Journal of Hydrology 315 (2005) 274–294
www.elsevier.com/locate/jhydrol
Energy dynamics and modeled evapotranspiration
from a wet tropical forest in Costa Rica
H.W. Loeschera,*, H.L. Gholza,b, J.M. Jacobsc, S.F. Oberbauerd,e
a
School of Forest Resources and Conservation, University of Florida, Gainesville, FL 32611, USA
b
National Science Foundation, Division of Environmental Biology, Arlington, VA 22230, USA
c
Civil Engineering Department, University of New Hampshire, Portsmouth, NH 03801, USA
d
Department of Biological Sciences, Florida International University, Miami FL 33199, USA
e
Fairchild Tropical Garden, 11935 Old Cutler Road, Miami, FL 33156, USA
Received 4 July 2003; revised 24 March 2005; accepted 31 March 2005
Abstract
The effects of albedo, net radiation (Rn), vapor pressure deficit (VPD), and surface conductances on energy fluxes and
evapotranspiration (ET) were determined for a wet tropical forest in NE Costa Rica from 1997 to 2000. Sensible heat fluxes (H)
were estimated by the combination of eddy-covariance and the change in below-canopy heat profiles. Above-canopy latent heat
fluxes (lE) were estimated by the residuals from Rn and H, and below canopy lE fluxes. Surface reflectance (albedo) was
w12% of incident solar radiation and did not differ seasonally. Rn was significantly different among years and explained w79%
of the variation in H and lE fluxes. The effects of VPD did not explain any additional variation in heat fluxes. lE fluxes were
always greater than H fluxes when RnO40 W mK2. Understory heat fluxes were small and contributed little towards daily
energy exchange, but may be significant when Rn is small. A dimensionless coefficient (U) was used to determine the relative
importance of aerodynamic conductance (ga) and bulk canopy conductance (gb) on lE flux. During the day, U was O0.6 and
peaked at 0.85 suggesting that the forest was decoupled from physiological controls, lE fluxes are more dependent on Rn than
water availability, and ga exerts more control on lE fluxes than gb. Because of these results, both the Priestly–Taylor and the
Penman–Monteith models performed well using only Rn. Because the canopy is wet w32% of the time, there was better
precision in estimating lE fluxes using the Priestly–Taylor model (with an empirically estimated aZ1.24), when the canopy
was wet. Annual ET were 1892, 2292 and 2230 mm for 1998, 1999 and 2000, respectively. Annual ET ranged from 54 to 66%
of bulk precipitation. Using a Rutter-type model, interception losses were 17–18% of bulk precipitation. The overall amount of
energy needed for annual ET accounted for w88 to 97% of total Rn.
q 2005 Elsevier B.V. All rights reserved.
Keywords: Tropical wet forest; Evapotranspiration; Latent heat; Sensible heat; Penman–Monteith; Priestly–Taylor; Eddy covariance
* Corresponding author. Address: Department of Forest Science, College of Forestry, Oregon State University, 321 Richardson Hall,
Corvallis, OR 97331, USA. Tel.: C1 541 737 8020; fax: C1 541 737 1393.
E-mail address: [email protected] (H.W. Loescher).
0022-1694/$ - see front matter q 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.jhydrol.2005.03.040
H.W. Loescher et al. / Journal of Hydrology 315 (2005) 274–294
1. Introduction
The energy balance of tropical forests is complex
due to feedback mechanisms among radiation, cloud
formation and precipitation (Wielicki et al., 2002;
Hartmann et al., 2001; Sohn and Smith, 1992). This
complexity extends to the potential role of the tropical
energy balance in affecting tropical and global
climates and general and anomalous circulations
(Kelly and Randall, 2001; Timmermann et al., 1999;
Chen and van den Dool, 1999; Fasullo and Webster,
1999; Larson et al., 1999). Much of our understanding
of these dynamics has relied on model results, which
have shown large spatial and temporal variability in
both sensible and latent energy budgets (Kelly and
Randall, 2001; Raman et al., 1998; Hulme and Viner,
1998; Shuttleworth, 1988). In situ studies have either
scaled leaf level measurements to whole canopies
(Bigelow, 2001; Avissar, 1993; Roberts et al., 1993),
or have estimated the energy balance components
using eddy covariance over short periods (cf. 8 d,
Shuttleworth et al., 1984; 1 yr, Malhi et al., 2002; 1 yr,
da Rocha et al., 2004). Quantifying the variation of
energy balance parameters and their biophysical
controls over longer periods (e.g. years) should
allow for better predictions of runoff and improved
models of regional and global climate.
Both physical and physiological factors influence
forest energy fluxes, including incident radiation,
surface albedo, rain, interception, canopy capacity,
and aerodynamic (ga) and bulk surface (gb) conductances. Incident radiation in the tropics varies less
seasonally than that at higher latitudes, and values at
the surface are more related to cloudiness than
changes in solar zenith angle. General circulation
models tend to underestimate net radiation in the
tropics because of uncertainties in estimating surface
albedo and cloud cover (Cramer et al., 1999; Ruimy
et al., 1995). Forest surface albedos range from 0.1 to
0.2 (Monteith and Unsworth, 1990), with annual
and seasonal differences affecting the available
energy. Large variability in annual rainfall totals
have been observed in the tropics and are thought to
be influenced by El Niño-Southern Oscillation
(ENSO) and other anomalous circulations (e.g.
McKinley et al., 2004; Malhi and Wright, 2004;
Bousquet et al., 2000). This in turn affects the amounts
of water available for evapotranspiration. A general
275
observation is that 50–60% of annual rainfall is recirculated to the atmosphere through transpiration and
evaporation of intercepted water, with the other 50%
as runoff from neotropical humid forests that receive
2400–3000 mm yrK1 (da Rocha et al., 2004; Malhi
et al., 2002; Shuttleworth, 1989). The results from
these studies imply that local hydrology is strongly
affected by how energy is partitioned at the surface.
Canopy conductances for tropical forests are
typically determined by estimating bulk surface
conductance (gb) by scaling leaf-level or sap-flow
measurements to the canopy (Whitehead, 1998;
Dolman et al., 1991) or by deducing gb in relation to
meteorological parameters and measured fluxes
(Wright and Gash, 1996; Stewart, 1988). Aerodynamic conductance is generally calculated as a
function of horizontal windspeed, zero-plane displacement, and roughness length (Denmead and Bradley,
1985). Unlike agronomic communities in which
evapotranspiration is dominated by transpiration and
controlled by the plant’s canopy conductance,
evapotranspiration from tropical forests is generally
thought to be strongly dependent on aerodynamic
conductance, because of the high rainfall and the
significant proportion of the time when the canopy is
wet, reducing the importance of gb in evapotranspiration (Shuttleworth, 1989).
Our research objective was to define the surface
controls on the energy fluxes from a wet tropical
forest in Costa Rica, which included the temporal
partitioning of Rn, below-canopy energy fluxes,
conductances and overall surface energy fluxes, and
to use these controls to model annual
evapotranspiration.
2. Methods
2.1. Study site
This study was conducted as part of a long-term
study of tropical forest carbon cycling, the CARBONO project, at the La Selva Biological Station,
Puerto Viejo de Sarapiquı́, Costa Rica (10825 0 51 00 N,
84800 0 59 00 W, elevation 80–150 m.a.s.l.). La Selva is
located in northeastern Costa Rica in the Caribbean
lowlands at the base of the central volcanic chain and
was classified as tropical wet forest in the Holdridge
276
H.W. Loescher et al. / Journal of Hydrology 315 (2005) 274–294
life zone system by Hartshorn and Peralta (1988). This
forest has an average 400 trees haK1O100 mm
diameter haK1 from w100 species (Lieberman
et al., 1985), and is dominated by the mimosoid
legume canopy species, Pentaclethra macroloba
(34% of the basal area, Clark and Clark, 2000).
Mean tree height is 20–25 m, with emergents
exceeding 60 m. Canopy gaps occupy w0.01–
0.04 ha haK1 (Denslow and Hartshorn, 1994) making
the overall canopy very aerodynamically rough.
Incident mean (1993–1998) daily solar radiation was
14.9 MJ m K2 dK1, with a range from 0.4 to
31.3 MJ mK2 dK1. Mean annual temperatures were
24.6 8C (1982–1998, Organization for Tropical
Studies, OTS, unpublished weather records). Mean
annual precipitation was 4000 mm (from 1963 to
2000), with a short drier period from December to the
end of May, but with no month receiving less than
100 mm (Sanford et al., 1994). Soils range from
relatively fertile Inceptisols in riverine areas to low
pH, low phosphorus Ultisols in upland areas (Sollins
et al., 1994).
Moisture-laden northeast trade winds originating
over the Caribbean Sea dominate surface winds
(Hastenrath, 1991). During most (85%) daytime
hours, the annual mean surface wind direction is
908. The wetter season (June through November) and
drier season (mid December through May) are
controlled by the movement of the equatorial lowpressure trough, i.e. the eastern Pacific intertropical
convergence zone (ITCZ). During the drier season,
the sub-tropical Hadley cell dominates general
circulations, while the tropical cell dominates wet
season circulations (Sanford et al., 1994). There are
no data showing the exact passing of the ITCZ in
Costa Rica. Hence, seasons are defined here as either
wet season beginning on May 1 (DOY 121), or dry
season, beginning December 20 (DOY 354). These
dates are w10 d after the average date that the ITZC
passes through Barro Colorado Island, Panama (Steve
Paton, Smithsonian Tropical Research Institute,
personal communication). Other circulations may
influence wet season climate including temporales,
polar air masses that move down the North American
continent generating depressions and prolonged rain
events, chiefly occurring in November and December
(Schultz et al., 1998). Veranillos, temporary and often
irregular movement of the South Pacific anticyclone
northward, create short dry periods, typically lasting
7–10 days in September or October. Sanford et al.
(1994) and Holdridge et al. (1971) provide further site
information for La Selva, and Waylen et al. (1996a,b)
and Hastenrath (1991) provide more details on its
climatology.
Because La Selva is located at 108N latitude, there
is little diurnal change in sunlight over the course of
the year, with only a 40 min difference in day length
between solstices. For this study, sunrise and sunset
were defined as 06:00 and 18:00 h, delineating
daytime and nighttime periods.
A 42 m tower (Upright, Inc., Selma, CA) was used
to access the canopy environment and to support
meteorological instrumentation. The site was a
relatively flat ridgetop in an area of generally rolling
topography, with w20–30 m relief between stream
bottoms and ridgetops (OTS unpublished digital
elevation model). After accounting for stability
effects, a source area model (Schuepp et al., 1990)
estimated that under stable conditions, 95% of the
cumulative flux was derived from within 1.2 km of the
tower (at a mean horizontal windspeed of 3 m sK1),
indicating the source area is within the boundaries of
this forest for both day- and night-time measured
fluxes. By placing the tower away from treefall gaps
and on a flat area, effects due to forest edges, belowcanopy advection either to or from the site, and any
major directional differences due to forest composition and structure were minimized.
2.2. Experimental data
All measurements were collected from September
1997 to December 2000. Instrumentation for measuring air temperature, relative humidity, bulk precipitation, and net radiation were mounted at the top of the
tower (Upright, Inc., Selma, CA). Prior to March 1,
1999, air temperature (Ta) was measured with a
CS500 probe (Campbell Scientific, Inc., Logan, UT)
installed within a radiation shield, and linearly backcorrected to fit the response of the aspirated
temperature sensor (R2Z0.98). After March 1, 1999,
Ta was measured with platinum resistance thermometers (100 U PRT, Omega Engineering, Stamford, CT) mounted in an aspirated shield. Rainfall
was measured with a tipping bucket rain gauge (model
TE525, metric, Texas Electronics, Dallas, TX).
H.W. Loescher et al. / Journal of Hydrology 315 (2005) 274–294
From August 1, 1997 to March 1, 1999 Rn was
measured with a Fritschen-type sensor (model Q*7.1,
Radiation Energy Balance Systems, Seattle, WA). Net
radiation, Rn, was measured from March 1, 1999 to
December 2000 with a closed-cell thermopile-style
sensor (NR-lite, Kipp and Zonen, Delft, the Netherlands). All data collected with the Q*7.1 were linearly
back-corrected to fit the response of the NR-lite (R2Z
0.97) and corrected for advected sensible heat across
the sensor faces. Another radiometer was also used to
measure upward and downward welling short-and
longwave radiation (model CNR1, Kipp and Zonen,
Delft, the Netherlands) to estimate albedo during
February–April, July, and September 2000.
Specific humidity was sampled from six inlets at
0.50, 7.30, 11.95, 16.55, 21.20, and 27.60 m on the
tower. Solenoids switched the flow (w3.0 lpm) from
each inlet through an infra-red gas analyzer (model
Li-6262, Li-Cor, Lincoln, NE) for 5 min during each
30 min averaging period; data were only used for the
final 4 min in the 5 min interval. A second Li-6262
sampled specific humidity continuously at 42 m and
30 min averages were used in the vertical integrations
of Zx (Eqs. (2a) and (2b)). The profile temperature, qP,
was measured with PRTs housed in radiation shields
and co-located with each inlet and when sampling
occurred, the airflow acted to aspirate the PRTs. The
temperature probe mounted at the top of the tower was
used for qP at 42 m.
Soil heat flux plates (model HFT-3, Radiation
Energy Balance Systems) were installed at a depth of
5 cm, in each of three 1!1 m plots O20 m distance
apart near the base of the tower. Atmospheric pressure
was measured at w3 m (PB105, Vaisala, Helsinki,
Finland). All meteorological data were collected at 5 s
intervals and compiled as 30 min averages with a
datalogger (CR10X, Campbell Scientific, Inc., Logan,
UT). Instruments were cleaned, leveled as necessary,
and recalibrated according to manufacturers’
instructions.
The eddy covariance system was comprised of a
sonic anemometer (K-probe, Applied Technologies,
Inc., Boulder, CO) and a laptop computer. The sonic
anemometer measured the wind velocities in three
dimensions at 10 Hz, where w is vertical windspeed,
and u and v are the two horizontal windspeed
components, as well as the sonic potential temperature, qs. Covariance fluxes were calculated in real time
277
at 10 Hz using a software program (McMillen, 1988),
which also collected raw eddy covariance data files.
To determine the turbulent fluctuations, instantaneous
measurements of qs and w were subtracted from a
600 s recursive mean. Contributions by low frequency
(i.e. longer, 30 min averaging operators) only caused
additional variability in 30 min averages of sensible
heat (H) flux (Loescher et al., 2003). However, these
data were observed only during early morning hours
and were filtered out because of non-stationarity (cf.
Foken and Wichura, 1996). Data were also removed
when rain occurred, the 30 min data collection periods
were incomplete, or when signals from either the
sonic anemometer were out-of-range (Hollinger et al.,
1995; Anthoni et al., 1999). Protocols for accuracy,
precision, and quality control and assurance were used
as defined by the AmeriFlux Science Plan (http://
cdiac.esd.ornl.gov/programs/ameriflux/scif.htm).
Leaf area estimates were made monthly by a plant
canopy analyzer (Li-2000, Li-Cor, Inc., Lincoln, NE)
and constrained by litterfall estimates, hemispherical
photography and destructive sampling after harvest.
2.3. Energy flux estimates
The energy equation uses a control volume
approach where the top of the control volume is the
tower top (42 m) and the bottom is the ground surface.
Here, we account for energy fluxes in and out of the
control volume and the change in energy stored in the
control volume. The storage fluxes were estimated by
the changes of heat through the forest profile in the air
column and through the leaf area (Aubinet et al.,
2001; Moore and Fisch, 1986). Ecosystem-level
energy balance was estimated by
Rn K lE K H K G
ð Zx
ð Zy
vqp
vqp
dz C Swt Cpw Larea
dz C lEBC
Z C p ra
Z0 vt
Z0 vt
(1)
assuming horizontal homogeneity in source area, Rn is
net radiation, lE is the latent heat flux, H is the
sensible heat flux, and G is the soil heat flux (all
units are W mK2). The right-hand terms are the
non-turbulent heat fluxes (W mK2, i.e. change in
below-canopy heat profiles), where the first term is the
change in sensible heat in the air column (HBC),
278
H.W. Loescher et al. / Journal of Hydrology 315 (2005) 274–294
the second term is the change in sensible heat in the
leaf biomass (Hleaf), the third term is the change in
latent heat stored in the air column (lEBC), such that
Cp is specific heat capacity of air (J kgK1 KK1), ra is
the density of air (kg mK3), Zx is height of the air
column determined as the tower top (42 m), Zy is the
maximum canopy height (27 m), Z0 is the ground
surface (0 m), qp is the potential temperature from
profile measurements (8C), Swt is the specific weight
of leaf water (kg mK3 leaf area), Cpw is the heat
capacity of water (J kgK1 H2O KK1), and Larea is the
one-sided leaf area based on LAI from polyculture
plantations at La Selva (i.e. 4.5 m2 mK2, S. Bigelow
and J. Ewel, pers. comm.). Changes of sensible heat in
the leaf biomass were included in Eq. (1) because
leaves have small thermal inertia but significant
amounts of water. It was assumed that the temperature
profiles were similar throughout the flux source area.
Net ecosystem exchange of sensible heat was
estimated by the combination of turbulent and storage
fluxes (Eq. (2a)), and the turbulent latent energy
exchange was calculated as a residual flux Eq. (2b),
such that
ð Zx
vqp
HNEE Z Cp ra w 0 qs0 C Cp r
dz
Z0 vt
ð Zy
vqp
C Swt Cpw Larea
dz
(2a)
Z0 vt
ð Zx
lE Z Rn K G K HNEE K f rw l
Z0
vq
dz
vt
(2b)
where w 0 , and qs0 are the deviations of instantaneous
values from a running mean of vertical windspeed
(m sK1) and qs as a function of the speed of sound and
changes in air density (excluding water vapor, i.e.
potential temperature), w 0 q 0 is the turbulent exchange
of sensible heat flux as estimated by eddy covariance
method, f is molar volume of air (mol mK3), rw is the
molar weight of water (18.015 g molK1), and q is the
specific humidity (mol H2O molK1 air). Hence, lE is
calculated by residual, and in practice, forces energy
balance closure (e.g. Twine et al., 2000; van der Tol
et al., 2003). Errors associated with this technique
are discussed later.
Storage flux through stem biomass was not
estimated here. Moore and Fisch (1986) modeled
heat flux in stem biomass from an Amazonian forest.
Using their relationships based on air temperature
profiles, heat flux in stem biomass from La Selva was
likely w7 W mK2 (30 min average, cf. Fig. 3). Moore
and Fisch (1986) also report that their model error is
G7 W m K2 and estimates should not exceed
15 W mK2 averaged over a 30 min period. Moreover,
their model was developed in a more structurally
complex forest that had twice the amount of the
above-ground biomass with a greater proportion of
radiation penetrating the canopy than the La Selva
forest. Hence, the modeled estimate of 7 W mK2
(similarly, averaged over a 30 min period) is likely an
overestimate and considered inconsequential in light
of other uncertainties.
2.4. Modeled conductance and evapotranspiration
Evapotranspiration estimates were partitioned into
whole forest transpiration and the evaporation of
intercepted precipitation. lE was modeled using the
Penman–Monteith equation (Monteith and Unsworth,
1990)
lEpm Z
DRn C ra Cp ½es ðTa Þ K ea ðTa Þga
h
i
D C g 1 C ggba f
(3)
where lEpm is latent energy flux (W mK2), D is the is
the rate of increase in saturated water vapor pressure
with temperature (kPa KK1), es is the saturated water
vapor pressure at Ta, ea is the actual ambient water
vapor pressure (kPa), ga is the aerodynamic conductance (mol mK2 sK1), l is the latent heat of
vaporization (J kgK1), g is the psychrometric constant
at 25 8C (0.0665 kPa KK1), gb is the bulk canopy
conductance (mol mK2 sK1). To change units of
energy to depth, lEpm was multiplied by a conversion
factor that included molar volume (f, mol mK3) and
weight (kg molK1). The use of the notation lE
denotes energy flux (W mK2), and ET, water depth
(mm per unit time), hence evapotranspiration depth
estimated by Eq. (3) is noted in the text as ETpm.
Based on Monin–Obukov similarity theory, a
negative exponential wind profile was assumed
H.W. Loescher et al. / Journal of Hydrology 315 (2005) 274–294
279
and aerodynamic conductance estimated by
evapotranspiration (Jarvis and McNaughton, 1986):
k2 u
ga Z h i2
h if
zxKd
zm
ln zzxKd
C
ln
K
J
ln
C
J
m
h
zm
z42
m
UZD
(4)
where k is von Karman’s constant (0.40), d is the zeroplane displacement (m), zm is the aerodynamic
roughness length (m), and Jm and Jh are the diabatic
correction factors (dimensionless) for momentum and
sensible heat, respectively (Yasuda, 1988; Arya,
1988). Zero-plane displacement and aerodynamic
roughness changed with stability and were empirically estimated for this study period (Loescher et al.,
2003). Diabatic correction factors are a function of
stability (i.e. Monin–Obukov length) and are
explained in Denmead and Bradley (1985), Kaimal
and Finnigan (1994), and Panofsky and Dutton
(1984).
Bulk canopy conductance, gb was estimated by
gb Z
ga
lE
C
K 1 ra D42
DH
Cp lE
(5)
where D42 is the specific humidity deficit at the
measurement height (kg kgK1).
Annual ET was modeled because gaps in long-term
datasets were inevitable due to sensor calibrations,
and sensor and power failures. Consequentially,
empirical relationships were developed for ga with
horizontal windspeed, and for gb, relationships with
Rn and the vapor pressure deficit (VPD) for use in Eq.
(3). gb was normalized by the maximum value of gb,
i.e. to unity, and the upper limit to Rn determined (cf.
Jarvis, 1976; Livingston and Black, 1987). This limit
function was used to estimate a theoretical maximum,
gmax, by increasing gb as though Rn was not limiting.
Then gmax, in turn, was related to VPD, assuming that
maximum conductance would take place with 0 VPD
and high Rn. This technique has been tested and
rigorously applied by Martin et al. (1997), but differs
slightly from Wright and Gash (1996), which
involved a multi-variate optimization and a hypothetical gmax that was never reached in practice.
A dimensionless decoupling coefficient, U, was
used to determine the relative effects of ga and gb on
g
D
g
C1
(6)
C 1 C ggba
A second method of modeling lE, the Priestly–
Taylor equation, was used to compare with the
Penman–Monteith results. The Priestly–Taylor
equation (Eq. (7)) simplifies the transfer process that
is explicit in Eq. (3), and in doing so, is thought to be
appropriate for large-scale, well-watered vegetative
canopies, like those typically found in the wet tropics
(Priestly and Taylor, 1972), and is defined as
D
lEPT Z aRn
(7)
D Cg
where a is a coefficient estimated by fitting the model
results to measures of lE from Eq. (2b). Monteith
(1981) estimated an average a on a theoretical basis as
1.26, but values observed over rough canopies have
varied greatly (Jones, 1992).
To determine annual ET, evaporation of intercepted water by the canopy was modeled using a
Rutter-type model (Calder et al., 1986). Canopy water
storage increased exponentially with precipitation to a
maximum capacity (Cmax). The rate by which the
canopy filled with water was estimated by a unitless
fill constant (a value of 0.28 was used from a
broadleaf plantation forest at La Selva, Bigelow,
2001). An empirical estimate of 1.53 mm was used for
Cmax (Loescher et al., 2002). The stemflow component of interception was ignored because it was
assumed to be a small volumetric flux, i.e. !2% of
rain (Schroth et al., 1999; Neal et al., 1993). The
canopy conditions were modeled as either completely
wet or completely dry for any 30 min period (Calder
et al., 1986). If the amount of ‘free’ canopy water was
less than the amount of modeled ET, then the
remainder of canopy free water was modeled as
transpiration (gbO0) not evaporation (i.e. setting ga/
gb in the denominator of Eq. (3) to zero, cf. Bigelow,
2001; Ubarana, 1996).
To characterize the stability within at the tower
top, Zx, a stability parameter (2) was used, such that
2 ZK
Zx K d
;
L
LZ
u3 ra Cp Ta
kgH
(8)
280
H.W. Loescher et al. / Journal of Hydrology 315 (2005) 274–294
d is the zero plane displacement (m) which is the
height at which the wind profile extrapolates to
w0 m sK1 within the canopy, L is the Monin–Obukov
length (m), ratio of convective to mechanical
turbulent production, with u* defined as friction
velocity (m sK1) and determined by the covariance
of w 0 u 0 as measured by the sonic anemometer, and g
the acceleration due to gravity (m sK2).
3. Results
3.1. Energy fluxes
Here, the measured atmospheric stability, the
energy balance components, and below canopy
energy exchanges are analyzed. Subsequently, the
modeled net ecosystem energy fluxes are examined.
The stability index, 2 differed only with incident
radiation, and did not differ with year or season. On a
diurnal basis, 2 was neutral (w0 m) during the night,
and decreased during the day time until 14:00
when the boundary layer became weakly unstable
(2wK125 m, Fig. 1). After 14:00 h, 2 sharply
increased, and the boundary-layer became weakly
stable (w100 m) at 16:00 h, but then returned to
neutral conditions by nightfall.
Total daily Rn ranged from 1.47 to 27.54 MJ dK1
during the measurement period, and differed among
years (using a general linear model with an alphaZ
0.05, p!0.0001, Fig. 2) with mean daily totals for
1998–2000 of 13.31G0.028, 17.48G0.050, and
15.33G0.040 MJ dK1 (meanG1SE), respectively.
Rn also varied significantly with season (p!0.0001).
Mid-day albedo did not change seasonally, and ranged
from 0.118 to 0.135.
Diurnal temperature and water vapor profiles in the
canopy (Fig. 3) followed trends similar to those of
other forests (Shaw et al., 1988). Heating of the
canopy air column during the day increased with
height, i.e. there was a positive temperature gradient.
However, negative temperature (counter) gradients
were often observed between 21 and 27 m, where the
leaf area was concentrated. Cooling during the night
often produced neutral or slightly negative gradients,
often with warmer temperatures at ground levels.
150
ζ
100
ζ (dimensionless)
50
0
–50
–100
–150
A
B
C
D
A
–200
0
200
400
600
800 1000 1200 1400 1600 1800 2000 2200 2400
Time (hour)
Fig. 1. Diurnal patterns of the stability parameter over an old growth wet tropical forest. Data are averages using all data from 1998 to 2000.
Intervals A, B, C, and D indicate neutral, unstable, weakly unstable, and weakly stable boundary conditions, respectively. Error bars are G1SE.
H.W. Loescher et al. / Journal of Hydrology 315 (2005) 274–294
281
7000
1998, mean 13.31 MJd–1
1999, mean 17.48 MJd–1
2000, mean 15.33 MJd–1
Net radiation, cumulative MJ
6000
5000
4000
3000
2000
1000
0
0
50
100
150
200
250
300
350
DOY
Fig. 2. Cumulative net radiation for 1998–2000 over an old-growth forest in La Selva, Costa Rica. Daily means were derived using first-order
regression. Each year was significantly different at the aZ0.05 level, p!0.001, and R2O0.99.
Negative or neutral water vapor gradients were
observed all times, with negative gradients present
during non-rain days between 11 and 21 m height.
Soil heat fluxes followed very similar diurnal
patterns throughout the year, and ranged G16 W m–2
at any point in time, with negative flux into the system
during the daytime (Fig. 4A). On a day-to-day basis,
the contribution of G varied little, and though the soil
thermal properties for these soils were not known, the
change in annual G was expected to be negligible. At
night, lEBC was w3–5 W mK2, larger lEBC occurred
in the early morning hours, presumably from
convective winds mixing the below-canopy airspace
and evaporating free water. lEBC flux continued to be
positive throughout the afternoon, but was more
variable, with mean daytime values ranging from w6
to 7 W mK2. This flux remained positive throughout
the day across all years. Mean nighttime storage of
sensible heat (HBCCHleaf) ranged from ca. K1 to
K14 W mK2, and became negative in the early hours
as the air space increased in temperature. The
maximum average HBCCHleaf was w18 W mK2,
which occurred at w08:00 h when the convective
boundary layer was developing. HBCCH leaf
decreased and became negative at w14:00 h, which
coincided with a late afternoon weakly stable/unstable
boundary layer (Fig. 1). The below-canopy environment continued to lose sensible heat until w19:00 h,
when neutral canopy conditions prevailed. The
maximum daytime soil flux lagged peak fluxes of
lEBC and HBCCHleaf by w7.5 h. The total storage
flux (Fig. 4B) became negative (out of the ecosystem)
at w16:00 corresponding to the decrease in HBCC
Hleaf. During nighttime neutral conditions, the flux
was w8–10 W mK2. The daytime minima were ca.
K5 W mK2 and the peak efflux was 30 W mK2,
which occurred during weakly stable conditions
(Fig. 4B).
The average 30-min lE was greater than HNEE for
all daytime hours and across seasons and years (i.e.
H/lEZb!1.0, Fig. 5). The largest relative contribution of HNEE occurred between 09:00 and 12:00 h
(Fig. 5), coinciding closely with the observed diurnal
pattern of Ta. A statistical (linear) model that included
second-order effects of year, season, VPD and Rn,
explained 79% of the total variation in HNEEClE.
282
H.W. Loescher et al. / Journal of Hydrology 315 (2005) 274–294
30
0600
0530
1330
1800
25
Height (m)
20
15
COOLING
HEATING
10
5
0
DAY
NIGHT
22 23 24 25 26 27 28 29 30 22 23 24 25 26 27 28 29 30
θp (C)
30
0600
0530
1430
1800
25
Height (m)
20
15
LOSS
GAIN
10
5
0
DAY
38
39
40
41
42
NIGHT
43
44
45 38
39
40
41
42
43
44
45
water vapor (mmol mol–1)
Fig. 3. Representative diurnal changes in below-canopy temperature and water vapor profiles from an old-growth forest, La Selva, Costa Rica.
Data are median values for all of 1999. Standard errors were !0.05 8C and !0.1 for q and q, respectively.
Because Rn and VPD are auto-correlated and VPD did
not explain any additional variation, it was removed
from the linear model. Rn alone accounted for 69 and
68% of the variation in HNEE and lE, respectively.
3.2. Modeled conductance and evapotranspiration
ga was linearly correlated with horizontal windspeed (Fig. 6A) and followed the diurnal patterns of
convective activity throughout the day (Fig. 1). These
diurnal patterns in ga agree with observations over
Amazonian forests (e.g. Grace et al., 1996; Wright
and Gash, 1996). Values of jH and jM were small,
contributed little, and all other parameters were
relatively constant. The upper boundary of
normalized gb was a square hyperbolic function to
Rn (Fig. 6B). gmax was expected to be negatively
related to VPD because of the negative physiological
H.W. Loescher et al. / Journal of Hydrology 315 (2005) 274–294
283
20
A
λEBC
G
HBC + Hleaf
15
10
5
Storage energy flux (W m–2)
0
–5
–10
–15
40
700
B
λEBC + HBC + Hleaf + G
Rn
30
600
500
400
300
10
200
Rn (W m–2)
20
0
100
–10
0
–20
–100
0
200
400
600
800 1000 1200 1400 1600 1800 2000 2200 2400
Time (hour)
Fig. 4. Diurnal patterns of below-canopy storage energy fluxes from an old-growth wet tropical forest in Costa Rica where (A) includes each of
the components of the energy flux, and (B) the sum of the components contrasted with Rn. Data are mean values from 1998 to 2000 with error
bars G1SE.
response of canopies to VPD (Martin et al., 1997).
This relationship was not found, so an upper limit for
gmax as a function of VPD could not be determined
(Fig. 6C). gb was modeled only as a function of Rn, in
concurrence with statistical results from a first-order
regression. The minimum ga was 1.0 mol mK2 sK1,
which increased to O2.0 mol mK2 sK1 during midday (10:00–15:00 h, Fig. 7A). Estimated values of gb
followed a similar diurnal pattern as Rn, and at dawn,
gb was w1 mol mK2 sK1 until 08:00 h, then steadily
increased till noon, decreased during the later after-
noon, and approached 0.1 mol mK2 sK1 at dusk
(Fig. 7B). Modeled values of ga and gb behaved
similarly to those derived using Eqs. (4) and (5)
(Fig. 7). U ranged from w0.6 during the night to
w0.85 by 08:30 h. For the majority of daytime hours
(06:00–16:00), U was O0.75 (Fig. 7C).
Both the Penman–Monteith and Priestly–Taylor
equations performed well in estimating lE (Fig. 8A).
The slopes of all the regression lines were w1.0. The
Penman–Monteith equation explained 95% of the
observed variation in lE, with the variation being
284
H.W. Loescher et al. / Journal of Hydrology 315 (2005) 274–294
0.45
0.40
0.35
2000
1999
1998
0.30
Bowen ratio (H/ E)
0.25
0.20
0.15
0.10
0.05
0.00
– 0.05
– 0.10
– 0.15
– 0.20
– 0.25
600
800
1000
1200
Time (hour)
1400
1600
Fig. 5. Daytime Bowen ratios for each year. Estimates of lE were calculated by Eq. (2b). Data are median values (G1SE) with RnO40 W mK2.
equally distributed and increasing with increasing
energy (Fig. 8A). There was no significant difference
between using the Penman–Monteith equation for wet
and dry canopy conditions. In contrast, the simpler
Priestly–Taylor relationship accounted for R98% of
the observed variation (Fig. 8B). While there was also
no significant difference in the use of the Priestly–
Taylor equation to estimate lE for wet and dry canopy
conditions, it did seem to overestimate lE when Rn
was !500 W mK2 under dry canopy conditions
(Fig. 8C). Annual ETPT ranged from 1892 mm in
1998 to 2294 mm in 1999, and from 54 to 66% of
bulk precipitation (Table 1). Mean daily ETPT
rates were also lower in 1998 and greatest in 1999
(5.2–6.3 mm dK1, respectively). Results using the
Penman–Monteith model were not significantly
different than the Priestly–Taylor model. Interception
loss was greatest in 2000, with an annual total of
708 mm, accounting for 18% of bulk precipitation.
3.3. Uncertainty in modeled estimates
All 3 years of measured data were used to
estimate the conductances, which in turn, were
used to model ET. A traditional approach to assess
uncertainty in modeled estimates would be to
divide the data, use one set to derive parameters
and test the results against the other set. For our
data, this technique would only assess the
uncertainty at w1 yr. Alternatively, a statistical
bootstrap technique (S-plus, v. 6.1, Lucent Technologies, Inc., Seattle, WA) resampled observations
from the total dataset, mimicked the process of
sampling each 30 min period observation (Shao
and Tu, 1995; Bradley and Tibshirani, 1993). To
assess uncertainty in our modeled conductance,
estimates were compared to measured values at
different timescales, resampling at 3, 7, 30, 165 d,
1 yr intervals (cf. Spencer et al., 2004; Magnussen
and Burgess, 1997). Ten replications were used
for the 3, 7 and 30 d intervals, and two
replications for 165 d and 1 yr intervals. Differences between modeled and measured mean ga
were !6% (of measured ga) with SE,
!0.08 mmol mK2 sK1 for 30, 165 d, and 1 yr
intervals. For intervals %7 d, modeled ga overestimated the measured ga by w50% during the
night, and underestimated measured ga by 25%
H.W. Loescher et al. / Journal of Hydrology 315 (2005) 274–294
5
285
A
ga = 1.37 + 0.66*u
ga (mol m–2 s–1)
4
R2 = 0.91
3
2
1
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
horizontal windspeed (m s–1)
1.2
normalized bulk canopy conductance
B
1.0
gmax = (1.11*Rn)/(374.21+Rn)+0.17
R2 = 0.98
0.8
0.6
0.4
0.2
0.0
200
400
600
800
1000
Net radiation, W m–2
1.2
residuals from gbmax - Rn
C
1.0
0.8
0.6
0.4
0.2
0.00
0.25
0.50
0.75
1.00
1.25
1.50
1.75
2.00
VPD, KPa
Fig. 6. Empirical relationships of both aerodynamic and bulk conductance used to model ETpm. Independent variables were averaged at different
intervals, i.e. 0.25 m sK1, 50 W mK2, and 0.025 kPa, for A, B, and C, respectively. All data were median values with G1SE.
H.W. Loescher et al. / Journal of Hydrology 315 (2005) 274–294
decoupling coefficient, Ω
bulk canopy conductance (mol m–2 s–1) aerody namic conductance (mol m–2 s–1)
286
4.0
3.5
A
measured
modelled
3.0
2.5
2.0
1.5
1.0
0.5
0.0
4.0
3.5
B
3.0
2.5
2.0
1.5
1.0
0.5
0.0
0.90
0.85
C
0.80
0.75
0.70
0.65
0.60
0.55
600
800
1000
1200
1400
1600
1800
Time (hour)
Fig. 7. The diurnal relationships for (A) aerodynamic conductance, (B) bulk canopy conductance, and (C) the decoupling coefficient. All data
are median values from 1997 to 2000. Error bars in graphs A and B are G1SE. SE for graph C are typically !0.006.
during the day (Fig. 9A). Measured ga also showed
higher daytime SE among 3- and 7 d intervals,
ranging 0.05–0.25 mmol mK2 sK1 (Fig. 9B). The
observed variability in modeled ga was consistent
among all resampling intervals with SE w0.01 and
!0.08 mmol mK2 sK1 during night and daytime,
respectively (Fig. 9C). Similar trends were observed
for gb. These results suggest that this modeling
H.W. Loescher et al. / Journal of Hydrology 315 (2005) 274–294
λE Penman-Monteith (W m–2)
1400
287
A
y = 0.99(x) – 4.22
R2 = 0.95
1200
1000
800
600
400
200
0
–200
1400
B
λE Priestly-Taylor(W m–2)
1200
y = 1.00(x) +1.00
R2 = 0.98
1000
800
600
400
200
0
–200
1400
λE Priestly-Taylor (W m–2)
1200
C y = 1.00(x) + 1.00
R2 = 0.99
1000
800
600
400
200
0
–200
–200
0
200
400
600
800
1000
1200
1400
E (W m–2)
Fig. 8. Relationship between empirical and modeled estimates of the latent energy flux, where in (A) is modeled using a Penman–Monteith
equation, and (B) using Priestly–Talyor equation with a dry canopy, and (C) using a Priestly–Taylor equation with a wet canopy. All graphs
used data from 1998 to 2000. An a of 1.24 was found for graphs B and C. All slopes were significant, p!0.0001. y-intercepts had p-values of
0.003, 0.04 and 0.04 for A, B and C, respectively.
approach cannot fully capture the short-term
variability observed in the measured data (i.e.
!7 d), but is robust for monthly, seasonal and
annual estimates of conductance and lE. Other
systematic and random errors in determining the
below canopy fluxes are discussed in Moore and
Fisch (1986) and for the eddy-covariance approach
by Kruijt et al. (2004).
ET fluxes calculated using the Priestly–Taylor equation unless otherwise noted. Units for total Rn are J yrK1, and ETpt/Rn is the fraction of the total annual energy needed to evaporate ETpt of
Rn. Evaporation of intercepted water is noted as Ei. Units for rain, ETpt, Ei are mm yrK1, and daily ETpt are mean mm G1SE. nd. denotes no data.
a
Luvall (1984).
b
Bigelow (2001).
c
Calder et al. (1986).
d
Malhi et al. (2002).
e
da Rocha et al. (2004).
f
Shuttleworth et al. (1984), Shuttleworth (1988).
0.30
0.26
0.32
0.35
0.06–0.25
0.41
nd.
nd.
0.30
0.55
0.66
0.54
0.47
nd.
0.52
0.66
0.60
0.48
564
587
708
760
74–375
595
nd.
nd.
nd.
0.95
0.88
0.97
1.25
0.79–0.90
0.96
0.65
0.86
0.92
1892
2294
2230
2172
1318–1509
1481
1124
1300
nd.
3495
3575
4127
4620
3156
2892
2089
2200
nd.
Old-growth
Old-growth
Old-growth
Old-growth
Three plantations
Secondary
Old-growth
Old-growth
Old-growth
This study
This study
This study
La Selva, Costa Ricaa
La Selva, Costa Ricab
Janlappa, Javac
Cuieiras, Brazild
Tapajos, Brazile
Ducke, Brazilf
1998
1999
2000
Aug 82–Mar 83
Dec 94–Nov 95
Aug 80–Jul 81
Sept 95–Aug 96
July 00–July 01
8 d in Sept 83
Annual
Ei/ETpt
Annual
ETpt/rain
Annual
Ei
Annual
ETpt/Rn
Annual
ETpt
Annual
rain
Period
Forest
Site
Table 1
Comparison of evaporative fluxes from the neo-tropics
0.17
0.17
0.18
0.17
nd.
0.21
nd.
nd.
nd.
H.W. Loescher et al. / Journal of Hydrology 315 (2005) 274–294
Annual
Ei/rain
288
4. Discussion
4.1. Ecosystem energy dynamics
Daytime profiles of temperature and water vapor in
the upper canopy were often counter to the abovecanopy gradients as a result of winds that did not fully
penetrate the canopy. The top of the canopy, between
21 and 27 m, acted as a physical barrier to the removal
of transpirated moisture from below. As a result, the
negative lEBC at night was due to adiabatic cooling
and consequent condensation. Storage fluxes can
contribute substantially to the overall ecosystem
energy flux when Rn is small or during the night. On
a diurnal basis, this was a very small component
(w2%). Malhi et al. (2002) did not find this to be the
case from a seasonally humid forest in Amazonia,
where storage fluxes calculated by the residual from
the energy balance accounted for as much as 30% of
overall flux on an hourly basis. This suggests a strong
contrast between tropical wet (O3200 mm yrK1) and
seasonally humid forests (2200–3100 mm yrK1).
Bowen ratios (b) were consistently !1 indicated that
water was not limiting lE at any time during the year (cf.
de Rocha et al., 2004). Soils in the wet tropics generally
do not exert hydraulic limitations on lE (De Bruin,
1983). This is likely also the case at La Selva, where soils
have high water-holding capacity and high hydraulic
conductivity (Wietz et al., 1997; Sollins et al., 1994).
The neglibile response of lE and gb, to VPD, further
suggests that water was not limiting in this forest.
Pentaclethra macroloba, the dominant tree species
(42% of the basal area) is known to close its stomata and
leaves in the late afternoon (w1530 h). However,
changes in either HNEE or lE due to associated changes
in P. macroloba physiology were not detected. This
result suggests that either single species do not control
energy partitioning, their controls cannot be detected at
the ecosystem-level in heterogenous tropical wet forests
with the eddy-covariance technique, or that the time of
day when P. macroloba closes its leaves and stomata is
inconsequential to the above-canopy energy fluxes.
4.2. Conductances and other limits to annual
energy fluxes
The general diurnal patterns found for ga and gb are
similar to those reported for other tropical forests
(measured-modelled)/measured
H.W. Loescher et al. / Journal of Hydrology 315 (2005) 274–294
0.6
289
A
3-d resampling ga
30-d resampling ga
0.4
0.2
0.0
–0.2
–0.4
–0.6
0.30
B
0.25
3-d resampling of measured ga
30-d resampling of measured ga
standard error from resampling, mmol m–2 s–1
0.20
0.15
0.10
0.05
0.00
0.08
C
0.07
3-d resampling of modelled ga
30-d resampling of modelled ga
0.06
0.05
0.04
0.03
0.02
0.01
0.00
0
200
400
600
800 1000 1200 1400 1600 1800 2000 2200 2400
Time (hour)
Fig. 9. Statistics from resampling the measured- and modeled aerodynamic conductance (ga) datasets at 3 and 30 d intervals with 10 replications
across time of day. (A) represents the difference between the resampled measured- and modeled ga as a fraction of the measured ga, and (B) and
(C) represent the standard error after resampling for measured and modeled ga data, respectively. Note different scales for (B) and (C).
(Bigelow, 2001; Wright and Gash, 1996; Shuttleworth
et al., 1984). We found however, higher late afternoon
gb rates for the old-growth forest than those found
for three monocultural plantations at La Selva
(w0.5–0.8 mol mK2 sK1, Bigelow, 2001). Our higher
observed afternoon gb was likely an integrated
response of all species, suggesting that the gas
exchange measurement acquired with cuvettes from
290
H.W. Loescher et al. / Journal of Hydrology 315 (2005) 274–294
these plantations do not fully capture the response
from old-growth forests. Moreover, our gb estimates
were also higher than Wright and Gash (1996) values
from a tropical humid forest in Brazil. The lower gb
estimates found by Wright and Gash (1996) were due
to VPD constraints on gb that were not found in this
study. For the most part, VPDs at La Selva were
relatively low, !1.5 kPa. VPDs may, however, affect
energy flux at the leaf-level in the upper canopy, but
were not detected by the methods used here,
suggesting that the lack of ecosystem-level response
to the observed VPDs may be due to increased
transpiration in the lower canopy, thereby offsetting
any leaf-level effects in the upper canopy. We found
significant differences in observed gb among years and
seasons that followed the same trends as Rn. Between
year differences in Rn explained much of the
interannual variability observed in measured and
modeled lE from La Selva. Increased rainfall in
2000 increased the absolute amount of interception,
but not the fraction of rainfall intercepted.
The direct effects of gb on lE could not be
determined because these variables were not be
independently measured. However, increasing values
of U (from 0.6 to w0.85) during the early morning
indicate that hours before 09:00 h were only times
when strong physiological control on lE occurs,
likely due to the opening and closing of stomata and
leaves. Values of UO0.75 suggest that mid-day lE is
controlled more by Rn and ga, than gb. ga explained
44% of the variation in lE during times when both
conductances were used to calculate ETPM (i.e. dry
canopy conditions). The importance of ga in controlling ETPM increased further because of the very high
precipitation at La Selva, where 32% of the time the
upper canopy was wet.
The Priestly–Taylor relationship for well-watered
conditions described the lE dynamics of this tropical
forest quite well, in particular for wet canopy
conditions (Fig. 8C). However, the Priestly–Taylor
relationship tended to overestimate the observed lE
flux for dry canopy conditions with Rn!500 W mK2.
Because the assumption that the a coefficient applies
for optimal conditions (free water and maximum Rn)
but not for hyperbolic change in gb with increasing Rn
(e.g. Fig. 6B), the fraction of available energy used for
ETPT was similar year-to-year, suggesting a thermodynamic constraint on H that limits maximum daily
temperature (Calder, 1986). Wright and Gash (1996)
and Calder et al. (1986) also found that potential
evapotranspiration accounted for a large fraction of Rn
(O0.80) from a humid Amazonian and Javanese
forest, respectively, as did Bigelow (2001) for
monocultural plantations at La Selva (0.79–0.90).
In a previous study in the old-growth forest at La
Selva, Luvall (1984) determined that the energy
required for evapotranspiration exceeded Rn by
25%. Although this seems counterintuitive, it may
very well be true. This phenomenon has been
observed over crops (Ham and Heilman, 1991)
as well as other tropical forests (Jones, 1992;
Shuttleworth, 1989; Calder, 1986). The likely explanation is that additional energy is locally advected
into the flux field from surrounding and contrasting
land use types, or from rapid movement of frontal
systems (Newson and Calder, 1989). This is certainly
possible at La Selva, where mean daytime wind
direction is w908 and the fetch is w2 km, beyond
which the landscape is dominated by pastures, crops
and patches of secondary forests extending for
w60 km to the Caribbean shore. Advection of drier
air masses with greater evaporative demand is
possible, particularly during Luvall’s study in the
early 1980s when much of the land in the Costa Rican
coastal plain was being converted from forests to
agriculture (Table 1). ETPT estimates reported here
did not exceed available Rn, suggesting advection was
not significant. But, we cannot rule out the possibility
however, that advection contributed additional
energy.
The magnitude of canopy capacity is in large part a
function of physical surface area of a canopy (Waring
and Schlesinger, 1985). At La Selva, high epiphytic
loads, bromeliad tanks and arboreal soil mats can
contribute towards additional capacity and may not
have been fully accounted for in our estimates. We
used a fixed estimate of capacity of 1.53 mm
(Loescher et al., 2002). The fraction of intercepted
rainfall was consistent from 1998 to 2000, as was
Luvall’s study. This suggests that the canopy surface
area at La Selva is often saturated, that the relative
annual amount of interception is constant, and that
changes in leaf surface area were not detectable over
our measurement period.
This old-growth wet tropical forest fairly consistently retained and recycled w50% of the annual bulk
H.W. Loescher et al. / Journal of Hydrology 315 (2005) 274–294
precipitation. For a given amount of Rn, limited water
losses to the atmosphere appears to be imposed by
forest structure and function. It remains unknown if
the 50% limit on bulk precipitation holds true for
other tropical wet forests with differing age and
structure. Annual Rn and bulk precipitation are likely
correlated to some degree. Data presented here
suggest that 88–97% of Rn is an upper bound on the
partition of bulk precipitation from tropical forests
having rainfall O2800 mm yrK1 (Table 1). How this
fraction may change with changes in climate is not
known. For example, if changes in climate increased
Rn relative to precipitation, there would likely be
ample water available for ET. In contrast, the amount
of available energy would likely limit annual ET if
annual precipitation increased relative to Rn, with a
resultant increase in runoff.
The relatively constant understory temperatures
with a modest diurnal range led to storage fluxes
that contributed very little (w2%) to the overall
daily energy fluxes. Daytime lE was always
greater than sensible heat, suggesting that the
trees in this forest have sufficient ground-water
reserves to minimize hydraulic stress among
seasons and years. Conductances followed similar
trends reported from other tropical forests, but
were higher in magnitude. Rn was the largest
determinant for the annual energy flux for this wet
tropical forest, as demonstrated by both models.
This was primarily because U was high and gb
exerted little control during daytime hours when
fluxes were at their greatest. Both the Penman–
Monteith and Priesty–Taylor models performed
well. The Priestly–Taylor model is more appropriate for wet canopy conditions that prevail for
much of time. This tropical wet forest behaved
more like a classical well-watered agronomic crop
rather than like many other forests that respond to
hydraulic limitations.
Acknowledgements
This work was supported, in part, by the US
Department of Energy Office of Science contract
IBN-9652699 within the Terrestrial Ecology and
Global Change (TECO) program as part of the
AmeriFlux network. The Florida Agricultural
291
Experiment Station, Journal Series Number is
R-10862. We are indebted to D.A. Clark, D.B.
Clark, J. Amthor, M. Leclerc, T. Meyers, T.
Prabha, and A. Karipot. Thanks also goes to D.
Hollinger for support and assistance with tower
site location, M. ja mon Schroeder for data
collection, S. Moore for programming assistance,
and two anonymous reviewers, R. Waring, S.
Mulkey, F. Putz for careful and thoughtful
reviews, OTS and SFRC for logistic support, and
lastly, to Flor de Caña for inspiring many a late
night scientific discussion.
References
Anthoni, P.M., Law, B.E., Unsworth, M.H., 1999. Carbon and water
vapor exchange of an open-canopied ponderosa pine ecosystem.
Agric. For. Meteorol. 95, 151–168.
Arya, S.P., 1988. Introduction to Micrometeorology. Academic
Press, London p. 307.
Aubinet, M., Chermanne, B., Vandenhaute, M., Longdoz, B.,
Yernaux, M., Laitat, E., 2001. Long term carbon dioxide
exchange above a mixed forest in the Belgain Ardennes. Agric.
For. Meteorol. 108, 293–315.
Avissar, R., 1993. Observations of leaf stomatal conductance at the
canopy scale—an atmospheric modeling perspective. Bound.
Layer Meteorol. 64, 127–148.
Bigelow, S., 2001. Evapotranspiration modelled from stands of
three broad-leaved tropical trees in Costa Rica. Hydrol. Proc.
15, 2779–2796.
Bousquet, P., Peylin, P., Ciais, P., Le Quere, C., Friedlingstein, P.,
Tans, P.P., 2000. Regional changes in carbon dioxide fluxes of
land and oceans since 1980. Science 290, 1342–1346.
Bradley, E., Tibshirani, R.J., 1993. An Introduction to the
Bootstrap. Chapman & Hall, New York p. 436.
Calder, I.R., 1986. What are the limits on forest evaporation?—A
further comment. J. Hydrol. 89, 33–36.
Calder, I.R., Wright, I.R., Murdiyaso, D., 1986. A study of
evaporation from a tropical rain forest-west Java. J. Hydrol.
89, 13–31.
Chen, W.Y., van den Dool, H.M., 1999. Significant change of
extratropical natural variability and potential predictability
associated with the El Niño Southern Oscillation. Tellus A51,
790–802.
Clark, D.B., Clark, D.A., 2000. Landscape-scale variation in forest
structure and biomass in a tropical rain forest. For. Ecol. Man.
137, 185–198.
Cramer, W., Kicklighter, D.W., Bondeau, A., Moore, B.,
Churkina, C., Nemry, B., Ruimy, A., Schloss, A.L., 1999.
Comparing global models of terrestrial net primary productivity (NPP): overview and key results. Glob. Change
Biol. 5, 1–15.
292
H.W. Loescher et al. / Journal of Hydrology 315 (2005) 274–294
De Bruin, H.A.R., 1983. A model for the Priestly-Taylor. J. Clin.
Appl. Meteorol. 22, 572–578.
da Rocha, H.R., Goulden, M.L., Miller, S.D., Menton, M.C.,
Pinto, L.D.V.O., de Freitas, H.C., Silva Figueira, A.M.E., 2004.
Seasonality of water and heat fluxes over a tropical forest in
eastern Amazonia. Ecol. Appl. 14, S22–S32.
Denmead, O.T., Bradley, E.F., 1985. Flux-gradient relationship in a
forest canopy. In: Hutchison, B.A., Hicks, B.B. (Eds.), The
Forest–Atmosphere Interaction. D. Reidel Publishing, Boston,
MA, pp. 421–442.
Denslow, J.S., Hartshorn, G.S., 1994. Tree-fall gap environments
and forest dynamic processes. In: McDade, L.A., Bawa, K.S.,
Hespenheide, H.A., Hartshorn, G.S. (Eds.), La Selva: Ecology
and Natural History of a Neotropical Rain Forest. University of
Chicago Press, Chicago, IL, pp. 120–127.
Dolman, A.J., Gash, J.H.C., Roberts, J., Shuttleworth, W.J., 1991.
Stomatal and surface conductance of tropical rain-forest. Agric.
For. Meteorol. 54, 303–318.
Fasullo, J., Webster, P.J., 1999. Warm pool SST variability
in relation to the surface energy balance. J. Clim. 12,
1292–1305.
Foken, T., Wichura, B., 1996. Tools for quality assessment
of surface-based flux measurements. Agric. For. Meteorol. 78,
83–105.
Grace, J., Lloyd, J., McIntyre, J., Miranda, A.C., Meir, P.,
Miranda, H.S., 1996. Carbon dioxide flux over Amazonian
rain forest in Rhondonia. In: Gash, J.H.C., Nobre, C.A.,
Roberts, J.M., Victoria, R.L. (Eds.), Amazonina Deforestation
and Climate. Institute of Hydology, pp. 307–318.
Ham, J.M., Heilman, J.L., 1991. Aerodynamic and surface
resistances affecting energy transport in a sparse crop. Agric.
For. Meteorol. 53, 267–284.
Hartshorn, G.S., Peralta, R., 1988. Preliminary description of
primary forests along the La Selva-Volcan Barva altitudinal
transect, Costa Rica. In: Alemda, F., Pringle, C. (Eds.), Tropical
Rainforests: Diversity and Conservation. California Academy of
Science, San Fransisco, pp. 281–295.
Hartmann, D.L., Moy, L.A., Fu, Q., 2001. Tropical convection and
the energy balance at the top of the atmosphere. J. Clim. 14,
4495–4511.
Hastenrath, S.L., 1991. Climate Dynamics of the Tropics. Kluwer
Academic, Boston, MA p. 488.
Holdridge, L.R., Grenke, W.C., Hatheway, W.H., Liang, T., Tosi
Jr., J.A., 1971. Forest Environments in Tropical Life Zones: A
Pilot Study. Pergamon Press, Oxford p. 747.
Hollinger, D.Y., Kelliher, F.M., Schulze, E-D., Vygodskaya, N.N.,
Varlagin, A., Milukova, I., Byers, J.N., Sogachov, A.,
Hunt, J.E., McSeveny, T.M., Kobak, K.I., Bauer, G.,
Arneth, A., 1995. Initial assessment of multi-scale measures
of CO2 and H2O flux in the Siberian taiga. J. Biogeol. 22,
425–431.
Hulme, M., Viner, D., 1998. A climate change scenario for the
tropics. Clim. Change 39, 145–176.
Jarvis, P.G., 1976. The interpretations of the variations in leaf water
potential and stomatal conductance found in canopies in the
field. Phil. Trans. R. Soc. B273, 593–610.
Jarvis, P.G., McNaughton, K.G., 1986. Stomatal control of
transpiration: scaling up from leaf to region. Adv. Ecol. Res.
15, 1–49.
Jones, H., 1992. Plants and Microclimate. Cambridge University
Press, New York p. 428.
Kaimal, J.C., Finnigan, J.S., 1994. Atmospheric Boundary-layer
Flows; Their Structure and Measurement. Oxford University
Press, Oxford, UK p. 453.
Kelly, M.A., Randall, D.A., 2001. A two-box model of a zonal
atmospheric circulation in the tropics. J. Clim. 19, 3944–3964.
Kruijt, B., Elbers, J.A., von Randow, C., Araujo, A.C.,
Oliveira, P.J., Culf, A., Manzi, A.O., Nobre, A.D., Kabat, P.,
Moors, E.J., 2004. The robustness in eddy correlation fluxes
for Amazonian rain forest conditions. Ecol. Appl. Suppl. 14,
S101–S111.
Larson, K., Hartmann, D.L., Klein, S.A., 1999. The role of clouds,
water vapor, circulation, and boundary layer structure in the
sensitivity of the tropical climate. J. Clim. 12, 2359–2374.
Lieberman, M., Lieberman, D., Hartshorn, G.S., Peralta, R., 1985.
Small-scale altitudinal variation in lowland wet tropical forest
vegetation. J. Ecol. 73, 505–516.
Livingston, N.J., Black, T.A., 1987. Stomatal characteristics and
transpiration of three species of conifer seedlings planted on a
high elevation south-facing clear-cut. Can. J. For. Res. 17,
1273–1282.
Loescher, H.W., Powers, J.S., Oberbauer, S.F., 2002. Spatial
variation of through fall volume in an old growth tropical wet
forest, Costa Rica. J. Trop. Ecol. 18, 397–407.
Loescher, H.W., Oberbauer, S.F., Gholz, H.L., Clark, D.B., 2003.
Environmental controls on net ecosystem-level carbon
exchange and productivity in a Central American tropical wet
forest. Glob. Change Biol. 9, 396–412.
Luvall, J.C., 1984. Tropical deforestation and recovery: the effects
on evaporation processes. Ph.D. Dissertation, University of
Georgia, Athens, Georgia, p. 146.
Magnussen, S., Burgess, D., 1997. Stochastic resampling techniques
for quantifying error propagations in forest field experiments.
Can. J. For. Res. 27, 630–637.
Malhi, Y., Wright, J., 2004. Spatial patterns and recent trends in the
climate of tropical rainforest regions. Phil. Trans. R. Soc.
London Ser. B 359, 311–329.
Malhi, Y., Pegoraro, E., Nobre, A.D., Pereira, M.G.P., Grace, J.,
Culf, A.D., Clement, R., 2002. Energy and water dynamics of a
central Amazonian rain forest. J. Geophys. Res. 107 (D20), 107
(art. no. 8061).
Martin, T.A., Brown, K.J., Cermák, J., Ceulemans, R., Kucera, J.,
Meinzer, F.C., Rombold, J.S., Sprugel, D.G., Hinckley, T.M.,
1997. Crown conductance and tree and stand transpiration in
a second-growth Abies amabilis Forest. Can. J. For. Res. 27,
796–808.
McKinley, G.A., Follows, M.J., Marshall, J., 2004. Mechanisms of air–sea CO2 flux variability in the equatorial
Pacific and the North Atlantic. Glob. Biogeochem. Cycles
18, GB2011–GB2025.
McMillen, R.T., 1988. An eddy-correlation technique with
extended applicability to non-simple terrain. Bound. Layer
Meteorol. 43, 231–245.
H.W. Loescher et al. / Journal of Hydrology 315 (2005) 274–294
Monteith, J.L., 1981. Evaporation and surface temperature. Q. J. R.
Meteorol. Soc. 107, 1–27.
Monteith, J.L., Unsworth, M.H., 1990. Principles of Environmental
Physics. Edward Arnold Publishers, New York p. 291.
Moore, C.J., Fisch, G., 1986. Estimating heat storage in Amazonian
tropical forest. Agric. For. Meteorol. 38, 147–169.
Neal, C., Robson, A.J., Bhardwaj, C.L., Conway, T.,
Jeffery, H.A., Neal, M., Ryland, G.P., Smith, C.J., Walls, J.,
1993. Relationships between precipitation, stemflow and
throughfall for a lowland beech plantation, Black-Wood,
Hampshire, Southern England - Findings on interception at a
forest edge and the effects of storm damage. J. Hydrol. 146,
221–233.
Newson, M.D., Calder, I.R., 1989. Forests and water resources:
problems of prediction on a regional scale. In: Jarvis, P.G.,
Montieth, J.L., Shuttleworth, W.J., Unsworth, M.H. (Eds.),
Forest, Weather and Climate. The Royal Society London, UK,
pp. 109–120.
Panofsky, H.A., Dutton, J.A., 1984. Atmospheric Turbulence,
Models and Methods for Engineering Applications. Wiley, New
York p. 397.
Priestly, C.H.B., Taylor, R.J., 1972. On the assessment of the
surface heat flux and evaporation using large scale parameters.
Month. Weath. Rev. 100, 81–92.
Raman, S., Niyogi, D.S., Prabhu, A., Ameenullah, S., Nagaraj, S.T.,
Kumar, U., Jayanna, S., 1998. VEBEX: vegetation and surface
energy balance experiment for the tropics. Proc. Ind. Acad. Sci.
107, 97–105.
Roberts, J., Cabral, O.M.R., Fisch, G., Molion, L.C.B., Moore, C.J.,
Shuttleworth, W.J., 1993. Transpiration from an Amazonian
rain-forest calculated from stomatal conductance measurements. Agric. For. Meteorol. 65, 175–196.
Ruimy, A., Jarvis, P.G., Baldocchi, D.D., Saugier, B., 1995. CO2
fluxes over plant canopies and solar radiation: a review. Adv.
Ecol. Res. 26, 1–81.
Sanford, R., Paaby, P., Luvall, J.C., Phillips, E., 1994. Climate,
geomorphology and aquatic systems. In: McDade, L.A.,
Bawa, K.S., Hespenheide, H.A., Hartshorn, G.S. (Eds.), La
Selva: Ecology and Natural History of a Neotropical
Rain Forest. University of Chicago Press, Chicago, IL,
pp. 19–33.
Schroth, G., Silva, L.F.D., Wolf, M.A., Teixeria, W.G.,
Zech, W., 1999. Distribution of throughfall and stemflow in
multi-strata agroforestry, perennial monoculture, fallow and
primary forest in central Amazonia, Brazil. Hydrol. Proc. 13,
1423–1436.
Schuepp, P.H., Leclerc, M.Y., Macpherson, J.I., Desjardins, R.L.,
1990. Footprint prediction of scalar fluxes from analytical
solutions of the diffusion equation. Bound. Layer Meteorol. 50,
355–373.
Schultz, D.M., Bracken, W.E., Bosart, L.F., 1998. Planetary- and
synoptic-scale signatures associated with central American cold
surges. Month. Weath. Rev. 126, 5–27.
Shao, J., Tu, D., 1995. The Jackknife and Bootstrap. Springer, New
York p. 516.
293
Shaw, R.H., Den Hartog, G., Neuman, H.H., 1988. Influence of
foliar density and thermal stability on profiles of Reynolds stress
and turbulence intensity in a deciduous forest. Bound. Layer
Meteorol. 45, 391–409.
Shuttleworth, W.J., 1988. Evaporation from Amazonian rainforest.
Phil. Trans. R. Soc. Lond. B233, 321–346.
Shuttleworth, W.J., 1989. Micrometeorology of temperate and
tropical forests. Phil. Trans. R. Soc. Lond. B324, 299–334.
Shuttleworth, W.J., Gash, H.C., Lloyd, C.R., Moore, C.J.,
Roberts, J., Marques Filho, A.D.O., Fisch, G., Silva
Filho, V.De.P, Nobre, J.C.A, Cabral, O.M.R., Patel, S.R.,
Moraes, J.C., 1984. Eddy correlation measurements of energy
partition for Amazonian forest. Q. J. R. Meteorol. Soc. 110,
1143–1163.
Sohn, B.J., Smith, E.A., 1992. Global energy transports and the
influence of clouds on transport requirements—a Satellite
analysis. J. Clim. 5, 717–734.
Sollins, P., Sancho, F.M., Mata, R.C., Sanford Jr.., R.L., 1994. Soils
and soil process research. In: McDade, L.A., Bawa, K.S.,
Hespenheide, H.A., Hartshorn, G.S. (Eds.), La Selva: Ecology
and Natural History of a Neotropical Rain Forest. University of
Chicago Press, Chicago, IL, pp. 34–53.
Spencer, H., Slingo, J.M., Davey, M.K., 2004. Seasonal predictability of ENSO teleconnections: the role of the remote ocean
response. Clim. Dyn. 22, 511–526.
Stewart, J.B., 1988. Modelling surface conductance of a pine forest.
Agric. For. Meteorol. 43, 19–35.
Timmermann, A., Oberhuber, J., Bacher, A., Esch, Latif, M.,
Roeckner, E., 1999. Increasing El Niño frequency in a climate
model forced by future greenhouse warming. Nature 398, 694.
Twine, T.E., Kustas, W.P., Norman, J.M., Cook, D.R., Houser, P.R.,
Meyers, T.P., Prueger, J.H., Starks, P.J., Wesely, M.L., 2000.
Correcting eddy-covariance flux underestimates over a grassland. Agric. For. Meteorol. 103, 279–300.
Ubarana, V.N., 1996. Observations and modeling of rainfall
interception at two experimental sites in Amazonia. In:
Gash, J.H.C., Nobre, C.A., Roberts, J.M., Victoria, R.L. (Eds.)
, Amazonia Deforestation and Climate. Institute of Hydrology,
pp. 151–162.
van der Tol, C., Gash, J.H.C., Grant, S.J., McNeil, D.D.,
Robinson, M.M., 2003. Average wet canopy evaporation for a
Sitka spruce forest derived using the eddy correlation-energy
balance technique. J. Hydrol. 276, 12–19.
Waring, R.H., Schlesinger, W.H., 1985. Forest Ecosystems:
Concepts and Management. Academic Press, San Diego, CA
p. 340.
Waylen, P.R., Caviedes, C.N., Quesada, M.E., 1996a. Interannual
variability of monthly precipitation in Costa Rica. J. Clim. 9,
2606–2612.
Waylen, P.R., Quesada, M.E., Caviedes, C.N., 1996b. Temporal and
spatial variability of annual precipitation in Costa Rica and the
Southern Oscillation. Int. J. Clim. 16, 173–193.
Whitehead, D., 1998. Regulation of stomatal conductance and
transpiration in forest canopies. Tree Physiol. 18, 633–644.
Wielicki, B.A., Wong, T.M., Allan, R.P., Slingo, A., Kiehl, J.T.,
Soden, B.J., Gordon, C.T., Miller, A.J., Yang, S.K.,
294
H.W. Loescher et al. / Journal of Hydrology 315 (2005) 274–294
Randall, D.A., Robertson, F., Susskind, J., Jacobowitz, H., 2002.
Evidence for large decadal variability in the tropical mean
radiative energy budget. Science 295, 841–844.
Wietz, A.M., Grauel, W.T., Keller, M., Veldkamp, E., 1997.
Calibration of time domain reflectometry technique using
undisturbed soil samples from humid tropical soils of volcanic
origin. Water Resour. Res. 33, 1241–2149.
Wright, I.R., Gash, J.H.C., 1996. Modelling surface conductance for
Amazonian pasture and forest. In: Gash, J.H.C., Nobre, C.A.,
Roberts, J.M., Victoria, R.L. (Eds.), Amazonia Deforestation
and Climate. Institute of Hydrology, pp. 437–467.
Yasuda, N., 1988. Turbulent diffusivity and diurnal variations in
the atmospheric boundary layer. Bound. Layer Meteorol. 43,
209–221.