Postfire energy exchange in arctic tundra: the importance and

Global Change Biology (2011) 17, 2831–2841, doi: 10.1111/j.1365-2486.2011.02441.x
Postfire energy exchange in arctic tundra: the importance
and climatic implications of burn severity
A D R I A N V. R O C H A and G A I U S R . S H AV E R
Marine Biological Laboratory, The Ecosystems Center, Woods Hole, MA 02543, USA
Abstract
Fires produce land cover changes that have consequences for surface energy balance and temperature. Three eddy
covariance towers were setup along a burn severity gradient (i.e. Severely, Moderately, and Unburned tundra) to
determine the effect of fire and burn severity on arctic tundra surface energy exchange and temperature for three
growing seasons (2008–2010) following the 2007 Anaktuvuk River fire. The three sites were well matched before the
fire, experienced similar weather, and had similar energy budget closure, indicating that the measured energy
exchange differences between sites were largely attributable to burn severity. Increased burn severity resulted in
decreased vegetation and moss cover, organic layer depth, and the rate of postfire vegetation recovery. Albedo and
surface greenness steadily recovered with Moderately matching Unburned tundra by the third growing season.
Decreased albedo increased net radiation and partly fueled increased latent and ground heat fluxes, soil temperatures,
and thaw depth. Decreases in moss cover and the organic layer also influenced the ground thermal regime and
increased latent heat fluxes. These changes either offset or decreased the surface warming effect from decreased
albedo, resulting in a small surface warming in Severely and a small surface cooling in Moderately relative to
Unburned tundra. These results indicate that fires have a significant impact on surface energy balance and highlight
the importance of moss and permafrost thaw in regulating arctic surface energy exchange and temperature.
Keywords: Anaktuvuk River fire, burn severity, energy balance, surface temperature
Received 24 January 2011 and accepted 18 March 2011
Introduction
The arctic is experiencing environmental changes that
challenge our ability to predict future conditions at high
latitudes. Increased temperatures may increase shrub
cover, which will result in a significant warming of the
atmosphere with additional implications for permafrost
and sea-ice cover (Sturm et al., 2001; Swann et al., 2010).
Another factor that could feed back strongly on future
northern climate is a change in the fire regime, which
alters land cover for years to decades through changes to
albedo, vegetation cover, surface roughness, and organic
layer depth (Amiro et al., 2006; Liljedahl et al., 2007;
McMillan & Goulden, 2008). These changes in response
to fire will change how solar energy is transferred into the
atmosphere as heat and water, or into the ground as heat.
Fires in the arctic are currently rare and limited by low
temperatures, low fuel loads, and few ignition sources
(Wein, 1976; Hu et al., 2010). Warmer temperatures are
expected to increase the frequency and area of arctic
wildfires (Hu et al., 2010; Kasischke et al., 2010), which
may initiate climate feedbacks that either amplify or
dampen future temperature increases (Jin & Roy, 2005;
Randerson et al., 2006; Euskirchen et al., 2009).
Correspondence: Adrian V. Rocha, e-mail: [email protected]
r 2011 Blackwell Publishing Ltd
Not all fires create equal damage to vegetation and
soils. Postfire vegetation recovery, residual plant and
moss cover, and soil thermal properties are often correlated with the severity of the burn (Brown, 1983; Keeley,
2009; Mack et al., unpublished results). The role of burn
severity in determining postfire energy exchange has
largely been overlooked (Amiro et al., 2006), but may be
particularly important in arctic systems as vegetation and
moss cover plays a large role in regulating both ground
and latent heat fluxes (McFadden et al., 2003; Beringer
et al., 2005; Engstrom et al., 2006; Douma et al., 2007).
Mosses play an important role in evapotranspiration, and
inadequate representation of their dynamics in climate
models results in overestimation of arctic evapotranspiration (Lynch et al., 1999; Eugster et al., 2000). Mosses also
regulate ground thermal properties and prevent permafrost degradation by insulating the ground from surface
heating (Liljedahl et al., 2007; Yi et al., 2009). Permafrost
thaw during the summer acts as a strong heat sink that
reduces surface temperatures and heat flux to the atmosphere (Eugster et al., 2000). Understanding how fire
alters surface biophysical properties and the surface
energy balance thus has important implications for
understanding future regional temperatures in the arctic.
We used three eddy covariance towers located along
a burn severity gradient (Severely Moderately, and
2831
2832 A . V. R O C H A & G . R . S H AV E R
Unburned tundra) in the 2007 Anaktuvuk River fire
scar to determine the effect of fire and burn severity on
energy exchange. The Anaktuvuk River fire was the
largest ever reported on the North Slope of Alaska
(burned area: 1039 km2) and created large areas
(41 km2) that differed in burn severity (Jones et al.,
2009; Boelman et al., 2011). Previous work demonstrated
that our three study sites exhibited similar prefire
vegetation characteristics, experienced similar weather,
and spanned the range of surface conditions observed
across the Anaktuvuk River fire scar (Rocha & Shaver,
2011). These conditions allowed us to determine
how burn severity affects surface energy exchange,
using the Unburned tundra site as a control. Turbulent
and ground heat fluxes, albedo, surface greenness, and
soil and surface temperatures were monitored for three
growing seasons immediately following the Anaktuvuk
River fire. Thaw depth and Terra & Aqua Moderate
Resolution Imaging Spectroradiometer (MODIS) surface temperatures and emissivity were used to corroborate soil and surface temperatures measured by
the eddy covariance towers. We hypothesized that
burn severity influences energy partitioning, energy
exchange, and surface temperatures by impacting biophysical properties such as albedo, moss and vegetation
cover, and organic layer depth.
Methods
Site description and field-based observations
The study was conducted along a burn severity gradient (i.e. Severely, Moderately, and Unburned tundra)
within the southern portion of the Anaktuvuk River
fire scar on the North Slope of Alaska (Jones et al., 2009).
Site details including species composition, vegetation
mortality, prefire vegetation conditions, and site representativeness within the fire scar can be found in Rocha
& Shaver (2011). Briefly, three sites were established
the year following the 2007 growing season before
any vegetation regrowth was observed. Sites had
similar prefire surface greenness [i.e. two band enhanced vegetation index (EVI2)] and were situated
within 7 km of each other. The Severely burned site
(68.991N, 150.281W) was a large (41 km2), gentle
(o5%), generally N-facing hillslope that experienced
30% mortality of the dominant tussock-forming sedge
(Eriophorum vaginatum) and had some areas of exposed
mineral soil. The Moderately burned site (68.951N,
150.211W) was an area of similar area, slope, and
exposure with smaller patches of completely and partially burned tundra intermixed across the landscape
with 5% tussock mortality. The Unburned site (68.931N,
150.271W) was located in a large flat area of tundra that
was unaffected by the fire, and composed of vegetation
that was typical of moist acidic tussock tundra (e.g.
Shaver & Chapin, 1991).
All three sites were located 50 km away from the
nearest road and were accessed by helicopter on a
single day every 3–4 weeks during the growing season
(June–August). During each site visit, we made 76
observations of thaw depth by inserting a rigid metal
rod ( 1 cm diameter) vertically into the soil and measuring the depth at which ice-bonded soil provided firm
resistance. Measurements were taken either from the
top of the moss layer or from the soil surface when moss
was absent. Thaw depth was measured in two areas
next to the eddy covariance towers. Fifty-eight percent
of thaw depth observations occurred every 10 m in a
100 40 m permanent grid located in a representative
area o50 m to the south of the eddy covariance towers.
The remaining thaw depth observations were randomly
taken along four 30 m transects that emanated from the
center of each eddy covariance tower in each cardinal
direction.
Eddy covariance tower instrumentation and data
processing
Flux measurements were made on three stainless steel
towers with identical instrumentation at a height of
2.6 m and were powered by solar panels and car batteries (see Rocha & Shaver, 2011 for further details).
Incoming (i) and reflected (r) solar (SW) radiation, and
incoming and outgoing (o) longwave (LW) radiation
were measured with a CNR-1 Radiometer (Campbell
Scientific; Logan, UT, USA). Incoming and reflected
photosynthetically active radiation were measured with
a silicon quantum sensor (LI-COR; Lincoln, NE, USA).
The EVI2 was calculated using incoming and reflected
radiation measurements and represented surface greenness, which is related to leaf area index and vegetation
cover (Rocha & Shaver, 2009). Air temperature and
relative humidity were measured with an HMP45C-L
sensor (Campbell Scientific) enclosed in a naturally
aspirated radiation shield, while surface temperature
was measured with an Apogee Infrared thermometer,
or calculated from LWo measured by the CNR-1 using
the Stefan–Boltzmann law. Soil temperatures at depths
of 2 and 6 cm were measured with two averaging soil
thermocouples (TCAV-L; Campbell Scientific), while
soil heat flux at a depth of 8 cm was measured with
four soil heat flux plates (HFP01; Campbell Scientific).
Soil instrumentation was installed on June 22, 2008
when the ground had thawed below 10 cm. Data were
logged as half hourly averages on either a CR5000 or a
CR1000 data logger and stored on 2 GB PCMIA card
that was downloaded during each site visit.
r 2011 Blackwell Publishing Ltd, Global Change Biology, 17, 2831–2841
POSTFIRE ENERGY EXCHANGE IN ARCTIC TUNDRA
Turbulent fluxes of momentum, sensible heat, and
latent heat were determined by the eddy covariance
technique (Baldocchi et al., 1988). Half hourly sensible
heat fluxes were calculated as the covariance between
the turbulent departures from the mean of the 10 Hz
vertical wind speed and temperature measured with a
3D sonic anemometer (CSAT3; Campbell Scientific),
while latent heat fluxes were calculated as the covariance between the turbulent departures from the mean
of the 10 Hz vertical wind speed and the H2O mixing
ratio measured with an open path InfraRed Gas Analyzer (LI7500; LI-COR). Ten hertz data were despiked,
rotated to the mean wind streamlines, and half hourly
fluxes were corrected for the density effect due to
sensible heat transfer using EDIRE software (University
of Edinburgh; Moncrieff et al., 1997). Gaps in the sensible and latent heat fluxes from instrument malfunction were filled using weekly parameterized linear
regressions with net radiation. Both sensible and latent
heat fluxes were averaged at the daily time step once a
complete dataset was assembled. Short gaps (o6 h) in
soil temperature and ground heat fluxes were filled
using a linear spline fit and then averaged to the daily
time step. Remaining gaps in the daily average soil
temperature and ground heat and turbulent flux record
were filled using site-to-site correlations (see Rocha &
Goulden, 2010) for each growing season. Site-to-site
correlations at the daily time step provided better
relationships than net radiation, and explained 480%
of the variation in the predictor variable.
Remote sensing data
Daily average surface temperatures obtained at the
towers were corroborated with 8-day 1 km resolution
Terra and Aqua MODIS surface temperature composite
images. Terra and Aqua MODIS satellites have different
overpass times (Wan et al., 2004), and both night and
day surface temperatures from both platforms were
averaged to obtain an 8-day composite daily average
surface temperature for each site. Surface emissivity
was also obtained from MODIS bands 31 and 32
(Petitcolin & Vermote, 2002) and all MODIS collection
5 data were obtained from a single pixel that was
centered on each eddy covariance tower. Periods with
cloud cover or low data quality were identified with
quality control flags provided by the MODIS products
and removed from further analyses.
Data analysis
The land surface energy balance is related to surface
temperature (Ts), albedo (a), and incoming short- (SWi)
and long- (LWi) wave radiation. The close proximity of
2833
our sites to one another means they receive essentially
identical SWi and LWi (Rocha & Shaver, 2011), allowing
comparisons of how differences in their biophysical
properties and surface energy balance influence surface
temperature (for further details see Juang et al., 2007):
2
3
IV
I
II
III
zfflfflfflffl}|fflfflfflffl{
zfflfflfflfflfflfflfflffl
ffl
}|fflfflfflfflfflfflfflffl
ffl
{
zfflfflfflfflffl}|fflfflfflfflffl{
zffl}|ffl{
1
6
7
DTs ¼
4ðTa Ts ÞDZ SWi Da DG sTs4 De5:
4seTs3 þ Z
ð1Þ
The four factors on the right hand side of Eqn (1) (i.e.
I–IV) describe how changes to biophysical properties
and surface energy balance influenced surface temperature changes. D describes the difference in each term
with each land cover conversion using Unburned tundra as the control (i.e. D 5 burned–unburned). A positive D indicates that a given parameter was increased in
burned tundra, while a negative D indicates that a given
parameter was decreased in burned tundra. Term I is
related to changes in latent and sensible heat measured
by eddy covariance, Term II is related to changes in
albedo measured by radiation instruments, Term III is
related to changes in ground heat flux (G) measured by
heat flux plates, and Term IV is related to changes in
surface emissivity (e) measured by MODIS. Z in Eqn (1)
is related to sensible (H) and latent (LE) heat using a
Priestley–Taylor like equation with available energy
(Qa 5 Net radiation [Qn]Gs), the slope of the saturation
vapor pressure–temperature curve (s), the psychrometric constant (g), and the Priestley–Taylor parameter
(aPT) (Priestley & Taylor, 1972; Juang et al., 2007):.
A
zfflffl}|fflffl{
H
Z¼
rCp ga
ðTs Ta Þ
s ¼
LE
1 aPT sþg
1Q
a
B
C
zfflfflfflfflfflfflfflffl}|fflfflfflfflfflfflfflffl{ zfflfflfflfflfflffl}|fflfflfflfflffl
ffl{
H
Qa
ðTs Ta Þ
¼
¼
:
a
ðTs Ta Þ
1 þ HQ
Q
ð2Þ
a
Z was calculated with air density (r), the specific heat
of moist air (Cp), and the bulk aerodynamic conductance (ga). Terms A–C in Eqn (2) determined the sensitivity of Z to incomplete energy budget closure, which is
observed in most eddy covariance studies (Wilson et al.,
2002). We incorporated this uncertainty into Eqn (1) by
calculating Z with Terms A, B and C and then calculated
the expected change in surface temperature from a
conversion of Unburned tundra to either Severely or
Moderately burned tundra with an average of the three
estimates. Surface temperature measurements at the
eddy covariance towers were used to determine how
well Eqn (1) predicted differences in average growing
season surface temperature between burned and unburned tundra in each year.
r 2011 Blackwell Publishing Ltd, Global Change Biology, 17, 2831–2841
2834 A . V. R O C H A & G . R . S H AV E R
Results
Environmental and biophysical conditions
Environmental conditions were similar, while biophysical variables differed among sites and years (Fig. 1,
Table 1). The mean absolute error (MAE) in average
daily incoming shortwave radiation (SWi) between the
Severely and Unburned sites was 1.6 and 0.5 W m2
between the Moderately and Unburned sites over the
three growing seasons. Average daily SWi and air
temperature also did not differ substantially between
years and were high in June and July and low in August
in most years. The start and end of the snow-free season
were within 10 days of each other between years, while
the average snow-free season length was 120 2 days.
Moss cover immediately following fire was 30% in
Moderately and 5% in Severely burned tundra (Rocha
& Shaver, 2011). Combustion related loss of mosses and
the upper soil organic mat also varied between sites
with Moderately burned tundra losing 5 cm and Severely burned tundra losing 8.7 cm of moss and soil
organic matter (Mack et al., unpublished results). Average MODIS surface emissivity was 0.97 and did not
differ among sites and years. Surface roughness exhibited little year-to-year change and was slightly higher in
Severely burned tundra. Albedo and surface greenness
(EVI2) were closely correlated and declined with increased burn severity following fire. Differences between burned and unburned tundra were greatest
during the first postfire growing season and subsequently declined. Albedo and EVI in Moderately were
identical to Unburned tundra, while albedo and EVI in
Severely were slightly lower than Unburned tundra by
the third postfire growing season. Recovering vegetation in burned tundra was mostly composed of resprouting tussocks with little to no moss recovery by
the third postfire growing season.
Energy budget closure and partitioning
Burn severity affected energy partitioning at the three
sites (Fig. 2). A higher proportion of energy was parti-
Fig. 1 Average daily incoming shortwave (SWi) (a–c), air temperature (d–f), albedo (g–i), and surface greenness (EVI2) (j–l) over three
growing seasons immediately following fire. (g)–(l) Albedo and EVI2 in Severely (solid line), Moderately (hatched line), and Unburned
(dotted line) tundra. Albedo and EVI2 are smoothed with an 8-day moving average and gray areas represent 95% confidence intervals.
EVI2, two band enhanced vegetation index.
r 2011 Blackwell Publishing Ltd, Global Change Biology, 17, 2831–2841
POSTFIRE ENERGY EXCHANGE IN ARCTIC TUNDRA
Table 1
2835
Environmental and biophysical conditions at the Anaktuvuk River burn sites from 2008 to 2010
Environmental variables
2008
2009
2010
Average SWi (W m2)*
Average air T ( 1C)*
Snow off (DOY)w
Snow on (DOY)w
Growing season length (days)
185 85
9.3 5
149
269
120
194 77
9.7 5
145
263
116
199 88
10 4
155
271
118
Biophysical variables
S
M
U
S
M
U
S
M
U
Moss cover (%)z
Organic depth (cm)§
MODIS emissivity*
Surface roughness* (m)
o5
12.5
0.97
0.02
30
16.2
0.97
0.02
40
21.2
0.97
0.02
–
–
0.97
0.03
–
–
0.97
0.02
–
–
0.97
0.02
–
–
0.97
0.03
–
–
0.97
0.02
–
–
0.97
0.01
Environmental conditions represent averages for all sites in each year, while biophysical conditions are separated by site and year.
*June through August average.
wObtained from daily MODIS false color images of the North Slope (http://rapidfire.sci.gsfc.nasa.gov/subsets/?subset=AERONET_Barrow) or personal observations. Surface Roughness calculated from Eqn (1) in Chambers et al. (2005) during atmospherically
unstable midday periods (10:00 to 14:00 hours PST).
zEstimate from Rocha & Shaver (2011).
§Estimate from Mack et al. (unpublished results).
S, Severely; M, Moderately; U, Unburned tundra.
tioned into latent heat LE at the two burned sites, resulting in less energy partitioning into sensible heat (H) in
burned relative to unburned tundra. A greater proportion
of energy also went into ground heat (G) in Severely
relative to Moderately burned tundra during the first 2
years. Differences in G partitioning between Moderately
and Unburned tundra were small to nonsignificant in all
3 years. Flux partitioning into G in Severely burned
tundra decreased with time and was similar among sites
by the third postfire growing season. The proportion of
missing energy (M) (i.e. the proportion needed to close
the surface energy budget) was similar to that reported
for other eddy covariance studies (Wilson et al., 2002) and
varied from 0.16 to 0.23. M did not dramatically differ
between sites in each year or exhibit year-to-year trends.
Differences in surface energy exchange
Differences in flux partitioning and net radiation created large differences in energy exchange among sites
(Fig. 3). Growing season averaged daily net radiation
for Unburned tundra was 104 W m2 in 2008, 109 W m2
in 2009, and 112 W m2 in 2010. Increased burn severity
increased net radiation with differences between
burned and unburned tundra decreasing throughout
the growing season and over the 3 years (Fig. 3a–c).
Latent heat fluxes were higher in burned tundra for
most of the first postfire growing season with small
differences between Moderately and Severely burned
Fig. 2 Growing season average flux partitioning of net radiation into latent (LE), sensible (H), and ground (G) heat fluxes, as
well as the missing (M) proportion required for energy budget
closure in Severely, Moderately, and Unburned tundra in the
three growing seasons immediately following fire.
r 2011 Blackwell Publishing Ltd, Global Change Biology, 17, 2831–2841
2836 A . V. R O C H A & G . R . S H AV E R
Fig. 3 Differences in net radiation (Qn: a–c), latent- (LE: d–f), sensible- (H: g–i), and ground-heat fluxes (G: j–l) in Severely (closed
circles/solid line) and Moderately (open circles/hatched line) relative to Unburned tundra over three growing seasons immediately
following fire. Lines represent 8-day moving averages.
tundra (Fig. 3d). Increased latent heat fluxes in burned
tundra were maintained throughout the second and third
postfire growing season, but were most pronounced
in early June and decreased to slightly above zero later
in the season (Fig. 3e–f). Sensible heat flux was lowest in
Moderately burned tundra with a weak seasonal and
small year-to-year declining trend. In Severely burned
tundra, sensible heat flux was similar to that observed in
Unburned tundra during the first 2 years, but slightly
lower than Unburned tundra during the third postfire
growing season (Fig. 3g–i). Ground heat fluxes were
elevated in Severely burned tundra with a slight declining trend over the years, while ground heat fluxes
exhibited small differences to the unburned control in
Moderately burned tundra (Fig. 3j–l).
Differences in soil temperature and thaw depth
The seasonality, magnitude, and rate of soil temperature
changes differed among Severely, Moderately, and
Unburned tundra (Fig. 4). Soil temperatures rose above
freezing earliest in Severely and latest in Unburned
tundra in all 3 years following fire. Soils also froze
about a week earlier in burned tundra in most years.
The rate of degree-day accumulation was highest
in Severely and lowest in Unburned tundra. These
patterns were largely driven by a 3.7–4 1C increase in
average growing season soil temperatures in Severely
and a 1.5–2.3 1C increase in Moderately burned tundra.
Soil temperature increases in burned relative to
unburned tundra were persistent over the 3 years with
small year-to-year differences.
Thaw depth increased with increased burn severity
(Fig. 5) and was deepest in Severely and shallowest in
Unburned tundra in all 3 years following fire (Fig. 5a–c).
Differences in thaw depth between burned and
unburned tundra exhibited strong seasonal and smaller
year-to-year trends (Fig. 5d–f). Differences in thaw depth
between burned and unburned tundra were small during
the start of the growing season, peaked by late June/early
r 2011 Blackwell Publishing Ltd, Global Change Biology, 17, 2831–2841
POSTFIRE ENERGY EXCHANGE IN ARCTIC TUNDRA
Fig. 4 Cumulative soil degree days above freezing (i.e. thawing
degree-days) for Severely (solid line), Moderately (hatched line),
and Unburned (dotted line) tundra, and difference in average
growing season soil temperature (DTsoil) in Severely and Moderately relative to Unburned tundra (inset plot) in three growing
seasons immediately following fire.
July, and decreased thereafter for both Severely and
Moderately burned tundra. The greatest differences in
thaw depth occurred during the first postfire growing
season and decreased slightly over the 3 years.
Differences in surface temperature
Surface temperature differences between burned and
unburned tundra were dependent on burn severity
(Fig. 6). The three independent measures of surface
temperature exhibited similar patterns among sites
2837
and years. Surface temperature differences derived
from the CNR-1 were smaller than those derived from
the Apogee sensor, but both were within the margin of
error for the MODIS data. Averaged growing season
surface temperature across all sensors was 0.6 1C
higher in Severely and 0.1 1C higher in Moderately
relative to Unburned tundra during the first postfire
growing season. Surface temperature differences between burned and unburned tundra decreased from
year-to-year and were 0.2 1C lower in Severely and
0.5 1C lower in Moderately relative to Unburned tundra
by the third postfire growing season. Temperature
differences between burned and unburned tundra
changed sign over the course of the day (Fig. 6 insets).
Surface temperatures in burned tundra were lower than
in unburned tundra during midday (10:00 to 15:00
hours PST) and higher at night (12:00 to 14:00 hours
PST). Midday surface temperature differences were also
more correlated with the daily average (R2: 0.83; n: 6;
P-value o0.01) than at night (R2: 0.31; n: 6; P-value: 0.25),
indicating the role of midday conditions in controlling the
daily average surface temperature difference.
Surface temperature differences resulted from differences in turbulent and ground heat fluxes, albedo, and
emissivity (Fig. 7). Sensible and latent heat fluxes cooled
the surface by 0.1 1C in Severely- and by 0.4 1C in Moderately burned tundra over the three postfire growing
seasons. These changes were counterbalanced by decreased albedo, which warmed the surface by 0.1 1C in
Severely and by 0.1 1C in Moderately burned tundra.
Increased ground heat flux in Severely burned tundra
lead to a surface cooling of o0.1 1C while little to no
cooling occurred in Moderately burned tundra. Emissivity did not increase following fire and its effect on surface
temperature change was negligible. MODIS emissivity
matched emissivity estimates calculated with average
growing season surface temperatures averaged across
sensors (see Fig. 6; MAE: 0.003) using the radiative energy
balance [i.e. Qn 5 (1a)SWi 1 (LWi–seST4S)], providing a
further check on MODIS derived emissivity. Model predictions of annual growing season surface temperature
changes in Severely and Moderately burned tundra
matched the observations when surface temperature
differences were averaged across sensors (Fig. 7 inset)
(R2: 0.81; Slope: 0.87 0.21; P-value: 0.01). Over the three
postfire growing seasons, the model indicated a 0.2 1C
increase in surface temperature in Severely burned tundra and a 0.2 1C decrease in surface temperature in
Moderately burned tundra.
Discussion
We determined the multiyear impacts of fire and burn
severity on surface energy exchange in arctic tundra.
r 2011 Blackwell Publishing Ltd, Global Change Biology, 17, 2831–2841
2838 A . V. R O C H A & G . R . S H AV E R
Fig. 5 Thaw depth (a–c) in Severely (closed circles), Moderately (open circles), and Unburned (closed triangles) tundra, and difference
in thaw depth (d–f) in Severely (closed circles) and Moderately (open circles) relative to Unburned tundra in three growing seasons
immediately following fire.
Sites were well matched before the fire, experienced
similar weather, and had similar energy budget closure,
indicating that the measured energy exchange differences
between sites were largely attributable to burn severity
(Figs 1 and 2; also see Rocha & Shaver, 2011). Differences
in burn severity were realized through changes to the
canopy biophysical properties (Table 1). Increased burn
severity decreased moss and plant cover and charred the
surface, which in turn decreased albedo until leaf area
recovered (Fig. 1). Recovery of albedo and leaf area were
related to burn severity as shown previously (Rocha &
Shaver, 2011), and resulted from differences in plant
mortality and consumption of biomass and soil organic
matter by burning. Soil organic layer thickness, which
plays a large role in determining the surfaces ability to
conduct heat into the ground (Liljedahl et al., 2007),
decreased with increased burn severity (Table 1), while
emissivity and surface roughness, which determine the
efficiency of energy transfer to the near-surface atmosphere, were minimally to unaffected by burn severity.
These changes had important implications for energy
exchange and surface temperature following fire in arctic
tussock tundra.
How did energy exchange change with burn severity and
following fire?
Changes to the canopy biophysical properties of burned
tundra altered energy exchange through changes in net
radiation and energy partitioning (Figs 1–3). Decreased
albedo largely drove increased net radiation in
burned tundra, as differences in net radiation between
burned and unburned tundra were highest in Severely
burned tundra and declined as albedo recovered
to prefire levels (Figs 1 and 3). The lack of a difference
in surface roughness and emissivity between burned
and unburned tundra also emphasized the importance
of albedo in increasing net radiation at the burned sites
(Table 1). The loss of the vegetation canopy in forest
fires dramatically decreases surface roughness, which
results in a decrease in net radiation from reduced
atmosphere and land coupling (Chambers et al., 2005;
Amiro et al., 2006). The increase in net radiation following fire that we observed is unique to tundra, and helps
to fuel increased latent and ground heat fluxes.
Ground heat fluxes, thaw depth, and soil temperatures typically increase following fire in northern landscapes (Rouse & Kershaw, 1971; Haag & Bliss, 1974;
Brown, 1983; Liljedahl et al., 2007), while latent heat
flux typically decreases following fire due to decreased
leaf area and transpiration (Chambers & Chapin, 2002;
Amiro et al., 2006; Liu & Randerson, 2008). Although we
found higher latent heat fluxes at both Moderately and
Severely burned sites in the Anaktuvuk River fire scar,
we believe that these observations are not an artifact of
our experimental design and are consistent with other
independent observations for several reasons. First,
both Severely and Moderately burned tundra exhibited
r 2011 Blackwell Publishing Ltd, Global Change Biology, 17, 2831–2841
POSTFIRE ENERGY EXCHANGE IN ARCTIC TUNDRA
2839
Fig. 7 The contribution of differences in sensible and latent
heat, albedo, ground heat, and emissivity to total surface temperature differences between burned and unburned tundra,
averaged over three growing seasons. Dark bars show the
difference between Severely and Unburned tundra; gray bars
show difference between Moderately and Unburned tundra.
Inset panel shows the observed vs. predicted surface temperature differences from Unburned tundra in each year for the
Severely (closed circles) and Moderately (open circles) burned
tundra. The solid line represents the 1 : 1 line and the hatched
line the regression between predicted and observed surface
temperature change for both sites in all years.
Fig. 6 Average daily surface temperature differences from Unburned tundra as measured by the Apogee (black bar), CNR-1 (gray
bar), and MODIS (dark gray bar) sensors in Severely (top panel) and
Moderately (bottom panel) burned tundra in three growing seasons
immediately following fire. Inset plots show the 3-year average
daytime (10:00 to 14:00 hours) and nighttime (12:00 to 14:00 hours)
surface temperature differences from Unburned tundra for each
sensor at each site. Error bars represent 95% confidence intervals.
increased latent heat fluxes during all 3 years following
fire indicating consistent behavior. Second, similar data
acquisition and processing programs were used to
calculate turbulent fluxes at all sites and sonic anemometers and IRGA’s were calibrated against each other
before each growing season, ensuring high quality
control. Third, the observed changes in surface temperature are consistent with increased latent heat surface cooling at the burned sites (Figs 6 and 7). Surface
temperature differences in Severely and Moderately
burned tundra were lower during the middle of the
day when latent heat flux differences between burned
and unburned tundra were highest (Fig. 6 insets). The
high latent and lower sensible heat fluxes in Moderately
burned tundra are also consistent with the lower
surface temperature changes when compared with
Severely burned tundra (Figs 3 and 6). Lastly, the close
agreement between measured and modeled surface
temperature differences (Fig. 7 inset) provides further
confidence and also indicates that the lack of energy
budget closure had little effect on the conclusions.
Why was latent heat flux high in burned tundra?
Changes to the moss cover, organic layer, and soil
characteristics increased latent heat flux in burned
tundra (Table 1; Figs 2 and 3). Arctic land surfaces are
commonly wet because annual evapotranspiration is
lower than precipitation and the underlying permafrost
prevents deep drainage (Harazono et al., 2003). Mosses
and organic material soak up and store a large part of
this moisture like a sponge (Oechel & Sveinbjornsson,
1978) and lose it through evaporation (Rouse et al., 2000;
Bond-Lamberty et al., 2010). The loss of this spongy
moss and organic layer at the burned sites resulted in
surface water pools that could be evaporated more
freely into the atmosphere. Surface moisture usually
increases following tundra fires (Brown, 1983; Liljedahl
et al., 2007), and surface water pooling is most pronounced following snow melt when water is abundant
and the subsurface is frozen. This may explain the
higher latent heat flux differences between burned
and unburned tundra at the beginning relative to the
end of the growing season (Fig. 3). The slightly elevated
latent heat fluxes toward the later part of the growing
season may have been caused by decreased latent heat
r 2011 Blackwell Publishing Ltd, Global Change Biology, 17, 2831–2841
2840 A . V. R O C H A & G . R . S H AV E R
flux in unburned tundra from moss desiccation. During
dry and sunny periods, radiation loading can dry
mosses from the top to bottom and effectively form a
seal for evaporation through leaf folding (McFadden
et al., 2003; Bond-Lamberty et al., 2010). Another factor
that may have fueled increased latent heat fluxes in
burned tundra is increased soil temperatures (Fig. 4),
which control both evaporation and the stomatal
conductance of arctic plants (Lynch et al., 1999; Starr
et al., 2004).
can lead to a positive radiative forcing and further
warming (Swann et al., 2010). Although less severe
tundra fires may cool the surface through the offsetting
effect of increased latent heat flux, the overall effect is to
warm the atmosphere because of the increase in energy
absorbed by the surface from decreased albedo. These
effects would be most pronounced during the summer
as winter darkness and snow cover would decrease the
differences in albedo that drive surface energy balance
differences between burned and unburned tundra.
Why did burn severity increase soil temperatures and
thaw depth?
Conclusions
Increased burn severity decreased moss cover, organic
layer thickness, and albedo resulting in increased soil
temperatures and thaw depth (Table 1, Figs 4 and 5).
Decreased albedo increased net radiation at the soil
surface and slightly increased ground heat flux and soil
temperature in burned tundra. Decreased organic layer
depth and moss cover decreased thermal insulation
resulting in deeper and warmer soils during the
summer in burned tundra but also more rapid freezing
in fall (Table 1; Figs 4 and 5). Changes in ground
thermal properties created persistent effects on the
ground thermal regime, and deeper thaw depths and
higher soil temperatures were maintained even as albedo differences between sites decreased (Figs 1, 4, and 5).
Changes to the ground thermal properties are expected
to last longer than albedo effects, as mosses, lichens,
and the organic layer may take decades to recover
(Racine, 2004; Mack et al., unpublished results).
What are the climatic implications of arctic fires?
Fires impacted the surface energy balance with important implications for surface temperature. Surface temperature changes in burned tundra were less than a
degree over the three growing seasons and much less
than reported for other ecosystems following fire
(Chambers et al., 2005; Liu et al., 2005). The subtle
changes in surface temperature were partly due to the
increased latent and ground heat fluxes that either
offset or ameliorated the surface warming effect caused
by decreased albedo. Increased latent heat fluxes have
been shown to offset albedo decreases resulting from
land use change (Juang et al., 2007), but it is unclear if
the same effect would occur at larger spatial scales.
Surface temperatures are more coupled to surface
energy exchanges, while regional to global temperatures tend to be more coupled to energy absorbed by the
land and atmosphere. In the present study, the energy
absorbed by the land surface fueled higher latent heat
fluxes, which if mixed thoroughly in the atmosphere
A wildfire in arctic tundra on the North Slope of Alaska
resulted in large changes to the land surface biophysical
properties, postfire surface energy exchange, and surface temperature. These effects were related to burn
severity. Severely burned tundra had lower albedo and
higher net radiation, ground heat fluxes, thaw depth,
and soil and surface temperatures than Moderately or
Unburned tundra (Figs 1, 3–5). Surface temperatures in
Severely were slightly higher than in Unburned tundra,
while surface temperatures in Moderately were slightly
lower than Unburned tundra. Vegetation regrowth led
to rapid recovery of albedo and greenness to prefire
levels by the third postfire growing season in Moderately burned tundra, but fire fundamentally altered
ecosystem energy partitioning through changes to organic layer depth and moss cover. These persistent
effects of burn severity have implications for permafrost
degradation, which can further enhance climate warming through increased greenhouse gas flux (Yoshikawa
et al., 2003; Schuur et al., 2008). Increased temperatures
are expected to fuel a higher fire regime (Hu et al., 2010;
Kasischke et al., 2010), and our results emphasize the
need to understand both the short- and long-term
climate implications of arctic fires.
Acknowledgements
We thank the Toolik Lake Field Station and CH2MHILL for
providing logistics and air support and Jim Laundre, Glenn
Scott, Jen Peters, Andrew Clark, Jeremy Caves, Verity Salmon,
and Marshall Moore for their assistance in the field. This work
was supported by NSF grants #0632139 (OPP-AON), #0808789
(OPP-ARCSS SGER), #0829285 (DEB-NEON SGER), and
#0423385 (DEB-LTER) to the MBL.
References
Amiro BD, Orchansky AL, Barr AG et al. (2006) The effect of post-fire stand age on the
boreal forest energy balance. Agricultural and Forest Meteorology, 140, 41–50.
Baldocchi DD, Hicks BB, Meyers TP (1988) Measuring biosphere–atmosphere exchanges of biologically related gases with micrometeorological methods. Ecology,
69, 1331–1340.
r 2011 Blackwell Publishing Ltd, Global Change Biology, 17, 2831–2841
POSTFIRE ENERGY EXCHANGE IN ARCTIC TUNDRA
2841
Beringer J, Chapin FS III, Thompson CC, McGuire AD (2005) Surface energy exchanges along a tundra-forest transition and feedbacks to climate. Agricultural and
Lynch AH, Chapin FS III, Hinzman LD, Lilly WWE, Vourlitis GL, Kim E (1999) Surface
energy balance on he arctic tundra: measurements and models. Journal of Climate,
Forest Meteorology, 131, 143–161.
Boelman NT, Rocha AV, Shaver GR (2011) Understanding burn severity sensing in
arctic tundra: exploring vegetation indices, sub-optimal assessment timing and the
impact of increasing pixel size. International Journal of Remote Sensing, in press.
Bond-Lamberty B, Gower ST, Amiro B, Ewers BE (2010) Measurement and modeling
of bryophyte evaporation in a boreal forest chronosequence. Ecohydrology, 4, 26–35.
Brown RJE (1983) Effects of fire on the Permafrost ground thermal regime. In: The Role
12, 2585–2606.
McFadden JP, Eugster W, Chapin FS III (2003) A regional study of the controls on
water vapor and CO2 exchange in arctic tundra. Ecology, 84, 2762–2776.
McMillan AMS, Goulden ML (2008) Age-dependent variation in the biophysical
properties of boreal forests. Global Biogeochemical Cycles, 22, GB2019, doi: 10.1029/
2007GB003038.
Moncrieff JB, Massheder JM, de Bruin H et al. (1997) A system to measure surface
of Fire in Northern Circumpolar Ecosystems (eds Wein RW, Maclean DA), pp. 97–110.
John Wiley & Sons Ltd, New York.
Chambers SD, Beringer J, Randerson JT, Chapin FS III (2005) Fire effects on net
radiation and energy partitioning: contrasting responses of tundra and boreal forest
ecosystems. Journal of Geophysical Research, 110, D09106, doi: 10.1029/2004JD005299.
Chambers SD, Chapin FS III (2002) Fire effects on surface-atmosphere energy exchange
fluxes of momentum, sensible heat, water vapour and carbon dioxide. Journal of
Hydrology, 188–189, 598–611.
Oechel WC, Sveinbjornsson B (1978) Photoynthesis of Arctic Bryophytes in Vegetation and
Production Ecology of an Alaskan Arctic Tundra (eds Tieszen LL), pp. 269–289.
Springer, New York.
Petitcolin F, Vermote E (2002) Land surface reflectance, emissivity, and temperature
in Alaskan black spruce ecosystems: implications for feedbacks to regional climate.
Journal of Geophysical Research, 108, D18145, doi: 10.1029/2001JD000530.
Douma JC, Van Wijk MT, Lang SI, Shaver GR (2007) The contribution of mosses to the
carbon and water exchange of arctic ecosystems: quantification and relationships
with system properties. Plant, Cell and Environment, 30, 1205–1215.
Engstrom R, Hope A, Kwon H, Harazono Y, Mano M, Oechel W (2006) Modeling
from MODIS middle and thermal infrared data. Remote Sensing of Environment, 83,
112–134.
Priestley CHB, Taylor RJ (1972) On the assessment of surface heat flux and evaporation
using large scale parameters. Mon. Weather Rev, 10, 81–92.
Racine CH (2004) Tundra fire effects on soils and three plant communities along a
hillslope gradient in the Seward Peninsula, Alaska. Arctic, 34, 71–84.
evapotranspiration in arctic coastal plain ecosystems using a modified BIOME–
BGC model. Journal of Geophysical Research, 111, G02021, doi: 10.1029/2005JG000102.
Eugster W, Rouse WR, Pielke RA Sr et al. (2000) Land–atmosphere energy exchange in
arctic tundra and boreal forest: available data and feedbacks to climate. Global
Change Biology, 6, 84–115.
Euskirchen ES, McGuire AD, Rupp TS, Chapin FS, Walsh JE (2009) Projected changes
in atmospheric heating due to changes in fire disturbance and the snow season in
Randerson JT, Liu H, Flanner MG et al. (2006) The impact of boreal forest fire on
climate warming. Science, 314, 1130–1132.
Rocha AV, Goulden ML (2010) Drought legacies influence the long-term carbon
balance of a freshwater marsh. JGR-Biogeosciences, 115, G00H02, doi: 10.1029/
2009JG001215.
Rocha AV, Shaver GR (2009) Advantages of a two band EVI calculated from solar and
photosynthetically active radiation fluxes. Agricultural and Forest Meteorology, 149,
the western arctic, 2003–2010. Journal of Geophysical Research, 114, G04022, doi:
10.1029/2009JG001095.
Haag RW, Bliss LC (1974) Energy budget changes following surface disturbance to
upland tundra. Journal of Applied Ecology, 11, 355–374.
Harazono Y, Mano M, Miyata A, Zulueta RC, Oechel WC (2003) Inter-annual carbon
dioxide uptake of a wet sedge tundra ecosystem in the Arctic. Tellus B, 55, 215–231.
1560–1563, doi: 10.1016/j.agrformet.2009.03.016.
Rocha AV, Shaver GR (2011) Burn severity influences post-fire CO2 exchange in arctic
tundra. Ecological Applications, 21, 477–489.
Rouse WR, Kershaw KA (1971) The effects of burning on the heat and water regimes of
lichen-dominated subarctic surfaces. Arctic and Alpine Research, 3, 291–304.
Rouse WR, Lafleur PM, Griffis TJ (2000) Controls on energy and carbon fluxes from
Hu FS, Higuera PE, Walsh JE, Chapman WL, Duffy PA, Brubaker LB, Chipman ML
(2010) Tundra burning in Alaska: linkages to climatic change and sea ice retreat.
Journal of Geophysical Research, 115, G004002, doi: 10.1029/2009JG001270.
Jin Y, Roy DP (2005) Fire-induced albedo change and its radiative forcing at the surface
in northern Australia. Geophysical Research Letters, 32, L13401, doi: 10.1029/
2005GL022822.
Jones BM, Kolden CA, Jandt R, Abatzoglou JT, Urban F, Arp CD (2009) Fire behavior,
select high-latitude wetlands: controls and extrapolation. Global Change Biology, 6,
59–68.
Schuur EAG, Bockheim J, Canadell J et al. (2008) Vulnerability of permafrost
carbon to climate change: implications for the global carbon cycle. BioScience, 58,
701–714.
Shaver GR, Chapin FS III (1991) Production/biomass relationships and element
cycling in contrasting arctic vegetation types. Ecological Monographs, 61, 1–31.
weather, and burn severity of the 2007 Anaktuvuk River Tundra Fire, North Slope,
Alaska. Arctic, Antarctic, and Alpine Research, 3, 309–316.
Juang JY, Katul G, Siqueira M, Stoy P, Novick K (2007) Separating the effects of albedo
from eco-physiological changes on surface temperature along a successional
chronosequence in the southeastern United States. Geophysical Research Letters, 34,
L211408, doi: 10.1029/2007GL031296.
Starr G, Neuman DS, Oberbauer SF (2004) Ecophysiological analysis of two arctic
sedges under reduced root temperatures. Physiologia Plantarum, 120, 458–464.
Sturm M, Racine C, Tape K (2001) Climate change: increasing shrub abundance in the
arctic. Nature, 411, 546–547.
Swann AL, Fung IY, Levis S, Bonan GB, Doney SC (2010) Changes in arctic vegetation amplify high-latitude warming through the greenhouse effect. PNAS, 107,
Kasischke ES, Verbyla DL, Rupp TS et al. (2010) Alaska’s changing fire regimeimplications for the vulnerability of its boreal forests. Canadian Journal of Forest
Research, 40, 1313–1324.
Keeley JE (2009) Fire intensity, fire severity and burn severity: a brief review and
suggested usage. International Journal of Wildland Fire, 18, 116–126.
Liljedahl A, Hinzman L, Busey R, Yoshikawa K (2007) Physical short-term changes
after a tussock tundra fire, Seward Peninsula, Alaska. Journal of Geophysical
1295–1300.
Wan Z, Zhang Y, Zhang Q, Li ZL (2004) Quality assessment and validation of the
MODIS global land surface temperature. International Journal of Remote Sensing, 25,
261–274.
Wein RK (1976) Frequency and characteristics of arctic tundra fires. Arctic, 29, 213–222.
Wilson K, Goldstein A, Falge E et al. (2002) Energy balance closure at FLUXNET sites.
Agricultural and Forest Meteorology, 113, 223–243.
Research-Earth Surface, 112, F02S07, doi: 10.1029/2006JF000554.
Liu H, Randerson JT (2008) Interannual variability of surface energy exchange
depends on stand age in a boreal forest fire chronosequence. Journal of Geophysical
Research, 113, G01006, doi: 10.1029/2007JG000483.
Liu H, Randerson JT, Lindfors J, Chapin FS (2005) Changes in the surface energy budget
after fire in boreal ecosystems of interior Alaska: an annual perspective. Journal of
Yi S, McGuire AD, Harden J et al. (2009) Interactions between soil thermal and
hydrological dynamics in the response of Alaska ecosystems to fire disturbance.
Journal of Geophysical Research, 114, G02015, doi: 10.1029/2008JG000841.
Yoshikawa K, Bolton WR, Romanovsky VE, Fukuda M, Hinzman LD (2003) Impacts of
wildfire on the permafrost in the boreal forest of Interior Alaska. Journal of
Geophysical Research, 108, D18148, doi: 10.1029/2001JD000438.
Geophysical Research – Atmospheres, 110, D13101, doi: 10.1029/2004JD005158.
r 2011 Blackwell Publishing Ltd, Global Change Biology, 17, 2831–2841