Agricultural and Forest Meteorology 161 (2012) 107–115

Contents lists available at SciVerse ScienceDirect

Agricultural and Forest Meteorology

journal homepage: www.elsevier.com/locate/agrformet

The role of permafrost in water exchange of a black spruce forest in Interior

Alaska
Hiroki Iwata a,∗ , Yoshinobu Harazono a,b , Masahito Ueyama b
a
International Arctic Research Center, University of Alaska Fairbanks, Fairbanks, AK 99775-7340, USA
b
Graduate School of Life and Environmental Sciences, Osaka Prefecture University, Sakai, Osaka 599-8531, Japan

a r t i c l e i n f o a b s t r a c t

Article history: Water exchange in a black spruce forest on ice-rich permafrost in Interior Alaska was examined over
Received 22 June 2011 seven years to clarify its seasonal and interannual variations. Partitioning of available energy to latent
Received in revised form 9 February 2012 heat flux was strongly constrained by leaf area and thaw depth of the soil. Water balance on a monthly
Accepted 20 March 2012
timescale showed that evapotranspiration was greater than precipitation in the early growing season.
It is likely that evapotranspiration was supported by soil thawing and snowmelt water stored in micro-
Keywords:
topographic hollows during this period. Precipitation increased in the latter half of the growing season,
Eddy covariance
whereas evapotranspiration decreased along with a decrease in net radiation and the leaf maturation
Energy balance
Evapotranspiration
and the senescence of understory deciduous vegetation. This resulted in an increase of stored water in
Seasonal freeze–thaw cycle the soil, which can be utilized by vegetation in the following growing season. The presence of ice-rich
Snowmelt permafrost helps the accumulation of water in the soil by blocking deep percolation of water. Cumulative
evapotranspiration during the snow-free period did not show noticeable interannual variability, in con-
trast to a significant variation in precipitation. Cumulative evapotranspiration was primarily controlled
by the available energy. Water resources for this black spruce forest remained sustainable under the
current climate due to net water gain in the late growing season, combined with the presence of ice-rich
permafrost. From a comparison to other black spruce forests with non-permafrost soil, we found that
whether ice-rich permafrost exits or not is an important factor for determining the magnitude in the
interannual variation of evapotranspiration in boreal forests.
© 2012 Elsevier B.V. All rights reserved.

1. Introduction increased precipitation. However, it is still debatable whether evap-

otranspiration is increasing or decreasing (Ryu et al., 2008, and
Water is a fundamental resource for all ecosystems, but it is par- references therein). In terrestrial ecosystems, evapotranspiration
ticularly important for forests, which generally grow in a climate is controlled not only by temperature, but also by other factors
where precipitation provides more water than the vegetation can such as radiation, atmospheric water vapor pressure deficit, soil
use (Waring and Running, 1998). A change of water availability for moisture, and leaf area. Jung et al. (2010) revealed that global ter-
vegetation affects surface energy partitioning on a short timescale restrial evapotranspiration decreased due to the limitation of soil
and vegetation dynamics on a longer timescale. In turn, regional cli- moisture in southern hemisphere in the last decade. Soil moisture
mate is influenced through changes in the surface fluxes of sensible varies with a longer timescale compared to other meteorological
heat and water vapor. variables. It has a memory effect during water exchange, influ-
Regarding water balance, precipitation and evapotranspiration encing evapotranspiration not only during the current year but
are two major components, and their differences may determine the following year as well (Ohta et al., 2008; Maximov et al.,
water availability. Observational evidence indicates that precipi- 2008). Hence, detailed understanding of the response of vegeta-
tation increased in northern high latitudes during the twentieth tion to long-term variability in climate is needed to predict both
century (Dai et al., 1997; Serreze et al., 2000; Kattsov and Walsh, future evapotranspiration from vegetation and water availability in
2000). The concurrent increase in air temperature (e.g., Serreze the soil.
et al., 2000; Jones and Moberg, 2003; Trenberth et al., 2007) At northern high latitude, permafrost is a characteristic soil con-
may have increased evapotranspiration and compensated for this dition. It is known that permafrost and the seasonal freeze–thaw
cycle of the active layer play an important role in controlling the
seasonal variation of hydrological processes at these latitudes. For
∗ Corresponding author. Tel.: +1 907 474 7337; fax.: +1 907 474 2691. example, large runoff occurs from snowmelt in the spring (Glenn
E-mail address: hiwata@alaska.edu (H. Iwata). and Woo, 1997) because infiltration of snowmelt water into frozen

0168-1923/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.agrformet.2012.03.017
108 H. Iwata et al. / Agricultural and Forest Meteorology 161 (2012) 107–115

soil is limited. Plant growth, and related transpiration, is controlled 3

by soil thawing (Ohta et al., 2001). Ice-rich permafrost impedes Average
infiltration, resulting in wet to saturated soil (Hinzman et al.,
2006). However, there is still a lack of knowledge about how water 2
exchange responds to long-term variability in climate in permafrost

LAI
regions. This lack is partly due to the severe environment that faces
continuous observation. In a long-term study in Siberia, Ohta et al. 1
(2008) discussed that trees use meltwater from the lower part of
the active layer during years with low precipitation (Sugimoto et al.,
2002) and that resulted in a small interannual variation of evapo- 0
transpiration in a larch forest on permafrost. 0 50 100 150 200 250 300 350
In North American boreal forests, black spruce is the dominant 100
species (O’Connell et al., 2003; Chapin et al., 2006), especially in per- 80 2003
mafrost areas. However, many study sites in black spruce forests in 2004

thaw depth (cm)

60 2005

Snow and soil

Canada (Jarvis et al., 1997; Pattey et al., 1997; Arain et al., 2003;
Amiro et al., 2006b; Barker et al., 2009; Zha et al., 2010) and Alaska
40 2006
20 2007
(Liu et al., 2005; Liu and Randerson, 2008), where evapotranspi- 2008
ration was observed with the eddy covariance technique, are not 0 2009
underlain by permafrost. One exception is a study by Ueyama et al. -20
(2006) in an Alaskan black spruce forest. However, that research -40
focused on carbon dioxide exchange. Hence, the role of permafrost -60
in water exchange in black spruce forests over a seasonal to inter- 0 50 100 150 200 250 300 350
annual timescale has yet to be clarified. DOY
In this article, we report variations of water flux observed with
the eddy covariance technique over seven years in a black spruce Fig. 1. Seasonal variations of leaf area index, LAI (upper panel), and snow depth and
forest on ice-rich permafrost in Interior Alaska. Specific objectives soil thaw depth (lower panel).
of this study are (1) to examine the seasonal and interannual vari-
ations of evapotranspiration in a boreal black spruce forest on
ice-rich permafrost and (2) to identify environmental variables that the average annual precipitation was 263 mm yr−1 (Shulski and
explain the variations in evapotranspiration. We also discuss the Wendler, 2007). The observation site was typically free of snow
role of permafrost and the seasonal freeze–thaw cycle of the active from late April through early October (Fig. 1).
layer in water exchange between the black spruce forest and the
atmosphere, by comparing our results with other studies in black 2.2. Observation
spruce forests without permafrost. This knowledge is necessary to
predict water balance in boreal forests with permafrost, in consid- Fluxes of momentum, sensible heat, and water vapor were
eration of projected climate warming. observed at a height of 6 m above the ground using an ultrasonic
anemometer (CSAT3, Campbell Scientific, USA) and an open-path
infrared gas analyzer (LI-7500, Li-Cor, USA). Turbulence data were
2. Observations and analysis recorded at 10 Hz with a DRM3 datalogger (TEAC, Japan) until the
end of 2008, and with a CR3000 datalogger (Campbell Scientific,
2.1. Study site USA) beginning in 2009.
Mean air temperature and relative humidity were measured
The data were obtained in a black spruce (Picea mariana) forest, at 1, 2, 4, and 8 m above the ground using ventilated platinum-
64◦ 52 N, 147◦ 51 W, 155 m a.s.l., which stands on ice-rich per- resistance thermometers and capacitive hygrometers (HMP45AC,
mafrost in Fairbanks, Alaska, USA (Ueyama et al., 2006, 2009; Iwata Väisälä, Finland). Micrometeorological data such as short-wave
et al., 2010). The forest is approximately 120 years old (Vogel et al., radiation (PCM-03 or CMP3, Kipp & Zonen, The Netherlands); net
2005), with an approximate mean canopy height of 3 m. The tree radiation (Q7.1, REBS, USA); rainfall (TR-525M-R3, Texas Elec-
density is 4500 trees ha−1 ; however, due to the narrow canopy tronics, USA); atmospheric pressure (PTB101B, Väisälä, Finland);
architecture of black spruce, the forest canopy is relatively open. ground surface temperature (IT450N, Horiba, Japan); soil temper-
The forest floor has a pronounced hummock-hollow microtopogra- ature (thermocouple); and soil water content (CS616, Campbell
phy. The understory is dominated by low evergreen shrubs (Ledum Scientific, USA) were obtained at and around the tower. The soil
groenlandicum, Vaccinium vitis-idaea), deciduous shrubs (Vaccinium water content sensor, with 30-cm-long probe rods, was inserted
uliginosum, Salix spp., and Betula glandulosa), and sedges (Carex vertically into the soil. Ground heat flux was observed at 10-cm
species). The ground is almost completely covered with mosses depth at four points around the tower, using heat flux plates (HFT3-
(Hylocomium splendens, Sphagnum capillifolium, and Tomenthyp- L, REBS, USA). These mean variables were observed at 10-s intervals,
num nitens). Leaf area index (LAI), measured with a plant canopy and the 30-min averages were recorded by dataloggers (CR23X and
analyzer (LAI-2000, Li-Cor, USA), varied from 0.2 m2 m−2 , when the CR10X, Campbell Scientific, USA).
ground was covered with snow, to 1.9 m2 m−2 during mid-summer Other manual observations such as active layer depth, soil water
(Fig. 1). The black spruce overstory is evergreen, so most of this vari- content (SWC), LAI, and snow depth were conducted approximately
ation in LAI can be explained by leaf growth and fall of understory once a week around the observation tower. The active layer depth
deciduous vegetation. The soil is silt-loam overlain by an organic was measured at ten points with five measurements at each point.
layer of 25–45 cm (Heijmans et al., 2004), and it is poorly drained SWC was observed in the same way as the observation of active
due to the presence of ice-rich permafrost. The depth of the active layer depth, using a water content sensor (CS620, Campbell Sci-
layer is 40–50 cm (Fig. 1). entific, USA) with 20-cm-long probe rods. When the thaw depth
The mean monthly air temperature in Fairbanks between 1971 was shallow, probe rods were inserted slantwise so that SWC in
and 2000 ranged from −23.2 ◦ C in January to 16.9 ◦ C in July, and the thawed soil layer was observed. LAI was observed at eight
 H. Iwata et al. / Agricultural and Forest Meteorology 161 (2012) 107–115 109

Table 1
Percentage of data gap-filled. The numbers in parentheses for net radiation indicate the percentage during the snow-cover period. The numbers in parentheses for turbulent
fluxes indicate the percentage during nighttime (solar radiation <10 W m−2 ).

Year 2003 2004 2005 2006 2007 2008 2009

Micrometeorology
Air temperature 2 8 1 2 1 3 3
Relative humidity 2 10 1 2 1 0 0
Rainfall 0 2 0 0 0 0 0
Net radiation 4 (3) 26 (10) 8 (7) 8 (6) 5 (5) 11 (10) 4 (3)
Soil heat flux 0 17 0 2 0 0 0
Turbulent flux
Sensible heat flux 66 (43) 66 (42) 37 (26) 68 (44) 62 (45) 59 (43) 58 (43)
Latent heat flux 74 (48) 78 (48) 82 (49) 74 (48) 67 (47) 67 (47) 66 (47)

points with ten measurements at each point. The snow depth was up to 10, 20, 30, and 40 cm; beginning of decline in LAI; and the
observed at three points. beginning of snow cover on the ground. The period of snowmelt
was determined as the days with a significant decrease in ˛H com-
2.3. Analysis pared to previous days during the snow cover period. The lengths
of intervals ranged from ten days (for the snowmelting period) to
2.3.1. Flux calculation five months (for the snow-cover period).
Turbulent fluxes were calculated with the eddy covariance
method for each half-hour period. Before calculation, data spikes 2.3.3. Calculation of bulk surface conductance
were removed (Vickers and Mahrt, 1997) and a rotation of the The bulk surface conductance (m s−1 ), gS , was calculated
sonic anemometer coordinate into a mean streamline coordinate by inverting the Penman–Monteith equation (Monteith, 1965;
was performed (Mahrt et al., 2000). The fixed streamline coordi- Monteith and Unsworth, 1990) as follows:
nate was used to prevent erroneous tilt angle correction due to low
1 1
 H  cp
wind speed, which prevails in the study site. The mean stream- = −1 + D, (2)
gS ga  E E
line coordinate was determined for the entire observation period.
The calculated fluxes were corrected for high-frequency attenu- where  is the slope of relation between saturation vapor pres-
ation due to sensor separation and line averaging (Horst, 1997; sure and air temperature (hPa K−1 ),  is psychrometric constant
Massman, 2000), influence of water vapor on temperature mea- (hPa K−1 ),  is air density (kg m−3 ), cp is specific heat capacity for
surement using the ultrasonic anemometer, and the effect of air constant pressure (J kg−1 K−1 ), D is vapor pressure deficit (hPa), and
density fluctuation on water vapor density measurement (Webb ga is aerodynamic conductance (m s−1 ), calculated as follows:
et al., 1980). For data selection, the data containing measurement
1 u
errors related to rain events and those exhibiting obvious non- = 2 + rb , (3)
stationarity were discarded before the analysis. Furthermore, data
ga u∗
when friction velocity (m s−1 ), u* , is smaller than 0.1 m s−1 were where u is mean horizontal wind speed (m s−1 ) and rb is additional
rejected for both daytime and nighttime. Below this u* threshold resistance (s m−1 ) for scalars relative to momentum calculated
value, the sum of sensible heat flux (W m−2 ), H, and latent heat flux using an empirical equation proposed by Thom (1972): rb =
(W m−2 ), E—where  is latent heat for vaporization (J g−1 ) and E −2/3
6.3u∗ . It should be noted that the coefficient in Thom (1972)
is evapotranspiration (g m−2 s−1 )—was less than 50% of available
was determined for u* in cm s−1 . For the calculation of gS , only
energy (W m−2 ), A (defined below), on average.
data obtained at times when the canopy was dry were used. It was
assumed that the canopy became dry in 12 h after rainfall. Only
2.3.2. Energy balance and energy partitioning observed data were used for the calculation of gS , and gap-filled
In this study, available energy, A, was evaluated as follows: data described below were not used unless specified.
A = Rn − G − J, (1)
2.3.4. Gap-filling of data
where Rn is net radiation (W m−2 ), G is heat flux into the soil or snow For meteorological variables except for rainfall, small gaps such
(W m−2 ), and J is the sum of sensible and latent heat storage in the as 1 h were interpolated linearly. Larger gaps were filled using rela-
atmospheric column below the level of flux observation (W m−2 ). tionships between missing data and related data, if available. For
The storage term, J, was determined from the difference of mean air example, if air temperature at 8 m was missing and air tempera-
temperature and humidity for one 30-min mean from the previous ture at 4 m was available, the relationship between air temperature
one. at 8 m and 4 m was used to fill the gap. The relationships were
The closure ratio of energy balance, ˛, is often used to evaluate obtained from approximately seven-day data around the data gaps.
the accuracy of fluxes observed with the eddy covariance tech- The remaining gaps were filled with a mean diurnal cycle around
nique. To evaluate ˛ and energy partitioning of A between H and E data gaps. Rainfall data were filled using data obtained at the
(denoted as ˛H and ˛E , respectively), half-hourly H and E were Fairbanks Airport, which is 6 km from the observation site. Also,
plotted against half-hourly A. A linear function, y = ˛x + ˇ, was fit- precipitation as snow during winter was filled using airport data,
ted to the data, and then ˛H , ˇH and ˛E , ˇE were determined for since the precipitation gauge used in this study was for collecting
H and E, respectively (Malhi et al., 2002). ˛ was calculated as the rainfall data only. All gaps in atmospheric pressure data were filled
sum of ˛H and ˛E . linearly.
The analysis above was conducted for several intervals within a Net radiation is a large component in the energy balance Eq.
year, in order to examine the characteristics of seasonal variations (1). However, observation of net radiation was affected by snow
of ˛, ˛H , and ˛E . The intervals were determined according to physi- cover on the sensor itself after snowfall. Net radiation when the
cal and phenological events using snow depth, thaw depth, and LAI, sensor was covered with snow was estimated as follows. First, peri-
as follows: beginning of snowmelt; end of snowmelt; soil thawing ods with snow cover on the radiation sensors were detected as
110 H. Iwata et al. / Agricultural and Forest Meteorology 161 (2012) 107–115

20 20

Mean air temperature, Net radiation, Rn

(MJ m day )
-1
15
10 10

-2
5
0 0
-5

Mean vapor pressure

20
20 Ta D

deficit, D (hPa)
15
0
Ta ( oC) 10
-20
5
-40
0
30 1.0
Continuous

Mean soil water

Precipitation, P

Manual

content, SWC
0.8
(mm day )

SWC
-1

20 SWC
0.6
0.4
10
0.2
0 0.0
2003 2004 2005 2006 2007 2008 2009
Year

Fig. 2. Daily meteorological variables and soil water content, SWC, observed during the seven years.

periods of unreasonable albedo calculated from short-wave radi- was estimated as an average value obtained from the period before
ation. Second, incident short-wave radiation during sensor snow and after the data gap. Then E was estimated from the estimated
cover was estimated using reflected short-wave radiation and an ˛E , ˇE , and A.
albedo value observed a few days later when the snow on the sen- Generally, large amounts of data during nighttime were gap-
sor was removed. Third, net radiation during sensor snow cover filled (Table 1) due to suppressed turbulence conditions, which
was estimated from a relationship between net radiation and net may be common in a low-wind environment. To test the gap-
short-wave radiation obtained when the sensor was not covered filling strategy, half of the data were artificially removed from
with snow. Table 1 summarizes the percentage of data gap-filled. the snow-free period in 2009, and then gap-filled data were
Most micrometeorological data were obtained nearly continuously, compared to the original data. A linear regression analysis
with a few data gap percentages, except for 2004. Relatively many indicated that y = 1.00x + 0.53 (R2 = 0.94 and root mean square
data concerned with net radiation during the snow-cover period error, RMSE = 21.05 W m−2 ) and y = 0.98x + 0.13 (R2 = 0.81 and
were gap-filled, due to snow cover on the radiation sensor. RMSE = 17.45 W m−2 ) for H and E, respectively. This degree of
To obtain continuous flux data, gaps in H and E were filled using accuracy may also be achieved for large gaps as described above,
A and ˛H , ˛E , ˇH , and ˇE (Gu et al., 2005; Amiro et al., 2006a) provided that the parameters (˛H,E and ˇH,E ) are reasonable.
obtained in the analysis described in Section 2.3.2. In 2004, there
was a problem with the datalogger during DOY149–187, and both 3. Results
H and E were unavailable. To fill this gap, ˛H,E and ˇH,E dur-
ing this period were assumed to be equal to those in the same 3.1. Meteorological conditions over the observation period
period in 2003, since the moisture conditions during the corre-
sponding season in 2003 were similar to those in 2004. Then, H and During the seven-year observation period, large variations of
E were estimated from A and assumed ˛H,E and ˇH,E . In 2005, the meteorological conditions were observed (Fig. 2 and Table 2), espe-
infrared gas analyzer had a problem during DOY133–246, and E cially in rainfall. The years 2004 and 2009 were dry, with lower
was not available. For this period, ˛E was estimated as the differ- rainfall and a higher water vapor pressure deficit, D. Especially
ence between observed ˛H during the period and ˛ averaged for the dry conditions occurred in 2004, when there were 40 consecutive
growing season (0.80) over the entire observation period, and ˇE days, June 12 through July 21 (Table 3), with no measurable rainfall.

Table 2
Summary of meteorological conditions over the observation period. 1971–2000 mean data were obtained from the Alaska Climate Research Center
(http://climate.gi.alaska.edu/).

Year (Jan–Dec)

2003 2004 2005 2006 2007 2008 2009 1971–2000 mean

Net radiation (GJ m−2 y−1 ) 1.76 1.63 1.69 1.50 1.75 1.50 1.60 N/A
GDD with 5◦ C threshold (◦ C day) 898.6 1193.4 1105.9 1036.2 1240.9 945.5 1154.0 1048
Mean vapor pressure deficit (hPa) 3.08 3.51 3.28 2.93 3.28 2.79 3.51 N/A
Rainfall (mm) 223.1 124.1 237.1 203.2 273.6 224.5 167.8 N/A
Precipitation (mm) 303.1 198.7 279.4 238.4 326.2 317.2 254.2 262.6
Number of snow-free days 187 174 164 169 170 147 174 N/A

Winter

02–03 03–04 04–05 05–06 06–07 07–08 08–09

Snowfall (mm water equivalent winter−1 ) 61.9 80.2 89.1 37.0 32.6 73.1 98.1
 H. Iwata et al. / Agricultural and Forest Meteorology 161 (2012) 107–115 111

Table 3
80

E and P (mm month )

-1
Major consecutive days with no measurable rainfall during the snow-free period
70 P (a)
(>10 days). The ends of consecutive days are indicated in parentheses.
60 E
Year Number of consecutive days
50
2003 13 (May 11), 18 (May 31), 18 (Jun 25), 14 (Sep 25), 14 (Oct 25) 40
2004 12 (May 5), 40 (Jul 21), 11 (Aug 12), 11 (Aug 25), 10 (Sep 12)
2005 14 (Aug 20)
30
2006 None 20
2007 15 (May 3) 10
2008 23 (Sep 9)
0
2009 11 (May 22), 10 (Jun 16), 10 (Jul 8), 15 (Jul 30), 17 (Oct 25)
1 3 5 7 9 11
Month
Widespread fires in Alaska in 2004 resulted in the largest fire year 1

Coefficient of variation
since record keeping began in 1950 (Todd and Jewkes, 2006). In E /E po (b)
contrast, 2007 was a wet year, with higher rainfall. Even so, it was 0.8 P
warmer than average in that year, and net radiation was higher. The 0.6
years 2003 and 2008 were cooler, with an intermediate amount of
rainfall. In 2008, late snowmelt and an early beginning of snow 0.4
accumulation (Fig. 1) resulted in the least number of snow-free
days.
0.2
0
3.2. Energy balance and partitioning 3 5 7 9
Month
The closure ratios of energy balance, ˛, were obtained as
described in Section 2.3.2, and the ratios were typically close to 0.8 Fig. 4. Seasonal variation of (a) monthly evapotranspiration, E, normalized with
(Fig. 3a) for this site during the growing season. Around DOY110, ˛ potential evapotranspiration, Epo , and (b) coefficients of variation for precipitation,
decreased to 0.4 at the minimum. This is because absorbed energy P, and E/Epo .

was utilized to melt snow accumulated on the ground (Fig. 1), and
˛H was highly reduced (Fig. 3b). During mid-winter (approximately
from minute H and A during this period. In reality, it is expected
November through February), ˛ was more variable, but it tended
that ˛H and ˛E , during the coldest period in winter, do not change
to be greater than that for the growing season.
between years, due to the extreme cold at the observation site.
Latent heat used for snowmelt is an important component to the
Hence, all midwinter data were gathered, and a single pair of ˛H,E
energy balance in the forest. This energy was estimated from the
and ˇH,E was determined. These were then used to fill the data
snow water equivalent at the beginning of snowmelt and the latent
gaps for the entire midwinter period.
heat of fusion. The snow water equivalent was obtained from snow
After the snow melted, about two-thirds of A was partitioned
depth and its typical density during snowmelt observed at this site
into H in the early growing season. With increased thaw depth and
(250 kg m−3 ). The latent heat used for snowmelt was typically 40%
LAI (Fig. 1), more A was partitioned into E during the latter half of
of A, and inclusion of this energy pushed ˛ up to 0.9.
the growing season (July and August). The midday (11:00–15:00,
Although the calculated ˛H during midwinter tended to vary
local noon is 13:00) Bowen ratio was the lowest (1.1–1.4) in July,
between years, this may be related to errors in determining ˛H
when LAI was at a maximum. This value of minimum Bowen ratio is
similar to that observed in black spruce forests in Alaska (Liu et al.,
1.2 2005) and at lower latitude in Canada (Jarvis et al., 1997; Pattey
(a) et al., 1997). In the late growing season, the fraction partitioned to
1 E decreased, along with a decrease of LAI.
0.8
α

0.6 3.3. Seasonal cycle of water balance

2006
0.4 2003 2007
2004 Fig. 4a shows mean seasonal cycles of evapotranspiration, E, and
0.2 2008
2005 precipitation (mm), P, over seven years. From November through
2009
0 March, E was very low due to low Rn and low air temperature, and
was lower than P. During this period, P usually occurred as snowfall,
1 and the snow accumulated on the ground. Regardless of the lim-
(b) ited P in the early growing season, E tended not to be limited and
αH
αH or αλE

0.8
to exceed P. Maximum monthly E was observed in July for most
0.6 years. Typical daily maximum E was 2.5 mm day−1 . Thereafter, E
was lower than P, a difference which resulted in an increase of SWC
0.4 at the end of the growing season (Fig. 2).
αλE
0.2 The first half of the growing season (May and June) had low P,
and P increased in the latter half (July and August). This is a typ-
0 ical seasonal cycle of P in Interior Alaska (Shulski and Wendler,
0 50 100 150 200 250 300 350 2007). However, the distribution of P within a year varied between
DOY years. For example, the years 2003 and 2006 were typical of this
pattern, but 2004 had more P in May compared to other years,
Fig. 3. Seasonal variations of (a) the closure ratio of energy balance, ˛, (b) energy and P during the rest of the growing season was extremely low
partitioning of available energy to sensible heat, ˛H , and latent heat fluxes, ˛E . (Fig. 2). In contrast, the seasonal variation of E did not change
112 H. Iwata et al. / Agricultural and Forest Meteorology 161 (2012) 107–115

8.0 From June through August, midday gS negatively correlated with

May (a) midday water vapor pressure deficit, D (p = 0.02, Fig. 5b), but it did
gs (mm s )

6.0 Jun
-1

not show a clear correlation with soil water content, SWC (p = 0.13,
Jul
4.0 Aug Fig. 5c). The independence of midday gS from SWC may be due to
Sep the fact that SWC in this forest was generally high, and did not
2.0 influence the stomatal conductance strongly.
The maximum monthly average of midday gS was approxi-
0.0 mately 6 mm s−1 . This value is similar to that reported for black
0.3 0.4 0.5 0.6 0.7 0.8 0.9
spruce forests in Canada (Zha et al., 2010).
NDVI
7.0 3.5. Annual water balance
(b)
6.0
gs (mm s )
-1

During winter, air temperature was extremely low and net radi-
5.0
ation was small, resulting in small magnitude and interannual
4.0 y =-0.16 x+6.82 variation of E. So we concentrate on the snow-free period to exam-
2
3.0 R =0.34 ine cumulative E and its interannual variation.
p=0.02 In general, it is expected that cumulative E during the snow-
2.0
0 5 10 15 20 25 free period is affected by climate variability and duration of the
snow-free period. However, cumulative E at the site did not change
D (hPa) remarkably compared to P during the seven years (Fig. 6). The
7.0 cumulative E during the snow-free period ranged from 195.1 mm
y =3.60x +3.03 (c) to 233.6 mm (average of 210.9 mm with a coefficient of variation
6.0
gs (mm s )

2
-1

R =0.21 of 0.07), whereas cumulative P ranged from 137.1 mm to 273.6 mm

5.0 p=0.13 (average of 202.2 m with a coefficient of variation of 0.22).
4.0 The relationships between cumulative E and environmental
variables during the snow-free period were examined. A strong cor-
3.0
relation between E and A (R2 = 0.62, p < 0.05) was found. Cumulative
2.0 E was also correlated with SWC and the number of snow-free days,
0.2 0.3 0.4 0.5 0.6 0.7 0.8 but the correlations were not statistically significant (R2 = 0.55,
SWC p = 0.15 for SWC and R2 = 0.30, p = 0.20 for the number of snow-free
days). Cumulative E was not controlled by the amount of cumula-
Fig. 5. (a) The relationship between the midday bulk surface conductance, gS and
NDVI; (b) the relationship between midday gS and midday vapor pressure deficit,
tive rainfall or cumulative precipitation including snowfall in the
D, during June–August; and (c) the relationship between midday gS and soil water previous winter. In addition, the ratio of E to Epo did not correlate
content, SWC, during June–August. with P (R2 = 0.03) or SWC (R2 = 0.05). This indicates that radiation
is an important variable for determining the cumulative E in this
significantly even in the driest year, 2004. Monthly E, normal- black spruce forest.
ized with potential evapotranspiration (g m−2 s−1 ), Epo , shows little
variation between years, whereas monthly P varied significantly 4. Discussion
between years (Fig. 4b). Epo was calculated with the Penman equa-
tion (Penman, 1948; Brutsaert, 2005). A relatively large variation in 4.1. Accuracy of observed fluxes
E/Epo was observed in September, which may be related to different
timing of the decline in E due to the occurrence of frost (Ueyama The observed ˛ at this site was within the range of typical ˛
et al., 2006). observed at FLUXNET sites (0.7–0.9, Wilson et al., 2002; Barr et al.,
2006; Franssen et al., 2010). In addition, ˛ during the growing sea-
3.4. Bulk surface conductance son can be underestimated, since heat flux plates were buried at
10-cm depth in this study, leading to underestimation of G at the
The relationship between bulk surface conductance (gS ) and surface. Meyers and Hollinger (2004) and Heusinkveld et al. (2004)
normalized difference vegetation index (NDVI), water vapor pres- showed that it is important to correct observed G for this under-
sure deficit (D), and soil water content (SWC) was examined in estimation, in order to close the energy balance. To examine the
Fig. 5. In this analysis, NDVI obtained from the MODIS MOD13Q1 importance of this correction, we estimated G at the surface using
product (collection 4, Huete et al., 2002) were used, since it is harmonic analysis (Horton and Wierenga, 1983; Heusinkveld et al.,
expected that the satellite data were observed more consistently 2004; Guyot et al., 2009) with one week of data without rainfall. The
over this analysis period compared to the manual observation of harmonic analysis estimates soil thermal properties from observed
LAI at the site. Only NDVI data with “good quality” were used. Data soil temperature and soil heat flux at a certain depth, and calculates
gaps were filled linearly, and monthly averages were calculated. soil heat flux at the surface. It was found that ˛ increased 10% by
Fig. 5a shows that gS is weakly correlated with NDVI. The mid- correcting G, resulting in ˛ of 0.9.
day (11:00–15:00) gS in May and September was relatively low Some of the residual in the energy balance may be explained
compared to that in June, July, and August (Fig. 5a). Generally, gS by overestimation of A through neglecting heat storage in canopy
depends on the amount of leaves, in addition to factors influenc- biomass and energy stored in carbohydrate bonds from photosyn-
ing stomatal conductance (e.g., Saigusa et al., 1998; Wilson and thesis (e.g., Meyers and Hollinger, 2004; Gu et al., 2007; Jacobs et al.,
Baldocchi, 2000). However, midday gS did not strongly depend on 2008; Michiles and Gielow, 2008; Moderow et al., 2009). Including
NDVI in June, July, and August, so we assumed that the variation this heat storage may improve energy balance closure to be nearly
of midday gS during these three months reflects the change in sto- closed. Hence, the turbulent fluxes observed at this site are likely
matal conductance under the influence of environmental changes, of reasonable accuracy for studying water and energy exchanges
and the change of LAI did not strongly affect gS . in this ecosystem. Underestimation of turbulent fluxes due to loss
 H. Iwata et al. / Agricultural and Forest Meteorology 161 (2012) 107–115 113

300
P P P P
250

Cumulative P
and E (mm)
E E E E
200
150
100
50 2003 2004 2005 2006
0
100 150 200 250 100 150 200 250 100 150 200 250 100 150 200 250
300
P P P
250
Cumulative P
and E (mm) E E E
200
150
100
50 2007 2008 2009
0
100 150 200 250 100 150 200 250 100 150 200 250
DOY

Fig. 6. Cumulative precipitation, P, and evapotranspiration, E, during the snow-free period

of low-frequency transport and/or advection may be attributable plenty of available water in the soil. The fact that the correlation
to the part of the imperfect energy balance closure; however, their between E and A is high suggests that this may be true, since a
contributions are probably less than several percentages, according shortage of available water would result in lower partitioning of A
to the discussion above. to E and more scatter around the regression line.

4.2. Constraint of frozen soil on evapotranspiration

4.3. Role of permafrost on water cycle
The relationship between the monthly average of daily A and
In this black spruce forest, E was not limited by P during the
daily E (Fig. 7) clearly shows that a year can be divided into three
first half of growing season (Fig. 4a). The evaporated water was
stages regarding the constraint of frozen soil on E. These stages
probably supplied first by snowmelt (60–160 mm m−2 water equiv-
are: the dormant stage (October through April), the recovering
alent), and subsequently by soil thawing (Rouse, 1984; Kane et al.,
stage (May and June), and the unconstrained stage (July through
1990; Yoshimoto et al., 1996; Iwahana et al., 2005). Microtopo-
September). In the dormant stage, E increases gradually with A.
graphic hollows efficiently stored the snowmelt water, which was
Since the ground is frozen and covered with snow, and decidu-
probably supplied to the ecosystem for several weeks after the com-
ous vegetation does not have leaves, E is basically dominated by
plete snowmelt. Otherwise, most snowmelt water would be lost as
sublimation from the snow surface and evaporation of snowmelt
runoff, because of the shallow thawed layer, just after the complete
water. In the recovering stage, E increases rapidly with A, sug-
snowmelt.
gesting a rapid recovery of E capacity of the ecosystem, due to an
Regardless of the lower midday gS under high VPD (Fig. 5b),
increase in LAI and deepening of the thawed soil down to about
monthly E did not significantly decrease in the months with low P;
30 cm (Fig. 1). Vogel (2004) found that the majority of roots were
rather, it stayed at a level similar to the corresponding months of
within the organic layer, which is 25–40 cm from the ground sur-
different years with high P (data not shown). This confirms that
face at the same site. This implies that there is little constraint on
water stress on transpiration was weak in this forest, and that
E from frozen soil after the recovering stage. In the unconstrained
higher radiation compensated for lower evaporation efficiency (i.e.,
stage, E decreases more gradually, with a decrease in A and LAI.
lower gS ) in the months with low P.
During this stage, E can be close to the maximum rate the ecosys-
In the latter half of the growing season, E decreased along with
tem can achieve with the leaf condition at that time, if there is
decreasing Rn and LAI, resulting in net gain of water by the ecosys-
tem (Fig. 4a). The presence of ice-rich permafrost helps to increase
6
Jan-Apr, Oct-Dec water storage by blocking deep percolation of water (Glenn and
May, Jun Woo, 1997; Woo and Young, 2006; Hinzman et al., 2006), and
5 Jul-Sep
consequently, the water resource for the ecosystem at the site
remained in a sustainable condition under the current climate and
4 y =0.47x -0.28 permafrost conditions. The excess water was preserved in the soil
λE (MJ m day )
-1

2
R =0.91 during winter, and it supported vegetation growth in the next early
p<0.0001
growing season.
-2

3
It was reported by Ohta et al. (2008) that the interannual varia-
2 tion of E/Epo was controlled by SWC in a Siberian larch forest. The
y =0.17x +0.13 difference between our results and theirs may stem from a differ-
R2=0.89
p<0.0001 ence in the amount of water in the soil at the two sites. In their site,
1 y =0.69x -3.99
2
R =0.60 maximum SWC was about 0.3, whereas the maximum SWC in this
p=0.001 black spruce forest was about 0.8 (Fig. 2). The shallow active layer
0 may have contributed to the higher water content at our study site,
-2 0 2 4 6 8 10 12 14
-2 -1 and the soil may contain enough water even in low P conditions to
A (MJ m day )
supply water to the vegetation.
Fig. 7. The relationship between monthly average of daily available energy, A, and A black spruce forest on drier soil without permafrost in Alaska,
daily latent heat flux, E in fact, experienced larger variability of E in the early half of the
114 H. Iwata et al. / Agricultural and Forest Meteorology 161 (2012) 107–115

growing season (Liu and Randerson, 2008). Yuan et al. (2010) con- References
cluded that seasonality in P, rather than the amount, has a large
Amiro, B.D., Barr, A.G., Black, T.A., Iwashita, H., Kljun, N., McCaughey, J.H., Morgen-
impact on cumulative E in the Alaskan black spruce forest with- stern, K., Murayama, S., Nesic, Z., Orchansky, A.L., Saigusa, N., 2006a. Carbon,
out permafrost, and P in the early growing season determines energy and water fluxes at mature and disturbed forest sites, Saskatchewan,
cumulative E. In our black spruce forest with ice-rich permafrost, Canada. Agric. Forest Meteorol. 136, 237–251.
Amiro, B.D., Orchansky, A.L., Barr, A.G., Black, T.A., Chambers, S.D., Chapin III, F.S.,
seasonality in P is less important (Fig. 4) due to the sufficient supply Goulden, M.L., Litvak, M., Liu, H.P., McCaughey, J.H., McMillan, A., Randerson,
of water from soil thawing and snowmelt even in months with low J.T., 2006b. The effect of post-fire stand age on the boreal forest energy balance.
P in the early growing season. These results suggest that whether Agric. Forest Meteorol. 140, 41–50.
Arain, M.A., Black, T.A., Barr, A.G., Griffis, T.J., Morgenstern, K., Nesic, Z., 2003. Year-
ice-rich permafrost exists or not is an important factor in determin-
round observations of the energy and water vapour fluxes above a boreal black
ing the magnitude of interannual variability and spatial variation spruce forest. Hydrol. Process 17, 3581–3600.
in E of boreal forests. A small interannual variation in E was also Barker, C.A., Amiro, B.D., Kwon, H., Ewers, B.E., Angstmann, J.L., 2009. Evapotran-
reported in a Canadian black spruce forest (Zha et al., 2010). In that spiration in intermediate-aged and mature fens and upland black spruce boreal
forests. Ecohydrology 2, 462–471.
case, a shallow ground water table may support high E in low P Barr, A.G., Morgenstern, K., Black, T.A., McCaughey, J.H., Nesic, Z., 2006. Surface
conditions. Thus, soil conditions such as the presence of ice-rich energy balance closure by the eddy-covariance method above three boreal for-
permafrost and the depth of the ground water table should be est stands and implications for the measurement of the CO2 flux. Agric. Forest
Meteorol. 140, 322–337.
taken into account for accurate prediction of E, in addition to plant Brutsaert, W., 2005. Hydrology: An Introduction. Cambridge University Press, Cam-
functional types. bridge, 605 p.
Chapin III, F.S., Hollingsworth, T., Murray, D.F., Viereck, L.A., Walker, M.D., 2006.
Floristic diversity and vegetation distribution in the Alaskan boreal forest. In:
Chapin, F.S.JIII, Oswood, M., Van Cleve, K., Viereck, L.A., Verbyla, D.L. (Eds.),
5. Conclusions Alaska’s Changing Boreal Forest. Oxford University Press, Oxford, pp. 81–99.
Dai, A., Fung, I.Y., del Genio, A.D., 1997. Surface observed global land precipitation
variations during 1900–88. J. Climate 10, 2943–2962.
This study examined seasonal and interannual variations of Franssen, H.J.H., Stöckli, R., Lehner, I., Rotenberg, E., Seneviratne, S.I., 2010. Energy
water and energy exchanges in a black spruce forest on ice-rich balance closure of eddy-covariance data: a multisite analysis for European
permafrost in Interior Alaska. The seasonal variation in partitioning FLUXNET stations. Agric. Forest Meteorol. 150, 1553–1567.
Glenn, M.S., Woo, M.-K., 1997. Spring and summer hydrology of a valley-
of available energy to latent heat flux was strongly affected by the
bottom wetland, Ellesmere Island, Northwest Territories, Canada. Wetlands 17,
vegetation activity and thaw depth of the soil. In the early growing 321–329.
season, transpiration of evergreen plants and growth of decidu- Gu, L., Meyers, T., Pallardy, S.G., Hanson, P.J., Yang, B., Heuer, M., Hosman, K.P., Liu, Q.,
Riggs, J.S., Sluss, D., Wullschleger, S.D., 2007. Influences of biomass heat and bio-
ous plants were limited by frozen soil, resulting in the low fraction
chemical energy storages on the land surface fluxes and radiative temperature.
of partitioning to latent heat flux. The partitioning to latent heat J. Geophys. Res. 112, D02107.
flux increased with the growth of deciduous plants and deepening Gu, S., Tang, Y., Cui, X., Kato, T., Du, M., Li, Y., Zhao, X., 2005. Energy exchange between
of the thawed soil layer. In the latter half of the growing season, the atmosphere and a meadow ecosystem on the Qinghai-Tibetan Plateau. Agric.
Forest Meteorol. 129, 175–185.
evapotranspiration was not constrained by the frozen soil. Guyot, A., Cohard, J.-M., Anquetin, S., Galle, S., Lloyd, C.R., 2009. Combined analysis
In Interior Alaska, the early growing season usually has low of energy and water balances to estimate latent heat flux of a sudanian small
precipitation. Regardless of the limited precipitation, evapotranspi- catchment. J. Hydrol. 375, 227–240.
Heijmans, M.M.P.D., Arp, W.J., Chapin III, F.S., 2004. Carbon dioxide and water vapour
ration was not controlled here by the amount of precipitation. Soil exchange from understory species in boreal forest. Agric. Forest Meteorol. 123,
thawing and storage of snowmelt water in the microtopographic 135–147.
hollows likely provided water to vegetation for transpiration. In Heusinkveld, B.G., Jacobs, A.F.G., Holtslag, A.A.M., Berkowicz, S.M., 2004. Surface
energy balance closure in an arid region: role of soil heat flux. Agric. Forest
the latter half of the growing season, evapotranspiration decreased Meteorol. 122, 21–37.
along with a decrease in net radiation and leaf area in contrast to Hinzman, L.D., Bolton, W.R., Petrone, K.C., Jones, J.B., Adams, P.C., 2006. Watershed
increased precipitation, resulting in an increase of soil water con- hydrology and chemistry in the Alaskan boreal forest. the central role of per-
mafrost. In: Chapin, F.S.JIII, Oswood, M., Van Cleve, K., Viereck, L.A., Verbyla,
tent. Excess water in the frozen soil was preserved during winter
D.L. (Eds.), Alaska’s Changing Boreal Forest. Oxford University Press, Oxford, pp.
and was utilized in the next growing season. 269–284.
Due to a sufficient water supply from precipitation and soil Horst, T.W., 1997. A simple fornula for attenuation of eddy fluxes measured with
first-order-response scalar sensors. Boundary-Layer Meteorol. 82, 219–233.
thawing, this forest did not experience strong water stress on
Horton, R., Wierenga, P.J., 1983. Estimating the soil heat flux from observations of
transpiration. The bulk surface conductance can be lower under soil temperature near the surface. Soil Sci. Soc. Am. J. 47, 14–20.
a high vapor pressure deficit, but the decrease in conductance Huete, A., Didan, K., Miura, T., Rodriguez, E.P., Gao, X., Ferreira, L.G., 2002. Overview of
was small. Higher available energy in low precipitation conditions the radiometric and biophysical performance of the MODIS vegetation indices.
Remote Sens. Environ. 83, 195–213.
compensated for the lower canopy conductance rate. This resulted Iwahana, G., Machimura, T., Kobayashi, Y., Fedorov, A.N., Konstantinov, P.Y., Fukuda,
in insignificant variations in the annual amount and the seasonal M., 2005. Influence of forest clear-cutting on the thermal and hydrological
pattern of evapotranspiration. The amount of available energy pri- regime of the active layer near Yakutsk, eastern Siberia. J. Geophys. Res. 110,
G02004.
marily determined annual evapotranspiration in this forest. Iwata, H., Harazono, Y., Ueyama, M., 2010. Influence of source/sink distributions on
The water resource for the ecosystem at this site remained in a flux-gradient relationships in the roughness sublayer over an open forest canopy
sustainable condition under the current climate due to a net water under unstable conditions. Boundary-Layer Meteorol. 136, 391–405.
Jacobs, A.F.G., Heusinkveld, B.G., Holtslag, A.A.M., 2008. Towards closing the sur-
gain in the late growing season, in combination with the presence of face energy budget of a mid-latitude grassland. Boundary-Layer Meteorol. 126,
ice-rich permafrost. This condition may persist in the future unless 125–136.
a large change occurs in either the precipitation pattern or the state Jarvis, P.G., Massheder, J.M., Hale, S.E., Moncrieff, J.B., Rayment, M., Scott, S.L., 1997.
Seasonal variation of carbon dioxide, water vapor, and energy exchanges of a
of the permafrost.
boreal black spruce forest. J. Geophys. Res. 102, 28953–28966.
Jones, P.D., Moberg, A., 2003. Hemispheric and large-scale surface air temperature
variations: an extensive revision and an update to 2001. J. Climate 16, 206–223.
Jung, M., Reichstein, M., Ciais, P., Seneviratne, S.I., Sheffield, J., Goulden, M.L., Bonan,
Acknowledgements G., Cescatti, A., Chen, J., de Jeu, R., Dolman, A.J., Eugster, W., Gerten, D., Gianelle,
D., Gobron, N., Heinke, J., Kimball, J., Law, B.E., Montagnani, L., Mu, Q., Mueller,
This study was partly supported by the Carbon Cycle Pro- B., Oleson, K., Papale, D., Richardson, A.D., Roupsard, O., Running, S., Tomelleri,
gram of IARC/NSF by the U.S. National Science Foundation and IJIS E., Viovy, N., Weber, U., Williams, C., Wood, E., Zaehle, S., Zhang, K., 2010. Recent
decline in the global land evapotranspiration trend due to limited moisture
(IARC/JAXA Information Systems). We would like to thank T. Saito, supply. Nature 467, 951–954.
A. Nojiri, and R. Okada for their support in the field work, and T. Kane, D.L., Gieck, R.E., Hinzman, L.D., 1990. Evapotranspiration from a small Alaskan
Alton and N. Bauer for editing the manuscript. arctic watershed. Nord. Hydrol. 21, 253–272.
 H. Iwata et al. / Agricultural and Forest Meteorology 161 (2012) 107–115 115

Kattsov, V.M., Walsh, J.E., 2000. Twentieth-century trends of Arctic precipita- Shulski, M., Wendler, G., 2007. The Climate of Alaska. University of Alaska Press,
tion from observational data and a climate model simulation. J. Climate 13, Fairbanks, p. 216.
1362–1370. Sugimoto, A., Yanagisawa, N., Naito, D., Fujita, N., Maximov, T.C., 2002. Importance
Liu, H., Randerson, J.T., 2008. Interannual variability of surface energy exchange of permafrost as a source of water for plants in east Siberian taiga. Ecol. Res. 17,
depends on stand age in a boreal forest fire chronosequence. J. Geophys. Res. 493–503.
113, G01006. Thom, A.S., 1972. Momentum, mass and heat exchange of vegetation. Quart. J. Roy.
Liu, H., Randerson, J.T., Lindfors, J., Chapin III, F.S., 2005. Changes in the surface energy Meteorol. Soc. 98, 124–134.
budget after fire in boreal ecosystems of interior Alaska: an annual perspective. Todd, S.K., Jewkes, H.A., 2006. Fire in Alaska: a history of organized fire suppres-
J. Geophys. Res. 110, D13101. sion and management in the last frontier. Agric. Forest Exp. Station Bull. 113.
Mahrt, L., Lee, X., Black, A., Neumann, H., Staebler, R.M., 2000. Nocturnal mixing in a University of Alaska Fairbanks, Fairbanks, AK.
forest subcanopy. Agric. Forest Meteorol. 101, 67–78. Trenberth, K.E., Jones, P.D., Ambenje, P., Bojariu, R., Easterling, D., Klein Tank, A.,
Malhi, Y., Pegoraro, E., Nobre, A.D., Pereira, M.G.P., Grace, J., Culf, A.D., Clement, R., Parker, D., Rahimzadeh, F., Renwick, J.A., Rusticucci, M., Soden, B., Zhai, P., 2007.
2002. The energy and water dynamics of a central Amazonian rain forest. J. Observations: Surface and atmospheric climate change. In: Solomon, S., Qin, D.,
Geophys. Res., 107. Manning, M., Chen, Z., Marquis, M., Averyt, K.B., Tignor, M., Miller, H.L. (Eds.), Cli-
Massman, W.J., 2000. A simple method for estimating frequency response correc- mate Change 2007: The Physical Science Basis. Contribution of Working Group
tions for eddy covariance systems. Agric. Forest Meteorol. 104, 185–198. I to the Fourth Assessment Report of the Intergovernmental Panel on Climate
Maximov, T., Ohta, T., Dolman, A.J., 2008. Water and energy exchange in east siberian Change. Cambridge University Press, Cambridge, pp. 235–336.
forest: a synthesis. Agric. Forest Meteorol. 148, 2013–2018. Ueyama, M., Harazono, Y., Kim, Y., Tanaka, N., 2009. Response of the carbon cycle
Meyers, T.P., Hollinger, S.E., 2004. An assessment of storage terms in the surface in sub-arctic black spruce forests to climate change: reduction of a carbon sink
energy balance of maize and soybean. Agric. Forest Meteorol. 125, 105–115. related to the sensitivity of heterotrophic respiration. Agric. Forest Meteorol.
Michiles, A.A.S., Gielow, R., 2008. Above-ground thermal energy storage rates, trunk 149, 582–602.
heat fluxes and surface energy balance in a central Amazonian rainforest. Agric. Ueyama, M., Harazono, Y., Ohtaki, E., Miyata, A., 2006. Controlling factors on the
Forest Meteorol. 148, 917–930. interannual CO2 budget at a subarctic black spruce forest in interior Alaska.
Moderow, U., Aubinet, M., Feigenwinter, C., Kolle, O., Lindroth, A., Mölder, M., Tellus 58B, 491–501.
Montagnani, L., Rebmann, C., Bernhofer, C., 2009. Available energy and energy Vickers, D., Mahrt, L., 1997. Quality control and flux sampling problems for tower
balance closure at four coniferous forest sites across europe. Theor. Appl. Clima- and aircraft data. J. Atmos. Oceanic Technol. 14, 512–526.
tol. 98, 397–412. Vogel, J.G., 2004. Carbon cycling in three mature black spruce (Picea mariana [Mill.]
Monteith, J.L., 1965. Evaporation and environment. Symp. Soc. Explor. Biol. 19, B.P.S) forests in interior Alaska. Ph.D. Thesis. University of Alaska Fairbanks, 178
205–234. p.
Monteith, J.L., Unsworth, M., 1990. Principles of Environmental Physics, 2nd ed. Vogel, J.G., Valentine, D.W., Ruess, R.W., 2005. Soil and root respiration in mature
Butterworth-Heinemann Ltd., Oxford, 291 p. Alaskan black spruce forests that vary in soil organic matter decomposition rates.
O’Connell, K.E.B., Gower, S.T., Norman, J.M., 2003. Comparison of net primary pro- Can. J. Forest Res. 35, 161–174.
duction and light-use dynamics of two boreal black spruce forest communities. Waring, R.H., Running, S.W., 1998. Forest Ecosystems: Anlysis at Multiple Scales,
Ecosystems 6, 236–247. 2nd ed. Academic Press, San Diego, 370 p.
Ohta, T., Hiyama, T., Tanaka, H., Kuwada, T., Maximov, T.C., Ohata, T., Fukushima, Y., Webb, E.K., Pearman, G.I., Leuning, R., 1980. Correction of flux measurements for
2001. Seasonal variation in the energy and water exchanges above and below a density effects due to heat and water vapour transfer. Quart. J. Roy. Meteorol.
larch forest in eastern Siberia. Hydrol. Process. 15, 1459–1476. Soc. 106, 85–100.
Ohta, T., Maximov, T.C., Dolman, A.J., Nakai, T., van der Molen, M.K., Kononov, A.V., Wilson, K., Goldstein, A., Falge, E., Aubinet, M., Baldocchi, D., Berbigier, P., Bernhofer,
Maximov, A.P., Hiyama, T., Iijima, Y., Moors, E.J., Tanaka, H., Tobaand, T., Yabuki, C., Ceulemans, R., Dolman, H., Field, C., Grelle, A., Ibrom, A., Law, B.E., Kowalski,
H., 2008. Interannual variation of water balance and summer evapotranspiration A., Meyers, T., Moncrieff, J., Monson, R., Oechel, W., Tenhunen, J., Valentini, R.,
in an eastern Siberian larch forest over a 7-year period (1998–2006). Agric. Forest Verma, S., 2002. Energy balance closure at FLUXNET sites. Agric. Forest Meteorol.
Meteorol. 148, 1941–1953. 113, 223–243.
Pattey, E., Desjardins, R.L., St-Amour, G., 1997. Mass and energy exchanges over a Wilson, K.B., Baldocchi, D.D., 2000. Seasonal and interannual variability of energy
black spruce forest during key periods of BOREAS 1994. J. Geophys. Res. 102, fluxes over a broadleaved temperate deciduous forest in North America. Agric.
28967–28975. Forest Meteorol. 100, 1–18.
Penman, H.L., 1948. Natural evaporation from open water, bare soil and grass. Proc. Woo, M.-K., Young, K.L., 2006. High Arctic wetlands: their occurrence, hydrological
R. Soc. Lond. A 193, 120–145. characteristics and sustainability. J. Hydrol. 320, 432–450.
Rouse, W.R., 1984. Microclimate at arctic tree line. 2. soil microclimate of tundra and Yoshimoto, M., Harazono, Y., Vourlitis, G.L., Oechel, W.C., 1996. The heat and water
forest. Water Resour. Res. 20, 67–73. budgets in the active layer of the arctic tundra at Barrow, Alaska (in Japanese
Ryu, Y., Baldocchi, D.D., Ma, S., Hehn, T., 2008. Interannual variability of evapotran- with English abstract). J. Agric. Meteorol. 52, 293–300.
spiration and energy exchange over an annual grassland in California. J. Geophys. Yuan, W., Liu, S., Liu, H., Randerson, J.T., Yu, G., Tieszen, L.L., 2010. Impacts of pre-
Res. 113, D09104. cipitation seasonality and ecosystem types on evapotranspiration in the Yukon
Saigusa, N., Oikawa, T., Liu, S., 1998. Seasonal variations of the exchange of CO2 and River Basin, Alaska. Water Resour. Res. 46, W02514.
H2 O between a grassland and the atmosphere: an experimental study. Agric. Zha, T., Barr, A.G., van der Kamp, G., Black, T.A., McCaughey, J.H., Flanagan, L.B., 2010.
Forest Meteorol. 89, 131–139. Interannual variation of evapotranspiration from forest and grassland ecosys-
Serreze, M.C., Walsh, J.E., Chapin III, F.S., Osterkamp, T., Dyurgerov, M., Romanovsky, tems in western Canada in relation to drought. Agric. Forest Meteorol. 150,
V., Oechel, W.C., Morison, J., Zhang, T., Barry, R.G., 2000. Observational evidence 1476–1484.
of recent change in the northern high-latitude environment. Climate Change 46,
159–207.