assessment
Report of the Intergovernmental Panel on Climate Change
(Cambridge: Cambridge University Press) p 1535
[3] Joshua L A, Klein E S, Oberbauer S F and Welker J M 2016
Coupled long-term summer warming and deeper snow
alters species composition and stimulates gross primary
productivity in tussock tundra Oecologia 181 287–97
[4] Park T, Ganguly S, Tømmervik H, Euskirchen E S,
Høgda K-A, Karlsen S R, Brovkin V, Nemani R R and
Myneni R B 2016 Changes in growing season duration and
productivity of northern vegetation inferred from
long-term remote sensing data Environ. Res. Lett. 11 084001
[5] Khorsand Rosa R, Oberbauer S F, Starr G, Parker La
Puma I, Pop E, Ahlquist L and Baldwin T 2015 Plant
phenological responses to a long-term experimental
extension of growing season and soil warming in the
tussock tundra of Alaska Glob. Change Biol. 21 4520–32
[6] Oberbauer S F, Elmendorf S C, Troxler T G, Hollister R D,
Rocha A V, Bret-Harte M S, Dawes M A, Fosaa A M,
Henry G H R and Høye T T 2013 Phenological response of
tundra plants to background climate variation tested using
the International Tundra Experiment Phil. Trans. R. Soc. B
368 20120481
[7] Wahren C H A, Walker M D and Bret-Harte M S 2005
Vegetation responses in Alaskan arctic tundra after 8 years
of a summer warming and winter snow manipulation
experiment Glob. Change Biol. 11 537–52
[8] Davidson S J, Sloan V L, Phoenix G K, Wagner R, Fisher J P,
Oechel W C and Zona D 2016a Vegetation type dominates
the spatial variability in CH4emissions across multiple
arctic tundra landscapes Ecosystems 19 1116–32
[9] Higuera P E, Brubaker L B, Anderson P M, Brown T A,
Kennedy A T and Hu F S 2008 Frequent fires in ancient
shrub tundra: implications of paleorecords for arctic
environmental change PLoS One 3e0001744
[10] Loranty M M, Goetz S J and Beck P S A 2011 Tundra
vegetation effects on pan-Arctic albedo Environ. Res. Lett.
6024014
[11] Nadelhoffer K J, Giblin A E, Shaver G R and Laundre J A
1991 Effects of temperature and substrate quality on
element mineralization in six Arctic soils Ecology
72 242–53
[12] Nauta A L, Heijmans M M P D, Blok D, Limpens J,
Elberling B, Gallagher A, Li B, Petrov R E, Maximov T C
and Van Huissteden J 2015 Permafrost collapse after shrub
removal shifts tundra ecosystem to a methane source Nat.
Clim. Change 567–70
[13] Pearson R G, Phillips S J, Loranty M M, Beck P S A,
Damoulas T, Knight S J and Goetz S J 2013 Shifts in Arctic
vegetation and associated feedbacks under climate change
Nat. Clim. Change 3673–7
[14] Sturm M and Douglas T 2005 Changing snow and shrub
conditions affect albedo with global implications
J. Geophys. Res. 110 1–13
[15] Bjorkman A D, Myers-Smith I H, Elmendorf S C,
Normand S, Rüger N, Beck P S A, Blach-Overgaard A,
Blok D, Cornelissen J H C and Forbes B C 2018 Plant
functional trait change across a warming tundra biome
Nature 562 57–62
[16] Mekonnen Z A, Riley W J and Grant R F 2018 Accelerated
nutrient cycling and increased light competition will lead
to 21st century shrub expansion in North American Arctic
Tundra J. Geophys. Res. Biogeosci. 123 1683–701
[17] Soudzilovskaia N A, Elumeeva T G, Onipchenko V G,
Shidakov I I, Salpagarova F S, Khubiev A B, Tekeev D K and
Cornelissen J H C 2013a Functional traits predict
relationship between plant abundance dynamic and
long-term climate warming Proc. Natl Acad. Sci.
110 18180–4
[18] Cornwell W K, Cornelissen J H C, Amatangelo K,
Dorrepaal E, Eviner V T, Godoy O, Hobbie S E, Hoorens B,
Kurokawa H and Pérez-Harguindeguy N 2008 Plant species
traits are the predominant control on litter decomposition
rates within biomes worldwide Ecol. Lett. 11 1065–71
[19] Díaz S, Kattge J, Cornelissen J H C, Wright I J, Lavorel S,
Dray S, Reu B, Kleyer M, Wirth C and Prentice I C 2016
The global spectrum of plant form and function Nature
529 167–71
[20] Reich P B, Walters M B and Ellsworth D S 1997 From
tropics to tundra: global convergence in plant functioning
Proc. Natl Acad. Sci. 94 13730–4
[21] Shipley B, Lechowicz M J, Wright I and Reich P B 2006
Fundamental trade-offs generating the worldwide leaf
economics spectrum Ecology 87 535–41
[22] Wright I J, Reich P B, Westoby M, Ackerly D D, Baruch Z,
Bongers F, Cavender-Bares J, Chapin T, Cornelissen J H C
and Diemer M 2004 The worldwide leaf economics
spectrum Nature 428 821–7
[23] Hudson J M G and Henry G H R 2009 Increased plant
biomass in a high Arctic heath community from 1981 to
2008 Ecology 90 2657–63
[24] Juszak I, Erb A M, Maximov T C and Schaepman-Strub G
2014 Arctic shrub effects on NDVI, summer albedo and
soil shading Remote Sens. Environ. 153 79–89
[25] Myers-Smith I H, Thomas H J D and Bjorkman A D 2019
Plant traits inform predictions of tundra responses to
global change New Phytol. 221 1742–8
[26] Thomas H J et al 2020 Global plant trait relationships
extend to the climatic extremes of the tundra biome Nat.
Commun. 11 1–12
16

Environ. Res. Lett. 16 (2021) 055006 E R Thomson et al
[27] Metcalfe D B, Hermans T D G, Ahlstrand J, Becker M,
Berggren M, Björk R G, Björkman M P, Blok D,
Chaudhary N and Chisholm C 2018 Patchy field sampling
biases understanding of climate change impacts across the
Arctic Nat. Ecol. Evol. 21443–8
[28] Raynolds M K, Walker D A, Balser A, Bay C, Campbell M,
Cherosov M M, Daniëls F J A, Eidesen P B, Ermokhina K A
and Frost G V 2019 A raster version of the circumpolar
Arctic vegetation map (CAVM) Remote Sens. Environ.
232 111297
[29] Myneni R B, Keeling C D, Tucker C J, Asrar G and
Nemani R R 1997 Increased plant growth in the northern
high latitudes from 1981 to 1991 Nature 386 698–702
[30] Bhatt U S, Walker D A, Raynolds M K, Bieniek P A,
Epstein H E, Comiso J C, Pinzon J E, Tucker C J and
Polyakov I V 2013 Recent declines in warming and
vegetation greening trends over pan-Arctic tundra Remote
Sens. 54229–54
[31] Zhang Y, Song C, Band L E, Sun G and Junxiang L 2017
Reanalysis of global terrestrial vegetation trends from
MODIS products: browning or greening? Remote Sens.
Environ. 191 145–55
[32] Myers-Smith I H, Kerby J T, Phoenix G K, Bjerke J W,
Epstein H E, Assmann J J, John C, Andreu-Hayles L,
Angers-Blondin S and Beck P S A 2020 Complexity
revealed in the greening of the Arctic Nat. Clim. Change
10 106–17
[33] Blok D, Schaepman-Strub G, Bartholomeus H,
Heijmans M M P D, Maximov T C and Berendse F 2011
The response of arctic vegetation to the summer climate:
relation between shrub cover, NDVI, surface albedo and
temperature Environ. Res. Lett. 6035502
[34] Frost G V, Epstein H E, Walker D A, Matyshak G and
Ermokhina K 2018 Seasonal and long-term changes to
active-layer temperatures after tall shrubland expansion
and succession in Arctic tundra Ecosystems 21 507–20
[35] Loranty M M and Goetz S J 2012 Shrub expansion and
climate feedbacks in Arctic tundra Environ. Res. Lett.
7011005
[36] Bonfils C J W, Phillips T J, Lawrence D M,
Cameron-Smith P, Riley W J and Subin Z M 2012 On the
influence of shrub height and expansion on northern high
latitude climate Environ. Res. Lett. 7015503
[37] Blok D, Heijmans M M P D, Schaepman-strub G,
Kononov A V, Maximov T C and Berendse F 2010 Shrub
expansion may reduce summer permafrost thaw in siberian
tundra Glob. Change Biol. 16 1296–305
[38] Arndt K A, Santos M J, Ustin S, Davidson S J, Stow D,
Oechel W C, Tran T T P, Graybill B and Zona D 2019 Arctic
greening associated with lengthening growing seasons in
Northern Alaska Environ. Res. Lett. 14 125018
[39] Fauchald P, Park T, Tømmervik H, Myneni R and
Hausner V H 2017 Arctic greening from warming promotes
declines in Caribou populations Sci. Adv. 3e1601365
[40] Ims R A, Henden J-A, Strømeng M A, Thingnes A V,
Garmo M J and Jepsen J U 2019 Arctic greening and bird
nest predation risk across tundra ecotones Nat. Clim.
Change 9607–10
[41] Tape K D, Christie K, Carroll G and O’Donnell J A 2016
Novel wildlife in the a rctic: the influence of changing
Riparian ecosystems and shrub habitat expansion on
snowshoe hares Glob. Change Biol. 22 208–19
[42] Opedal Ø H, Armbruster W S and Graae B J 2015 Linking
small-scale topography with microclimate, plant species
diversity and intra-specific trait variation in an alpine
landscape Plant Ecol. Divers. 8305–15
[43] Suvanto S, Le Roux P C and Luoto M 2014 Arctic-alpine
vegetation biomass is driven by fine-scale abiotic
heterogeneity Geogr. Ann. A96 549–60
[44] Räsänen A and Virtanen T 2019 Data and resolution
requirements in mapping vegetation in spatially
heterogeneous landscapes Remote Sens. Environ.
230 111207
[45] Pattison R R, Jorgenson J C, Raynolds M K and Welker J M
2015 Trends in NDVI and tundra community composition
in the arctic of NE Alaska between 1984 and 2009
Ecosystems 18 707–19
[46] Guay K C, Beck P S A, Berner L T, Goetz S J, Baccini A and
Buermann W 2014 Vegetation productivity patterns at high
northern latitudes: a multi-sensor satellite data assessment
Glob. Change Biol. 20 3147–58
[47] Assmann J J, Myers-Smith I H, Kerby J T, Cunliffe A M and
Daskalova G N 2020 Drone data reveal heterogeneity in
tundra greenness and phenology not captured by satellites
Environ. Res. Lett. 15 125002
[48] Siewert M B and Olofsson J 2020 Scale-dependency of
Arctic ecosystem properties revealed by UAV Environ. Res.
Lett. 15 094030
[49] Asner G P, Martin R E and Suhaili A B 2012 Sources of
canopy chemical and spectral diversity in lowland Bornean
forest Ecosystems 15 504–17
[50] Serbin S P, Singh A, McNeil B E, Kingdon C C and
Townsend P A 2014 Spectroscopic determination of leaf
morphological and biochemical traits for northern
temperate and boreal tree species Ecol. Appl. 24 1651–69
[51] Serbin S P, Jin W, Ely K S, Kruger E L, Townsend P A,
Meng R, Wolfe B T, Chlus A, Wang Z and Rogers A 2019
From the Arctic to the tropics: multibiome prediction of
leaf mass per area using leaf reflectance New Phytol.
224 1557–68
[52] Roth K L, Roberts D A, Dennison P E, Alonzo M,
Peterson S H and Beland M 2015 Differentiating plant
species within and across diverse ecosystems with imaging
spectroscopy Remote Sens. Environ. 167 135–51
[53] Curran P J 1989 Remote sensing of foliar chemistry Remote
Sens. Environ. 30 271–8
[54] Asner G P, Knapp D E, Kennedy-Bowdoin T, Jones M O,
Martin R E, Boardman J and Flint Hughes R 2008 Invasive
species detection in hawaiian rainforests using airborne
imaging spectroscopy and LiDAR Remote Sens. Environ.
112 1942–55
[55] Durgante F M, Higuchi N, Almeida A and Vicentini A 2013
Species spectral signature: discriminating closely related
plant species in the amazon with near-infrared
leaf-spectroscopy For. Ecol. Manage. 291 240–8
[56] Lopatin J, Fassnacht F E, Kattenborn T and Schmidtlein S
2017 Mapping plant species in mixed grassland
communities using close range imaging spectroscopy
Remote Sens. Environ. 201 12–23
[57] Asner G P and Martin R E 2009 Airborne spectranomics:
mapping canopy chemical and taxonomic diversity in
tropical forests Front. Ecol. Environ. 7269–76
[58] Bratsch S N, Epstein H E, Buchhorn M and Walker D A
2016 Differentiating among four arctic tundra plant
communities at Ivotuk, Alaska using field spectroscopy
Remote Sens. 851
[59] Davidson S J, Santos M J, Sloan V L, Watts J D,
Phoenix G K, Oechel W C and Zona D 2016b Mapping
Arctic tundra vegetation communities using field
spectroscopy and multispectral satellite data in north
Alaska, USA Remote Sens. 8978
[60] Beamish A L, Coops N, Chabrillat S and Heim B 2017 A
phenological approach to spectral differentiation of
low-Arctic tundra vegetation communities, north slope,
Alaska Remote Sens. 91200
[61] Buchhorn M, Walker D A, Heim B, Raynolds M K,
Epstein H E and Schwieder M 2013 Ground-based
hyperspectral characterization of Alaska tundra vegetation
along environmental gradients Remote Sens. 53971–4005
[62] Huemmrich K F, Gamon J A, Tweedie C E, Campbell P K E,
Landis D R and Middleton E M 2013 Arctic tundra
vegetation functional types based on photosynthetic
physiology and optical properties IEEE J. Sel. Top. Appl.
Earth Obs. Remote Sens. 6265–75
[63] Liu N, Budkewitsch P and Treitz P 2017 Examining spectral
reflectance features related to arctic percent vegetation
17

Environ. Res. Lett. 16 (2021) 055006 E R Thomson et al
cover: implications for hyperspectral remote sensing of
Arctic tundra Remote Sens. Environ. 192 58–72
[64] Bratsch S, Epstein H, Buchhorn M, Walker D and Landes H
2017 Relationships between hyperspectral data and
components of vegetation biomass in low Arctic tundra
communities at Ivotuk, Alaska Environ. Res. Lett. 12 025003
[65] Kennedy B E, King D J and Duffe J 2020 Comparison of
empirical and physical modelling for estimation of
biochemical and biophysical vegetation properties: field
scale analysis across an Arctic bioclimatic gradient Remote
Sens. 12 3073
[66] Riihimäki H, Luoto M and Heiskanen J 2019 Estimating
fractional cover of tundra vegetation at multiple scales
using unmanned aerial systems and optical satellite data
Remote Sens. Environ. 224 119–32
[67] Mora C, Vieira G, Pina P, Lousada M and Christiansen H H
2015 Land cover classification using high-resolution aerial
photography in adventdalen, Svalbard Geogr. Ann. A
97 473–88
[68] Tømmervik H, Karlsen S-R, Nilsen L, Johansen B,
Storvold R, Zmarz A, Beck P S, Høgda K A, Goetz S and
Park T 2014 Use of unmanned aircraft systems (UAS) in a
multi-scale vegetation index study of Arctic plant
communities in adventdalen on Svalbard EARSeL eProc.
13 47–52
[69] Baena S, Moat J, Whaley O and Boyd D S 2017 Identifying
species from the air: uAVs and the very high resolution
challenge for plant conservation PLoS One 12 11
[70] Lu B and Yuhong H 2017 Species classification using
unmanned aerial vehicle (UAV)-acquired high spatial
resolution imagery in a heterogeneous grassland ISPRS J.
Photogramm. Remote Sens. 128 73–85
[71] Oldeland J, Große-Stoltenberg A, Naftal L and
Strohbach B J 2017 The potential of UAV derived image
features for discriminating savannah tree species The Roles
of Remote Sensing in Nature Conservation (Berlin: Springer)
pp 183–201
[72] Constantin D, Rehak M, Akhtman Y and Liebisch F 2015
Detection of crop properties by means of hyperspectral
remote sensing from a micro UAV Bornimer
Agrartechnische Berichte 88 129–37
[73] Capolupo A, Kooistra L, Berendonk C, Boccia L and
Suomalainen J 2015 Estimating plant traits of grasslands
from UAV-acquired hyperspectral images: a comparison of
statistical approaches ISPRS Int. J. Geo-Inf. 42792–820
[74] Thomson E R, Malhi Y, Bartholomeus H, Oliveras I,
Gvozdevaite A, Peprah T, Suomalainen J, Quansah J,
Seidu J and Adonteng C 2018 Mapping the leaf economic
spectrum across West African tropical forests using
UAV-acquired hyperspectral imagery Remote Sens. 10 1532
[75] Dash J P, Pearse G D and Watt M S 2018 UAV multispectral
imagery can complement satellite data for monitoring
forest health Remote Sens. 10 1216
[76] Lucieer A, Turner D, King D H and Robinson S A 2014
Using an unmanned aerial vehicle (UAV) to capture
micro-topography of Antarctic moss beds Int. J. Appl. Earth
Obs. Geoinf. 27 53–62
[77] Turner D, Lucieer A, Malenovský Z, King D and
Robinson S A 2018 Assessment of Antarctic moss health
from multi-sensor uas imagery with random forest
modelling Int. J. Appl. Earth Obs. Geoinf. 68 168–79
[78] Fraser R H, Olthof I, Lantz T C and Schmitt C 2016 UAV
photogrammetry for mapping vegetation in the low-Arctic
Arctic Sci. 279–102
[79] Palace M, Herrick C, DelGreco J, Finnell D, Garnello A J,
McCalley C, McArthur K, Sullivan F and Varner R K 2018
Determining subarctic peatland vegetation using an
unmanned aerial system (UAS) Remote Sens. 10 1498
[80] Langford Z, Kumar J, Hoffman F M, Norby R J,
Wullschleger S D, Sloan V L and Iversen C M 2016
Mapping Arctic plant functional type distributions in the
barrow environmental observatory using WorldView-2 and
LiDAR datasets Remote Sens. 8733
[81] Lawrimore J H, Ray R, Applequist S, Korzeniewski B and
Menne M J 2016 Global Summary of the Month (GSOM),
version 1.0. NOAA National Centers for Environmental
Information (Accessed 26 October 2020)
[82] Jones A, Stolbovay V, Tarnocai C, Broll G, Spaargaren O
and Montanarella L 2010 Soil Atlas of the Northern
Circumpolar Region (Luxembourg: European Commission)
[83] Zwolicki A, Zmudczy´
nska-Skarbek K M, Iliszko L and
Stempniewicz L 2013 Guano deposition and nutrient
enrichment in the vicinity of planktivorous and piscivorous
seabird colonies in spitsbergen Polar Biol. 36 363–72
[84] Henry G H R and Molau U 1997 Tundra plants and climate
change: the International Tundra Experiment (ITEX) Glob.
Change Biol. 31–9
[85] Bjerke J W, Treharne R, Vikhamar-Schuler D, Karlsen S R,
Ravolainen V, Bokhorst S, Phoenix G K, Bochenek Z and
Hans T 2017 Understanding the drivers of extensive plant
damage in boreal and arctic ecosystems: insights from field
surveys in the aftermath of damage Sci. Total Environ.
599 1965–76
[86] Breiman L 2001 Random Forests Mach. Learn. 45 5–32
[87] Ghimire B, Rogan J and Miller J 2010 Contextual
land-cover classification: incorporating spatial dependence
in land-cover classification models using random forests
and the Getis statistic Remote Sens. Lett. 145–54
[88] Kotsiantis S and Pintelas P 2004 Combining bagging and
boosting Int. J. Comput. Intell. 1324–33
[89] Haaland D M and Thomas E V 1988 Partial least-squares
methods for spectral analyses. 1. Relation to other
quantitative calibration methods and the extraction of
qualitative information Anal. Chem. 60 1193–202
[90] Asner G P, Martin R E, Anderson C B and Knapp D E 2015
Quantifying forest canopy traits: imaging spectroscopy
versus field survey Remote Sens. Environ. 158 15–27
[91] Asner G P and Martin R E 2015 Spectroscopic remote
sensing of non-structural carbohydrates in forest canopies
Remote Sens. 73526–47
[92] Hijmans R J, Van Etten J, Cheng J, Mattiuzzi M, Sumner M,
Greenberg J A, Lamigueiro O P, Bevan A, Racine E B and
Shortridge A 2015 Package ‘raster’ R Package
[93] Bokhorst S, Tømmervik H, Callaghan T V, Phoenix G K
and Bjerke J W 2012 Vegetation recovery following extreme
winter warming events in the sub-Arctic estimated using
NDVI from remote sensing and handheld passive proximal
sensors Environ. Exp. Bot. 81 18–25
[94] Nieke J and Rast M 2018 Towards the copernicus
hyperspectral imaging mission for the environment
(CHIME) Proc. IEEE Int. Geosci. Remote Sens. Symp.
(IGARSS) (July 2018) pp 157–9
[95] Guanter L, Kaufmann H, Foerster S, Brosinsky A, Wulf H,
Bochow M, Boesche N, Brell M, Buddenbaum H and
Chabrillat S 2016 EnMAP Science Plan—Environmental
Mapping and Analysis Program
[96] Matsunaga T, Iwasaki A, Tsuchida S, Tanii J, Kashimura O,
Nakamura R, Yamamoto H, Tachikawa T and Rokugawa S
2013 Current status of Hyperspectral Imager Suite (HISUI)
Proc. IEEE Int. Geosci. Remote Sens. Symp. (IGARSS), 2013
[97] Carrere V, Briottet X, Jacquemoud S, Marion R,
Bourguignon A, Chami M, Dumont M,
Minghelli-Roman A, Weber C and Marie-José L-F 2013
HYPXIM: a second generation high spatial resolution
hyperspectral satellite for dual applications Paper Presented
at 5th Workshop on Hyperspectral Image and Signal
Processing: Evolution in Remote Sensing (Gainsville, FL,
25–28 June 2013)
[98] Lee C M, Cable M L, Hook S J, Green R O, Ustin S L,
Mandl D J and Middleton E M 2015 An introduction to the
NASA Hyperspectral Infrared Imager (HyspIRI) mission
and preparatory activities Remote Sens. Environ.
167 6–19
[99] Pignatti S, Palombo A, Pascucci S, Romano F, Santini F,
Simoniello T, Umberto A, Vincenzo C, Acito N and
Diani M 2013 The PRISMA hyperspectral mission: science
18

Environ. Res. Lett. 16 (2021) 055006 E R Thomson et al
activities and opportunities for agriculture and land
monitoring Proc. IEEE Int. Geosci. Remote Sens. Symp.
(IGARSS), 2013
[100] Turetsky M R, Mack M C, Hollingsworth T N and
Harden J W 2010 The role of mosses in ecosystem
succession and function in Alaska’s boreal forest Can. J. For.
Res. 40 1237–64
[101] Longton R E 1988 Adaptations and strategies of polar
bryophytes Bot. J. Linn. Soc. 98 253–68
[102] Hedderson T A and Longton R E 1996 Life history variation
in mosses: water relations, size and phylogeny Oikos
77 31–43
[103] Elumeeva T G, Soudzilovskaia N A, During H J and
Cornelissen J H C 2011 The importance of colony structure
versus shoot morphology for the water balance of 22
subarctic bryophyte species J. Veg. Sci. 22 152–64
[104] Turetsky M R, Bond-Lamberty B, Euskirchen E, Talbot J,
Steve Frolking A, McGuire D and Tuittila E - S 2012 The
resilience and functional role of moss in boreal and Arctic
ecosystems New Phytol. 196 49–67
[105] Elberling B 2007 Annual soil CO2effluxes in the high
arctic: the role of snow thickness and vegetation type Soil
Biol. Biochem. 39 646–54
[106] Baddeley J A, Woodin S J and Alexander I J 1994 Effects of
increased nitrogen and phosphorus availability on the
photosynthesis and nutrient relations of three Arctic dwarf
shrubs from svalbard Functional Ecology 8pp 676–85
[107] Asner G P and Martin R E 2008 Spectral and chemical
analysis of tropical forests: scaling from leaf to canopy
levels Remote Sens. Environ. 112 3958–70
[108] Ullah S, Skidmore A K, Ramoelo A, Groen T A, Naeem M
and Ali A 2014 Retrieval of leaf water content spanning the
visible to thermal infrared spectra ISPRS J. Photogramm.
Remote Sens. 93 56–64
[109] Asner G P, Martin R E, Carranza-Jiménez L, Sinca F,
Tupayachi R, Anderson C B and Martinez P 2014
Functional and biological diversity of foliar spectra in tree
canopies throughout the Andes to Amazon region New
Phytol. 204 127–39
[110] Mutanga O and Kumar L 2007 Estimating and mapping
grass phosphorus concentration in an African savanna
using hyperspectral image data Int. J. Remote Sens.
28 4897–911
[111] Porder S, Asner G P and Vitousek P M 2005 Ground-based
and remotely sensed nutrient availability across a tropical
landscape Proc. Natl Acad. Sci. 102 10909–12
[112] Asner G P, Knapp D E, Anderson C B, Martin R E and
Vaughn N 2016 Large-scale climatic and geophysical
controls on the leaf economics spectrum Proc. Natl Acad.
Sci. 113 E4043–51
[113] Mahajan G R, Sahoo R N, Pandey R N, Gupta V K and
Kumar D 2014 Using hyperspectral remote sensing
techniques to monitor nitrogen, phosphorus, sulphur and
potassium in wheat (Triticum Aestivum L.) Precis. Agric.
15 499–522
[114] Mahajan G R, Pandey R N, Sahoo R N, Gupta V K,
Datta S C and Kumar D 2017 Monitoring nitrogen,
phosphorus and sulphur in hybrid rice (Oryza Sativa L.)
using hyperspectral remote sensing Precis. Agric. 18 736–61
[115] Kokaly R F 2001 Investigating a physical basis for
spectroscopic estimates of leaf nitrogen concentration
Remote Sens. Environ. 75 153–61
[116] Milton N M, Eiswerth B A and Ager C M 1991 Effect of
phosphorus deficiency on spectral reflectance and
morphology of soybean plants Remote Sens. Environ.
36 121–7
[117] Thomas H J D, Myers-Smith I H, Bjorkman A D,
Elmendorf S C, Blok D, Cornelissen J H C, Forbes B C,
Hollister R D, Normand S and Prevéy J S 2019 Traditional
plant functional groups explain variation in economic but
not size-related traits across the tundra biome Glob. Ecol.
Biogeogr. 28 78–95
[118] Hofmann W 1968 Geobotanische Untersuchungen in
Sudost-Spitzbergen 1960 Ergebn. Stauferland-Exped.
1959/60,8 pp 1–83
[119] Vanderpuye A W, Elvebakk A and Nilsen L 2002 Plant
communities along environmental gradients of high-arctic
mires in Sassendalen, Svalbard J. Veg. Sci. 13 875–84
[120] Elven R, Eriksen M B, Elvebakk A, Johansen B and
Engelskjøn T 1990 Gipsdalen, central svalbard; flora,
vegetation and botanical values Environmental Atlas,
Gipsdalen, Svalbard. II. Nor. Polarinst. Rapp. Ser. vol 61
pp 27–66
[121] Callaghan T V, Press M C, Lee J A, Robinson D L and
Anderson C W 1999 Spatial and temporal variability in the
responses of Arctic terrestrial ecosystems to environmental
change Polar Res. 18 191–7
[122] Robinson S A, Turnbull J D and Lovelock C E 2005 Impact
of changes in natural ultraviolet radiation on pigment
composition, physiological and morphological
characteristics of the Antarctic Moss, Grimmia Antarctici
Glob. Change Biol. 11 476–89
[123] Royles J, Amesbury M J, Convey P, Griffiths H,
Hodgson D A, Leng M J and Charman D J 2013 Plants and
soil microbes respond to recent warming on the Antarctic
Peninsula Curr. Biol. 23 1702–6
[124] May J L, Parker T, Unger S and Oberbauer S F 2018 Short
term changes in moisture content drive strong changes in
normalized difference vegetation index and gross primary
productivity in four arctic moss communities Remote Sens.
Environ. 212 114–20
[125] Van Leeuwen W J D and Huete A R 1996 Effects of standing
litter on the biophysical interpretation of plant canopies
with spectral indices Remote Sens. Environ. 55 123–38
[126] Assmann J J, Kerby J T, Cunliffe A M and Myers-Smith I H
2018 Vegetation monitoring using multispectral
sensors—best practices and lessons learned from high
latitudes J. Unmanned Veh. Syst. 754–75
[127] Stow D, Nichol C J, Wade T, Assmann J J, Simpson G and
Helfter C 2019 Illumination geometry and flying height
influence surface reflectance and NDVI derived from
multispectral UAS imagery Drones 355
[128] Wookey P A, Aerts R, Bardgett R D, Baptist F, Bråthen K A,
Cornelissen J H C, Gough L, Hartley I P, Hopkins D W and
Lavorel S 2009 Ecosystem feedbacks and cascade processes:
understanding their role in the responses of arctic and
alpine ecosystems to environmental change Glob. Change
Biol. 15 1153–72
[129] Van Der Wal R and Brooker R W 2004 Mosses mediate
grazer impacts on grass abundance in Arctic ecosystems
Funct. Ecol. 18 77–86
[130] Theodose T A and Bowman W D 1997 The influence of
interspecific competition on the distribution of
an alpine graminoid: evidence for the importance of plant
competition in an extreme environment Oikos 79 101–14
[131] Tomassen H B M, Smolders A J P, Limpens J, Lamers L P M
and Roelofs J G M 2004 Expansion of invasive species on
ombrotrophic bogs: desiccation or high N deposition?
J. Appl. Ecol. 41 139–50
[132] Beamish A, Raynolds M K, Epstein H, Frost G V,
Macander M J, Bergstedt H, Bartsch A, Kruse S, Miles V
and Tanis C M 2020 Recent trends and remaining
challenges for optical remote sensing of Arctic tundra
vegetation: a review and outlook Remote Sens. Environ.
246 111872
[133] Buchhorn M, Raynolds M K and Walker D A 2016
Influence of BRDF on NDVI and biomass
estimations of Alaska Arctic tundra Environ. Res. Lett.
11 125002
[134] Pla M, Bota G, Duane A, Balagué J, Curcó A, Gutiérrez R
and Brotons L 2019 Calibrating Sentinel-2 imagery with
multispectral UAV derived information to quantify
damages in mediterranean rice crops caused by western
swamphen (Porphyrio Porphyrio) Drones 345
19

Environ. Res. Lett. 16 (2021) 055006 E R Thomson et al
[135] Kattenborn T, Lopatin J, Förster M, Braun A C and
Fassnacht F E 2019 UAV data as alternative to field
sampling to map woody invasive species based on
combined Sentinel-1 and Sentinel-2 data Remote Sens.
Environ. 227 61–73
[136] Immitzer, Markus, Clement Atzberger, and Tatjana Koukal
2012 Tree species classification with random forest using
very high spatial resolution 8-Band WorldView-2 satellite
data Remote Sens. 42661–93
[137] Fraser R H, Van Der Sluijs J and Hall R J 2017 Calibrating
satellite-based indices of burn severity from UAV-derived
metrics of a burned boreal forest in NWT, Canada Remote
Sens. 9279
[138] Pla M, Duane A and Brotons L 2017 Potencial De Las
Im´
agenes UAV Como Datos De Verdad Terreno
Para La Clasificación De La Severidad De Quema De
Im´
agenes Landsat: aproximaciones a Un Producto ´
Util
Para La Gestión Post Incendio Rev. Teledetección
49 91–102
[139] Leit˜
ao P J, Schwieder M, Pötzschner F, Pinto J R R,
Teixeira A M C, Pedroni F, Sanchez M, Rogass C,
Van Der Linden S and Bustamante M M C 2018 From
sample to pixel: multi-scale remote sensing data for
upscaling aboveground carbon data in heterogeneous
landscapes Ecosphere 9e02298
[140] Sivertson A H, Stodle D, Solbo S and Storvold R A new
airborne remote sensing platform at Svalbard Poster
Session Presented at Svalbard Science Conf.: Svalbard in
a pan–Arctic Perspective (Oslo, Norway, 5–6 November
2019)
20