Multisensor satellite monitoring of the cryosphere and phenology on
Svalbard
E. Malnes1, Karlsen SR1, Johansen B1 , Haarpaintner J.,1 Storvold R.1 & Høgda KA1
(1)Northern
Research Institute (Norut), Tromsø, Norway
Abstract
First and last day of snow on Svalbard
Climate models suggest that Svalbard will face larger temperature increases than most other
places on earth in the next century. Some places will experience up to 8°C increase of the
average annual temperature. Svalbard has become a large-scale laboratory of studying climate
change. It is thus utmost important to monitor the changes already taking place. Arctic
ecosystems display unique qualities dictated by the physical environment. The cryosphere, and
in particular the snow cover is an especially important ecological factor, affecting soil
moisture, plant survival, plant community composition and controlling the length of the
growing season. Short-stemmed perennial herbs, low-growing prostrate shrubs, lichens and
mosses characterize plant communities.
By using the five year time series over Svalbard we are able to establish pixel-wise the first and
last day of snow. We define the first snow free day as the first day with five consecutive days
with less than 30% snow cover fraction. In the same way the first snow day in the fall is
defined as the first day with five consecutive days of snow cover fraction above 30%. The
maps below shows the averages over the five year period where both ASAR and MODIS data
were available.
16°E
20°E
24°E
28°E
30°E
Last day with snow cover
Kvitøya
12°E
16°E
20°E
I
I
N
S
T
R
E
T
E
T
N O R D A U S T L A N D E T
E
E
N
N
E
P
P
S
O
O
S
S
T
T
E
E
E
HÅ K ON V II
LAND
R
R
R
IK
L
L
R
R
IK
S
E
N
S
T
R
E
T
E
T
IK
E
E
T
T
G
G
A
r
Fo
A
nd
T
n
su
T
n
su
Fø rs te dag me d
s n ø de kke
S
OS CA R I I
la
S
S is te d ag m ed
s n ø de kke
L
O L A V V
L A N D
L
OS CA R I I
K ON G K ARL S LA ND
( i k k e k a r t l a gt )
79°N
O
K ON G K ARL S LA ND
( i k k e k a r t l a gt )
O
O L A V V
L A N D
de
de
N
S T
O R
F J
O R
D E
N
IS
F
J
O
BARENTSØYA
R
D
E
N
T
E D G E Ø Y A
LA ND
78°N
E
LAND
< 1. juni
1 - 15. juni
16 - 30. juni
1 - 15. juli
16 - 30. juli
> 1. a ugus t
T
E
N ORD ENS K I ÖLD
N ORD ENS K I ÖLD
LAND
Ikke ka rtla gt
77°N
E
D
R
BARENTSØYA
R
T
O
E
J
T
F
E
IS
t
R
t
LA ND
78°N
S T
O R
F J
O R
D E
N
nd
The Svalbard Archipelago is located at around 78°N, 15°E. Since it is easily accessible with
regular flights from Europe, it is often use as an experimental laboratory for scientist to study
climate changes in the Arctic. Snow cover is a central environmental factor in various studies
of plants and wildlife. The current project have focused on establishing a base line climatology
for the snow cover on Svalbard.
E
D
D
E
R
N
N
E
IK
79°N
la
Observations
30°E
Kvitøya
H
80°N
N O R D A U S T L A N D E T
HÅ K ON V II
LAND
E D G E Ø Y A
< 1. s ept.
1 - 10. s ept.
11 - 20. s e pt.
21 - 30. se pt.
> 1. oktobe r
Ikke ka rtla gt
77°N
B A R E N T S -
The correlation between snow cover, vegetation composition and distribution of species have
been studied by comparing derived snow maps and satellite-based vegetation maps. At local
scale the amounts of snow varies in the terrain with a low snow cover fraction on ridges and
full snow cover in depressions. At a regional scale the most varied vegetation is located in the
inner Isfjord area with early snowmelt and a relatively long growing season. On the opposite
side the Arctic Polar Desert Zone, located to the eastern and northernmost areas is
characterized by an extremely short growing season. Due to an abbreviated growing season
only a few vascular plants are adapted to the harsh growing conditions. Moss and lichens
characterize the community types here.
28°E
First day with snow cover
(ikke kartlagt)
H
80°N
24°E
F ø r s te d a g m e d s n ø d e k k e p å S v a lb a rd
(ikke kartlagt)
r
Fo
Here we present the most recent snow cover maps developed for the entire Svalbard
archipelago. A ten-year long climate record on snow has been developed using Terra MODIS
data and since 2005 also using Envisat ASAR satellite data. Daily time-series of snow cover
are used to derive the spatial distribution of snow, assessing the first day of snow free
conditions in early summer, first day of snow in autumn and the annual number of snow free
days. Correspondent average maps for the ten-year period has been produced. The developed
map products provide valuable information about the exact timing of snow and its relation to
the variability of growing season, biodiversity and vegetation cover. A daily
multisensor/multi-temporal cloud free fractional snow cover area product is automatically
generated based on multi-temporal interpolation in combination with multi-sensoral fusion of
SAR and optical data. At local scales we also introduce high-resolution snow maps from SAR
sensors like TerraSAR-X and Radarsat-2 to study detailed snow melting patterns at 2-5 m
scale.
12°E
S is t e d a g m e d s n ø d e k k e p å S v a lb a r d
B A R E N T S H OP EN
H OP EN
( i k k e k ar t l a g t )
H A V E T
50 km
( i k k e k ar t l a g t )
H A V E T
50 km
Figure 3. Last day with snow cover (left) and first day with snow cover (right). Areas
above 80 degrees North are not covered.
Figure 3 shows that only 2.6% of the ice free areas are snow free before June 1 on average for
the five years of data.Most of these areas are in the inner fjord zone nearby Longyearbyen. 18%
of the areas melts in June and 38% in July. A large fraction (41%) of the ice free areas
(unglaciated land areas) comes in the category after August 1 and indicates that a large fraction
of the land areas are not snow free every year. This is the dominating category in the northern
and eastern areas.
Consequences for vegetation and phenology
The phenological phases on Svalbard has also been studied in the same project. By defining
thresholds for the Normalized Differential Vegetation Index (NDVI) based on field studies of
the coloring of the small willow plant (Salix polaris), which is abundant on Svalbard, we are
able to define the onset and the end of the growing season.
In the inner fjord zone nearby Longyearbyen we have studied the phenophases in more detail,
and compared them with the snow melting patterns. In general there is a very good
correspondence between the time of when snow disappear and when flowering begins.
Is
fj
o
rd
e
n
Is
Long ye a rbye n
fj
o
rd
e
n
Longye a rbye n
(
!(
!
(
!
(
! (
!
S ta rt p å
v e k s ts e s o n g e n
Figure 1. Left: Northern Europe with Svalbard archipelago in blue. Right: Landsat based
vegetation map over Svalbard developed at Norut
Satellite data
We use the Terra MOD10A1 v.5 product from NASA together with Envisat ASAR wide
swath data from ESA to produce a multisensor/multitemporal snow cover fraction map over
Svalbard. The level-1 ASAR product is classified into wet/dry snow and bare soil using a
standardized change detection technique developed by Norut through several previous projects
covering Scandinavia and is used operationally by Kongsberg Satellite Services. By
combining ASAR snow maps and MODIS snow maps into daily snow cover maps, we reduce
the data gaps due to cloud cover in MODIS data significantly. From this single day product,
we obtain a multi-temporal cloud free snow product by temporal interpolation.
The multisensor/multitemporal time series covers the time span March 2005 to October 2009
where ASAR and MODIS data were both available. We have processed daily snow cover
maps from March 15 to October 15 for each of these years. Snow maps for the winter can not
be calculated due to poor light conditions for optical data and insensitivity to dry snow for Cband SAR-data.
2009_170 ASAR SCA single day
Days without satellite coverage
Threshold: First/last snow day
Figure 2.
Left top: Snow cover fraction vs. day for a single pixel at Svalbard. Days
without satellite coverage are interpolated in time. Left bottom: Total snow cover fraction
for ten years. Right: Single day data from MODIS. Right: Multisensoral/multitemporal
snow cover map (MODIS, ASAR and multitemporal/multisensor combination) for June 17,
2009.
(
!
(
!
< 15. juni
15. juni
27. juni
3. juli
> 15. juli
> 15. juli
(e lle r lite va ria s jon i NDVI)
Uten ve ge ta s jon
(NDVI < 0,1)
Va n
rd
nfjo
Mij e
10 km
en
Sis te dag me d
s nøde kke
< 1. juni
1 - 15. juni
16 - 30. juni
1 - 15. juli
16 - 30. juli
> 1. a ugus t
Mi je
Va n
n fj o
rd e n
10 km
Figure 4. Left: Onset of growing season. Middle: Last day of snow. Right. Average onset
of growing season vs. last day of snow for five years 2005-2009.
Discussion and conclusion
We have provided a five year base line snow climatology from Svalbard and shown the
interconnections between snow coverage and onset and end of the growth season. The time series
will in the future be extended to also cover the years 2000-2005 using only MODIS data and 19822000 using NOAA AVHRR data. The results will also be validated using high-resolution satellite
data.
Acknowledgement
The work has partially been financed by the Norwegian Polar Institute. Data originates from ESA
under Envisat ASAR AO-785 project and NASA.