Tellus (2003), 55B, 657–668 Copyright 

C Blackwell Munksgaard, 2003

Printed in UK. All rights reserved
TELLUS
ISSN 0280–6509

Effects of land use and fine-scale environmental

heterogeneity on net ecosystem production over
a temperate coniferous forest landscape

By DAVID P. TURNER1∗ , MICHAEL GUZY1 , MICHAEL A. LEFSKY1 , STEVE VAN TUYL1 , OSBERT
SUN1 , CHRIS DALY2 and BEVERLY E. LAW1 , 1 Department of Forest Science; 2 Department of Geosciences,
Oregon State University, Corvallis OR 97331 USA

(Manuscript received 2 January 2002; in final form 26 August 2002)

ABSTRACT
In temperate coniferous forests, spatial variation in net ecosystem production (NEP) is often associated
with variation in stand age and heterogeneity in environmental factors such as soil depth. However,
coarse spatial resolution analyses used to evaluate the terrestrial contribution to global NEP do not
generally incorporate these effects. In this study, a fine-scale (25 m grid) analysis of NEP over a
164-km2 area of productive coniferous forests in the Pacific Northwest region of the United States was
made to evaluate the effects of including fine scale information in landscape-scale NEP assessments.
The Enhanced Thematic Mapper (ETM+) sensor resolved five cover classes in the study area and
further differentiated between young, mature and old-growth conifer stands. ETM+ was also used
to map current leaf area index (LAI) based on an empirical relationship of observed LAI to spectral
vegetation indices. A daily time step climatology, based on 18 years of meteorological observations,
was distributed (1 km resolution) over the mountainous terrain of the study area using the DAYMET
model. Estimates of carbon pools and flux associated with soil, litter, coarse woody debris and live
trees were then generated by running a carbon cycle model (Biome-BGC) to a state that reflected the
current successional status and LAI of each grid cell, as indicated by the remote sensing observations.
Estimated annual NEP for 1997 over the complete study area averaged 230 g C m−2 , with most of the
area acting as a carbon sink. The area-wide NEP is strongly positive because of reduced harvesting in
the last decade and the recovery of areas harvested between 1940 and 1990. The average value was
greater than would be indicated if the entire area was assumed to be a mature conifer stand, as in a
coarse-scale analysis. The mean NEP varied interannually by over a factor of two. This variation was
38% less than the interannual variation for a single point. The integration of process models with ground
surface information provided by remote sensing provides a framework for investigating mechanisms
regulating NEP and evaluating coarse resolution globally applied NEP scaling efforts.

1. Introduction marginal agricultural lands to forestland, regrowth of

harvested forests, reversion of rangeland to woodland,
Mid-latitude forests of the northern hemisphere are and fire suppression on forestland, and (2) those asso-
currently a significant carbon sink, but the geographic ciated with environmental factors, including nitrogen
distribution of the sink and the mechanisms generating deposition, rising atmospheric CO2 and climate warm-
it are uncertain (Fan et al., 1998; Ciais et al., 2000). ing (Pacala et al., 2001). The northern hemisphere ter-
The principal factors contributing to this sink include restrial sink is also subject to considerable interannual
(1) those associated with land use, notably reversion of variation, the mechanisms of which are likewise poorly
understood (Bousquet et al., 2000). From both policy
∗ Corresponding author. and science perspectives, there is a need to develop
e-mail: [Link]@[Link] improved approaches to scaling carbon cycle

Tellus 55B (2003), 2

658 D. P. TURNER ET AL.

processes such that the mechanisms accounting for suggested that spatial resolution of ≤250 m is desir-
the terrestrial carbon sources and sinks in different re- able for assessing spatial patterns in land cover and
gions, and their interannual variation, are understood. NEP in this region (Cohen et al., 1995; Turner et al.,
A critical variable for understanding the terrestrial 2000).
carbon flux is net ecosystem production (NEP), the
balance of gross photosynthesis and ecosystem respi-
ration (often calculated from biological measurements
2. Methods
as the net primary production minus the heterotrophic
respiration). NEP is sensitive to stand age, climate and
2.1. Overview
local factors such as soil depth. Forest stands typically
exhibit a negative NEP, i.e. act as a source of CO2 A biogeochemistry process model (Biome-BGC)
to the atmosphere, early in succession when residues was applied over a 25 m grid covering a 164-km2
from the recent disturbance are rapidly decomposing area of managed coniferous forests in western Ore-
and vegetation NPP has not recovered from the dis- gon. Model initiation required land cover classifica-
turbance (Sprugel, 1985; Schulze et al., 2000). NEP is tion, stand age and a reference leaf area index (LAI),
expected to be highest in mid-succession and to de- all of which were provided by satellite remote sensing.
cline in late succession as the carbon pools approach The model was driven with a distributed daily clima-
a steady state. In addition to stand age effects on tology. Model outputs included NPP, Rh and NEP.
NEP, local environmental heterogeneity and interan-
nual climate variation can affect NEP by dispropor-
tionately changing net primary production (NPP) and 2.2. Study area
heterotrophic respiration (Rh).
The study area surrounds the H.J. Andrews Exper-
Process-based biogeochemistry models represent
imental Forest (HJA) within the Willamette National
working hypotheses about the important mechanisms
Forest in the state of Oregon. The climate is temperate,
regulating NEP (Ryan et al., 1996). Both prognostic
with wet winters and dry summers. Mean annual pre-
models (e.g. Kicklighter et al., 1999), which are typi-
cipitation at the Headquarters meteorological station
cally driven by distributed climate data, and diagnostic
is 227 cm and mean annual temperature is 8.2 ◦ C. The
models (e.g. Potter et al., 1993), which employ re-
dominant conifer species are Douglas fir (Psuedotsuga
mote sensing to characterize solar radiation absorbed
menziesii) western hemlock (Tsuga heterophylla) and
by the canopy, provide insights into potential causes
western redcedar (Thuja plicata).
of the spatial and temporal patterns of terrestrial NEP
inferred by inverse modeling (Fan et al., 1998). How-
ever, these analyses have generally been made at rela-
2.3. Land cover
tively coarse spatial resolutions (grid cells 10–100 km
on a side) that do not account for land use and environ- Land cover for the study area was determined by up-
mental heterogeneity likely influencing NEP in many dating a previously published land cover map for 1988
forested areas (White and Running, 1994; Turner et al., (Cohen et al., 1998; 2001; 2002). In the creation of the
2000). 1988 map, satellite imagery was first used to develop
In this study, the Landsat Enhanced Thematic Map- continuous (0–100%) data layers for total foliage cover
per sensor (ETM+) with a spatial resolution of ∼30 m and conifer cover. These data layers were the basis for
was employed to initialize a process-based carbon creating five forest classes (Open: total cover <30%;
cycle model and examine the sensitivity of simulated Semi-open: 30% < total cover < 70%; Deciduous:
landscape-scale NEP to the inclusion of high spatial total cover > 70% and conifer cover < 30%; Mixed:
resolution satellite data. Relatively high spatial resolu- total cover > 70% and 30% < conifer cover < 70%;
tion climate and soils data were used in the simulation Conifer: total cover and conifer cover > 70%). There
as well. The study was conducted in the Pacific North- were also non-forest (primarily urban and agricultural
west region of the US, which is of particular interest lands) and water classes based on information from
with regard to the carbon cycle because of the large National Land Cover coverages (Vogelmann, 1998).
pools of carbon in biomass, the high productivity of In all, land cover mapping was performed over an area
the forests and the large impacts of management in that included much of western Oregon; the area around
recent decades (Cohen et al., 1996). Earlier work has HJA was subset for this study.

Tellus 55B (2003), 2

 NEP ANALYSIS OVER A CONIFEROUS FOREST 659

In the initial phase of the updating procedure, an Table 1. Landsat ETM+ cover types in the study area
analysis of changes associated with stand replacement
disturbance was performed using Landsat ETM+ im- Cover type Area (km2 ) Representative age
agery from the years 1988 and 2000. With the year
Open 2.2 8
2000 imagery, the reflectances in each of the seven in- Semi-open 28.3 20
dividual Landsat scenes covering the area of interest Conifer, Young 15.5 65
(Rows 28, 29 and 30 for Paths 45 and 46, and Row 28 Conifer, Mature 37.2 150
for Path 47) were first transformed using the “tasseled- Conifer, Old growth 41.4 400
cap” method of Crist and Cicone (1984). The tasseled- Closed mixed forest 34.7 30
cap transformation is a series of three indices, specif- Deciduous forest 4.0 50
Other 0.7 –
ically designed for use with Landsat images, which
summarizes spectral variation. It is intended to en- Total 164
hance the vegetation components of imagery, gener-
ally by contrasting them against soil and background
components. The corresponding tasseled-cap image
for 1988 was then obtained from the LARSE (Labora- These were aggregated to three conifer age classes
tory for Applications of Remote Sensing in Ecology, [young (30–100 yr), mature (101–200 yr) and old
Oregon State University) archive, to create a differ- (>200 yr)], and this information was carried over to
ence image. As in Cohen et al. (1998), iterative unsu- the year 2000 land-cover map. Representative ages for
pervised classification was used to define change and these classes were set to the mid-points of their age
no-change classes by visual inspection. ranges. Representative ages for the open, semi-open,
To apply the original 1988 classification to the areas mixed and deciduous classes (Table 1) were approxi-
that had changed, the 2000 imagery was first radiomet- mated from aged reference stands (Cohen et al., 1995;
rically normalized to the 1988 image employing the Turner et al., 2000).
ridge regression procedure (Song et al., 2001), using
only those areas that had been classified, in the pre-
2.4. Leaf area index
vious analysis step, as “no-change”. Ridge regression
is a technique used to match the spectral qualities of Leaf area index (LAI, half total leaf surface area)
two scenes by comparing their values in the area in values were measured on 76 plots, each 100 × 100 m,
which they overlap. Before land cover could be clas- at 13 locations broadly distributed throughout west-
sified, the 2000 data had to be further transformed to ern Oregon (hierarchical random sampling design).
duplicate the underlying variables used in creating the All plots were georeferenced using a high-resolution
1988 classification, namely green cover and conifer Global Positioning System instrument (Trimble
cover. To ensure that 2000 values for these variables Navigation Ltd, Sunnyvale, CA). The measurements
matched those for 1988, new regressions between the of LAI were made with a LAI-2000 (LICOR,
three tasseled cap spectral vegetation indices (bright- Lincoln, NE) in diffuse light conditions, and corrected
ness, greenness and wetness) from the 1988 images for leaf clumping within shoots, clumping at scales
and the 1988 estimates of green cover and conifer larger than shoot, and wood interception (Chen, 1996;
cover (percent) were performed for each scene. The Law et al., 2001a). Clumping within shoots was deter-
resulting equations were then applied to the year 2000 mined by taking into consideration the species com-
tasseled-cap images and changed areas were classified. position of the stand and the clumping factors for in-
Results from an independent verification of the final dividual conifer species found in the literature (Law
cover map (using aerial photography) indicate better et al., 2001a; Gower et al., 1999; Frazer et al., 2000).
than 80% accuracy in predicting the five forest cover Clumping at scales larger than shoot was determined
classes. from optical measurements with a TRAC device (3rd
For the purposes of initializing the biogeochem- Wave Engineering, Ontario, Canada) along two 100
istry model (see below) a representative stand age was m transects on each plot (Law et al., 2001a). Wood
needed. Cohen et al. (2001) concluded that Landsat- interception was calculated as half total surface area
TM could only estimate age accurately in the conifer of stems and branches.
class, and had created continuous estimates of conifer A single mosaic of Landsat ETM+ was created
age for the 1988 land cover map (Cohen et al., 2001). from six scenes using the ridge regression procedure

Tellus 55B (2003), 2

660 D. P. TURNER ET AL.

(Song et al., 2001). After the images were assem- Because most stands in the PNW region originated
bled, a dark-object subtraction was performed to re- from catastrophic disturbances (Wallin et al., 1996),
move atmospheric effects, following the conclusions two successive disturbances were simulated at the end
of Turner et al. (1999). Polygons were hand-digitized of each spin-up such that 1/3 of the live tree carbon
around each of the plots in the Landsat ETM+ imagery was transferred to the CWD pool at each disturbance
to ensure that a homogenous region was used in the (Turner et al., 1995). The disturbances were 90 yr apart
comparison of spectral characteristics and LAI. Both for all forest classes except the shorter-lived decidu-
the tasseled-cap indices and the normalized difference ous forest, where the interval was 60 yr. The model
vegetation index (NDVI) were calculated from the re- was then run forward in a final “succession run” to
sulting mosaic and stepwise multiple regressions were the representative stand age indicated by the remote
used to determine the best set of variables for predict- sensing classification (Table 1). This brought the sim-
ing LAI. The same equation was then used across all ulated live tree, CWD, litter and soil carbon pools into
cover types to generate a LAI data layer. agreement with the stand age.
To bring simulated LAI into agreement with the ref-
erence remote sensing-based LAI, spin-ups were run at
2.5. Meteorological data a range of soil depths. Potential LAI in PNW conifer
forests is closely related to water availability (Grier
The distributed meteorological data used to drive
and Running, 1977) and hence to soil depth (which
the model was from 18 yr (1980–1997) of daily me-
influences soil water holding capacity). The model is
teorological observations at a network of sites around
self-regulating with respect to LAI, so increasing soil
the conterminous United States. The meteorological
depth tends to result in an increase in the maximum
station data were interpolated to a 1 km grid using the
achievable LAI. Thus, by running the spin-up process
DAYMET program (Thornton et al., 1997; Thornton
at a series of soil depths in each cell of the climate
and Running, 1999; Thornton et al., 2000). The result-
grid, a range of potential LAIs was generated (Fig. 1).
ing database included daily maximum and minimum
Soil texture was first specified from Kern et al. (1997),
temperature, precipitation, daytime average solar ra-
where data layers for % sand, silt and clay were de-
diation and daytime average vapor pressure deficit.
veloped from the State Soil Geographic (STATSGO)
The gridded data were compared with meteorological
database and the US National Soil Characterization
observations made near the H.J. Andrews Headquar-
Database. Soil depth was then assigned based on the
ters (HJA, 2001).
best agreement between the simulated LAI and the re-
mote sensing based reference LAI. To limit the num-
ber of computationally intensive spin-ups, a mean LAI
2.6. Model application
per cover type (cover class by age class combination,
Biome-BGC is a daily time-step biogeochemistry Table 1) was determined for each 1 km cell and used as
model with physiologically based algorithms for pho- the reference LAI for that cover type and that cell. Then
tosynthesis, autotrophic respiration and heterotrophic
respiration (Running and Hunt, 1993; Thornton, 1998;
14
Thornton et al., 2002). A set of ecophysiological con-
0.2 m
stants, e.g. maximum stomatal conductance, is pre- 12 0.5 m
scribed for each forest cover type (White et al., 2000).
Leaf Area Index (m2 m2)

0.8 m
10
In this study, all cover types except the deciduous class 1.0 m
were run as conifer. The major carbon compartments, 8
or pools, include leaves and fine roots as well as bole,
6
coarse roots, coarse woody debris (CWD), litter and
two classes of soil organic matter. 4
To establish the initial conditions for those pools,
2
a model “spin-up” is run. The soil carbon pools are
brought into approximate equilibrium with the lo- 0
0 100 200 300 400
cal climate during this thousand-year model run. In
Year
the spin-up, the 18-yr climate time series was run
repeatedly. Fig. 1. Effect of alternative soil depths on simulated LAI.

Tellus 55B (2003), 2

 NEP ANALYSIS OVER A CONIFEROUS FOREST 661

all 25 m grid cells of that cover type were assigned the variation in NEP, while controlling for all vegetation
same soil depth based on the multiple spin-ups. In factors, the meteorological data for each of the other
the case of the relatively young age classes (open and 17 yr was successively substituted for the 1997 data at
semi-open), which may not have reached their maxi- the end of the base year run.
mum LAI, soil depth was set to the value derived by
Kern et al. (1997). The young stand age then limits the
achieved LAI and tends to generate agreement with the 3. Results
remote sensing based reference LAI. The root mean
square error for the reference LAI generated by remote The land cover in the study area (Fig. 2a) is pre-
sensing and the model generated LAI was 0.6 m2 m−2 dominantly coniferous forest. Sixty-nine percent of
over the complete study area. the area is closed canopy conifer forest, and much
For development of the base year 1997 NEP data of the remainder is land recovering from previous dis-
layer, the climate files were structured such that the turbance (Table 1). The land-cover map reveals dis-
year when the stand reached its representative age was tinct patchiness, with polygons corresponding primar-
the 1997 meteorological year. To examine interannual ily to areas clear-cut for timber harvest since 1940.

Fig. 2. (a) Land cover data layer for the H. J. Andrews study area. The location of the site headquarters in the lower left
corner of the figure is 44◦ 12
N, 122◦ 14
W. (b) LAI reference data layer for the H. J. Andrews study area.

Tellus 55B (2003), 2

662 D. P. TURNER ET AL.

16 layer (Fig. 2b) revealed patchiness associated with pre-

14 vious clear cuts.
Predicted LAI (m2 m-2)

Meteorological observations at HJA (not used in

12
creation of the gridded data) indicated a mean annual
10
temperature over the 18 yr within 1 ◦ C of the value
8 from the interpolated meteorological data. Mean an-
6 nual precipitation from the local measurements was
4
16% greater than that for the interpolated data.
Simulated successional trends in NPP and Rh for
2
RMSE = 1.6 a mid-elevation (950 m) conifer stand indicated a
0 large negative NEP predicted for early in succession,
0 2 4 6 8 10 12 14 16
a cross-over from source to sink at a stand age of 10–
Observed LAI (m2 m-2) 20 yr, a maximum NEP at ages 30–50 yr, followed
by a declining NEP until the old-growth stage which
Fig. 3. Observed and predicted LAI for the LAI measure-
ment plots. The one-to-one line is indicated. was nearly at carbon steady state (Fig. 5). The decline
in NEP is primarily the result of a decline in NPP,
which is driven by low nitrogen availability associated
Nineteen percent of the study area is in the open and with decay of dead wood generated by an increasing
semi-open classes, which are predominantly recent input from mortality. The average NEP by age class
clear cuts. Twenty-seven percent of the land is old- over the study area was consistent with this temporal
growth conifer forest. pattern (Fig. 6). For the complete study area, the av-
The optical measurements of LAI corrected for erage NEP for 1997 was 230 g C m−2 yr−1 (Figs. 6
clumping and wood interception resulted in values and 7).
ranging from 0.2 to 12.6 m2 m−2 . The best-fit equa- For a mid-elevation grid cell classified as old
tion relating measured LAI to ETM+ reflectances conifer, there was a large interannual variation in NEP
used the brightness and wetness indices and explained (Fig. 8). For the study area as a whole, the interannual
80% of variance. Most of the error was at LAI > 5.0 variation in NEP was about a factor of two, but that
(Fig. 3), where variation in LAI has little effect on range was only 62% of the range for the old conifer
simulated NEP. The range of LAI estimates extended stand (Fig. 8). Interannual NEP values ranged from
to 12 m2 m−2 , in agreement with the optical measure- −70 to 183 gC m−2 yr−1 at the old conifer stand, and
ments and with maximums recently measured in other 106 to 282 gC m−2 yr−1 for the study area as a whole.
Cascade forests (Thomas and Winner, 2000). Mean
LAI within a cover type increased from Open, to Semi-
open to Young Conifer then decreased in the oldest
4. Discussion
classes (Fig. 4). The remote sensing based LAI data

12
4.1. Sensitivity of scaled NEP to fine spatial
resolution input data
Leaf Area Index (m2 m-2)

10
The large differences in mean NEP as a function
8
of age class (Fig. 6) are indicative of the potential
6 error if this area were modeled as one large cell with
a single cover class. The prognostic NEP models are
4
generally run to near steady state before perturbations
2 (such as CO2 increase) are imposed, thus their results
0
would most likely resemble the behavior of the mid
Open Semi-open Young Mature Old elevation old-growth stand. In that case, average NEP
Cover Class over the 18-yr climate time series was 56 gC m−2 yr−1 ,
considerably less that the 230 gC m−2 yr−1 estimated
Fig. 4. Mean (and standard deviation) of leaf area index
as determined from remote sensing by cover class. The over the study area when cover type and stand age were
Open and Semi-open classes are early stages in the conifer taken into account. Examination of coarse scale NEP
succession. estimates that do not account for land use generally

Tellus 55B (2003), 2

 NEP ANALYSIS OVER A CONIFEROUS FOREST 663

1500

1000

Carbon Flux (gC m yr )

-1
-2
500

-500
Net Primary Production
-1000 Heterotrophic Respiration
Net Ecosystem Production
-1500
0 100 200 300 400
Year
Fig. 5. Trends in net primary production, heterotrophic respiration and net ecosystem production during succession for a
mid-elevation conifer stand.

show NEP values less than 50 gC m−2 yr−1 in the 1996). Federal lands in the study area are now in a
vicinity of the area studied here (e.g. Woodward et al., phase of recovery from disturbance during the period
2001). 1940–1990, thus they are largely carbon sinks. Fire
NEP is notably high in this area because harvesting and logging need to be considered in a complete car-
on public lands has slowed in recent decades (Fig. 9), bon budget (Turner et al., 1995; Schulze et al., 2000),
leaving large areas in the young, high NEP, age classes but neither was significant in the 1997 reference year
(age 20–100 yr). During former periods of rapid cut- for the study area.
ting, a larger proportion of the landscape had been The spatially explicit LAI extends the utility of the
recently cut and was acting as a carbon source. The age-based land cover classification by providing infor-
process of converting a landscape dominated by pri- mation on within-class variation of a model variable
mary forests to one dominated by secondary forests that strongly influences NPP. For a variety of site qual-
results in a sustained carbon source to the atmosphere, ity or site history factors, many PNW forest stands are
even when the carbon sink associated with forest at less than full stocking. The less vigorous stands
products is taken into account (Harmon et al., 1990; tend to have relatively low LAI (Gholz, 1982), which
is likely to be detectable by remote sensing (Peterson
et al., 1987).
The linkage of simulated LAI to remote sensing
Area-wide based LAI worked well in this study because of the
Deciduous strong influence of site water balance on LAI in the
Mixed PNW region. In other regions, ecosystem attributes
Old such as foliar nitrogen concentration and canopy ni-
Mature trogen content [also potentially detectable by remote
Young sensing (Martin and Aber, 1987)] are believed to be
Semi-open the dominant regulators of NPP. A major challenge in
Open scaling studies is to incorporate information from re-
mote sensing into a simulation modeling framework
-600 -400 -200 0 200 400 600
-2 -1
without compromising the model’s capacity for self-
Net Ecosystem Production (gC m yr ) regulation during simulation of stand development.
Fig. 6. Mean net ecosystem production by cover type in In this study, there was a tendency in the simulations
1997. to underestimate LAI in the Open class relative to the

Tellus 55B (2003), 2

664 D. P. TURNER ET AL.

Fig. 7. Net ecosystem production for the H. J. Andrews study area in 1997.

remotely sensed reference LAI values. This discrep- lated temperatures (Turner et al., 1996). The more
ancy was related to a lag in simulated production after restricted range of temperatures at coarse resolution
clearcuts. The heavy nitrogen demand associated with results in a lower range for respiration rates, with cor-
decomposition of the post-harvest residues restricted responding effects on NPP and NEP.
NPP and hence LAI in the model simulations. A po- The difference in the NEP range over the 18-yr test
tential approach to improving the simulations may be period for a point and as an average over the study
to assume significant nitrogen fixation early in succes- area (Fig. 8) suggests that incorporation of fine-scale
sion, as has been widely observed in early successional data lowers sensitivity to climatic variation. The area-
PNW conifer forests (Binkley et al., 1992). wide response is moderated to some degree because
Besides the high-resolution land cover and LAI, the of counteracting responses to climate by different age
1-km distributed meteorology was also beneficial to classes and different locations on the environmental
the analysis. The algorithms in Biome-BGC for pho- gradients.
tosynthesis, autotrophic respiration and heterotrophic
respiration all employ non-linear temperature-based
4.2. Prospects for validation of landscape-scale
functions. Earlier analysis in this mountainous region
NEP estimates
has shown how spatial resolution influences interpo-
The 25 m spatial resolution of this analysis closely
matches the plot size used in measurements of forest
Net Ecosystem Production (gC m-2 yr-1)

300

200 14000
Harvest (Million Board Feet)

12000 Private Lands

100 Public Lands
10000
0
8000
-100 6000
Area-wide Mean NEP 4000
-200
Point-based NEP
2000
-300
1980 1985 1990 1995 0
Year 1965 1970 1975 1980 1985 1990 1995 2000
Year
Fig. 8. Interannual variation in mean net ecosystem produc-
tion for the complete study area and for a mid elevation cell Fig. 9. Harvest volumes on public and private land in the
classified as old conifer. Pacific Northwest region.

Tellus 55B (2003), 2

 NEP ANALYSIS OVER A CONIFEROUS FOREST 665

wood production and NPP (Gower et al., 1999). Wood The heterotrophic respiration component of simu-
production is usually a large component of NEP and lated NEP is more difficult to validate that the NPP
provides a useful measure for partial model validation component. Chronosequence studies give some indi-
(Barford et al., 2001). Thus the network of perma- cation of mass loss of CWD over time, and the FIA
nent plots maintained by national forestry agencies, Program is beginning to include estimation of woody
e.g. the US Forest Inventory and Analysis (FIA) Pro- debris mass at a subset of the permanent plot net-
gram (USDA, 1992), represents a potential source of work (Waddell, 2002). Chamber-based studies of soil
validation data over large forested areas. However, the respiration (e.g. Davidson et al., 1998) are helping
information on plot locations and volume increments to quantify litter and soil organic matter decompo-
may be poorly documented or unavailable. The loca- sition rates, but it remains problematic to isolate het-
tions of the FIA plots in the Pacific Northwest are not erotrophic from autotrophic respiration (Hanson et al.,
released and plot level data are not available except 2000).
as raw diameter distributions. Because of the policy Eddy covariance flux towers can be used to verify
relevance of forests to national carbon accounting, in- simulated NEP. At tower sites, aggregated half-hourly
creased attention should be paid to making locations measurements of net ecosystem exchange can be com-
and associated bolewood productions estimates avail- pared to simulated daily carbon flux values to reveal
able to carbon-cycle researchers. The FIA Program model effectiveness with regard to day-to-day varia-
does report summary statistics on volume increment tion in site meteorology (Aber et al., 1996). In the PNW
at the county and state levels, and these data will pro- region, there are flux towers at three conifer sites (Law
vide validation at an aggregate level as the spatial scale et al., 2001b; Chen et al., 2002). Initial comparison of
of the modeling grows to include complete political flux site data with simulations from Biome-BGC indi-
units. cate that model estimates of major carbon flux compo-
Bolewood production estimates based on FIA data nents agree with budget-based observations to within
(non-georeferenced) for the Willamette National For- ±20%, with larger differences for NEP and for several
est, which surrounds the study area, decline signifi- storage terms (Law et al., 2001b). Annual NEP values
cantly after age 50 (Fig. 10). The temporal trend in at tower sites require filling in missing data (Goulden
bole production simulated by Biome-BGC shows a et al., 1996a), but coherent patterns in interannual vari-
similar pattern, lending confidence to the simulated ation in NEP are beginning to emerge from tower stud-
trends in NPP and NEP. The large variation in bole ies (Goulden et al., 1996b; 1998). Multiple-year ob-
production for a given stand age indicates the impor- servations of NEP may thus provide valuable checks
tance of accounting for site factors other than age class on model sensitivity to interannual climate variation.
(e.g. LAI and climate data) in the simulations.
4.3. Relationship of fine-scale
600 to coarse-scale analyses
Bolewood NPP (gC m-2 yr-1)

500 Global NEP models run at coarse resolution are in-

creasingly providing insights into spatial and tempo-
400
ral patterns in terrestrial carbon flux. These flux es-
300 timates are compared with results from global-scale
inverse modeling studies, or interannual anomalies in
200 the rate of CO2 increase in the atmosphere (e.g. Ito
and Oikawa, 2000). The modeled NEP estimates are
100
rarely evaluated for individual cells because of the
0 mismatch in scale between the 10–100 km cell size
0 100 200 300 400 500 of the simulations and the much finer scale of car-
Stand Age (years) bon flux measurements. The high spatial resolution
approach to scaling NEP used in this study provides
Fig. 10. Bole production estimates for all permanent plots
a means to bridge the gap between the scale of the
in the Willamette National Forest. Data on diameter distribu-
tions and 5-yr increments by species (FIA, 2001) were con- carbon flux measurements and the coarse resolution
verted to biomass and carbon using allometric relationships currently employed for global-scale NEP simulations.
in BIOPAK (Means et al., 1994). Comparisons of one or more cells in selected regions

Tellus 55B (2003), 2

666 D. P. TURNER ET AL.

could greatly inform model development at the global 5. Conclusions

scale.
The fine scale approach described in this paper could Globally applied NEP simulations are necessarily
potentially be extended beyond the landscape scale restricted to spatial resolutions with cell sizes on the
to the regional domain. Areas of intensive manage- order of 10–100 km on a side. At those coarse scales,
ment and land-use change, especially where carbon most information on land use and local environmental
stocks and growth rates are large, are of greatest inter- heterogeneity is omitted. Landscape to regional an-
est. This extension is not constrained by availability of alyzes which rely on fine-resolution remote sensing
satellite data. The ETM+ sensor and Moderate Imag- (≤250 m) to characterize land cover and LAI can sup-
ing Spectroradiometer (MODIS) sensor are provid- plement coarse scale analyses by examining effects of
ing global coverage suitable for mapping land cover fine-scale heterogeneity on mean NEP and sensitivity
to relevant scales. Mesoscale climate models which of mean NEP to interannual climate variation.
use General Circulation Models, or observations, for
boundary conditions are increasingly able to provide
distributed meteorological data appropriate for fine- 6. Acknowledgements
scale simulation modeling at the regional scale (e.g.
Pielke et al., 1996). As noted, the logistical constraints This research was supported by the U.S. Environ-
on the number of field measurements that can be made mental Protection Agency STAR Program on Regional
in support of fine scale simulation modeling are sig- Scale Analysis and Assessment (Grant no. R828309).
nificant, hence government-supported permanent plot Thanks are due to the US Forest Service for the per-
networks and the flux tower networks must be increas- manent plot data for the Willamette National Forest, to
ingly relied upon. Implementation of the fine-scale Darius Adams (Oregon State University) for the har-
approach over a regional domain opens the possibil- vest statistics, and to Scott Waichler (Battelle Pacific
ity of comparisons with fluxes inferred from inverse Northwest Laboratories) for the filled-in meteorolog-
modeling. ical data for the H.J. Andrews Experimental Forest.

REFERENCES

Aber, J. D., Reich, P. B. and Goulden, M. L. 1996. Extrap- Ciais, P., Peylin, P. and Bousquet, P. 2000. Regional bio-
olating leaf CO2 exchange to the canopy: a generalized spheric carbon fluxes as inferred from atmospheric CO2
model of forest photosynthesis validated by eddy correla- measurements. Ecol. Appl. 10, 1574–1589.
tion. Oecologia 106, 267–275. Cohen, W. B., Spies, T. A. and Fiorella, M. 1995. Estimating
Barford, C. C., Wofsy, S. C., Goulden, M. L., Munger, J. W., the age and structure of forests in a multi-ownership land-
Pyle, E. H., Urbanski, S. P., Hutyra, L., Saleska, S. R., scape of western Oregon, U.S.A. Int. J. Remote Sensing
Fitzgarrald, D. and Moore, K. 2001. Factors controlling 16, 721–746.
long- and short-term sequestration of atmospheric CO2 in Cohen, W. B., Harmon, M. E., Wallin, D. O. and Fiorella,
a mid-latitude forest. Science 294, 1688–1691. M. 1996. Two decades of carbon flux from forests of the
Binkley, D., Sollins, P., Bell, R., Sachs, D. and Myrold, Pacific Northwest. BioScience 46, 836–844.
D. 1992. Biogeochemistry of adjacent conifer and alder- Cohen, W. B., Fiorella, M., Gray, J., Helmer, E. and
conifer stands. Ecology 73, 2022–2033. Anderson, K. 1998. An efficient and accurate method for
Bousquet, P., Peylin, P., Ciais, P., Quere, C. L., Friedlingstein, mapping forest clearcuts in the Pacific Northwest using
P. and Tans, P. P. 2000. Regional changes in carbon diox- Landsat imagery. Photogram. Eng. Remote Sensing 64,
ide fluxes of land oceans since 1980. Science 290, 1342– 293–300.
1346. Cohen, W., Maiersperger, T. K., Spies, T. A. and Oetter,
Chen, J. M. 1996. Optically-based methods for measuring D. R. 2001. Modeling forest cover attributes as continuous
seasonal variation of leaf area index in boreal conifer variables in a regional context with Thematic Mapper data.
stands. Agric. For. Meteorol. 80, 135–163. Int. J. Remote Sensing 22, 2279–2310.
Chen, J., Falk, M., Euskirchen, E., PawU, K. T., Suchanek, Cohen, W. B., Spies, T. A., Alig, R. J., Oetter, D. R.,
T. H., Ustin, S. L., Bond, B. J., Brosofske, D. D., Phillips, Maiersperger, T. K and Fiorella, M. 2002. Characterizing
N. and Bi, R. 2002. Biophysical controls of carbon flows 23 years (1972–1995) of stand replacement disturbance in
in three successional Douglas-fir stands based on eddy- western Oregon forests with Landsat imagery. Ecosystems
covariance measurements. Tree Physiol. 22, 169–177. 5, 122–137.

Tellus 55B (2003), 2

 NEP ANALYSIS OVER A CONIFEROUS FOREST 667

Crist, E. P. and Cicone, R. C. 1984. A physically-based trans- United States. In: Soil processes and the carbon cycle (eds.
formation of Thematic Mapper data – the TM Tasseled R. Lal, J. M. Kimbal, R. Follett and B. A. Stewart). CRC
Cap. IEEE Trans. Geosci. Remote Sensing GE-22, 256– Press, Boca Raton FL, 29–43.
263. Kicklighter, D. W., Bruno, M., Donges, S. et al. 1999. A
Davidson, E. A., Belk, E. and Boone, R. D. 1998. Soil water first-order analysis of the potential role of CO2 fertilization
content and temperature as independent or confounding to affect the global carbon budget: a comparison of four
factors controlling soil respiration in a temperate mixed terrestrial biosphere models. Tellus 51B, 343–366.
hardwood forest. Glob. Change Biol. 4, 217–227. Law, B. E., Van Tuyl, S., Cescatt, i A. and Baldocchi, D. D.
Fan, S., Gloor, M., Mahlman, J., Pacala, S., Sarmiento, J., 2001a. Estimation of leaf area index in open-canopy pon-
Takahashi, T. and Tans, P. 1998. A large terrestrial carbon derosa pine forests at different successional stages and
sink in North America implied by atmospheric and oceanic management regimes in Oregon. Agric. For. Meteorol. 108,
carbon dioxide data and models. Science 282, 442–446. 1–14.
FIA. 2001. U.S. Forest Inventory and Analysis web site. Law, B. E., Thornton, P., Irvine, J., Van Tuyl, S. and Anthoni,
[Link] us/r6/survey/. P. 2001b. Carbon storage and fluxes in ponderosa pine
Frazer, G. W., Trofymow, J. A. and Lertzman, K. P. 2000. forests at different developmental stages. Glob. Change
Canopy openness and leaf area in chronosequences of Biol. 7, 755–777.
coastal temperate rainforests. Can. J. For. Res. 30, 239– Martin, M. E. and Aber, J. D. 1997. High spectral resolu-
256. tion remote sensing of forest canopy lignin, nitrogen and
Gholz, H. L. 1982. Environmental limits on above ground net ecosystem processes. Ecol. Appl. 7, 431–443.
primary production, leaf area and biomass in vegetation Means, J. E., Hansen, H., Koerper, G., Alaback, P. B. and
zones of the Pacific Northwest. Ecology 63, 469–481. Klopsch, M. W. 1994. Software for computing plant
Goulden, M. L., Munger, J. W., Fan, S., Daube, B. C. and biomass – BIOPAK users guide. USDA For. Serv. Gen.
Wofsy, S. C. 1996a. Measurements of carbon sequestra- Tech. Rep. PNW-GTR-340. 184 pp.
tion by long-term eddy covariance: methods and a critical Pacala, S. W., Hurtt, G. C., Baker, D. and coauthors. 2001.
evaluation of accuracy. Glob. Change Biol. 2, 169–182. Consistent land- and atmosphere-based U.S. carbon sink
Goulden, M. L., Munger, J. W., Fan, S.-M., Daube, B. C. and estimates. Science 292, 2316–2322.
Wofsy, S. C. 1996b. Exchange of carbon dioxide by a de- Peterson, D. L., Spanner, M. A., Running, S. W. and Teuber,
ciduous forest: response to interannual climate variability. K. B. 1987. Relationship of Thematic Mapper Simulator
Science 271, 1576–1578. data to leaf area index of temperate coniferous forests.
Goulden, M. L., Wofsy, S. C., Harden, J. W., Trumbore, S. E., Remote Sensing Environ. 22, 323–341.
Crill, P. M., Gower, S. T., Fries, T., Daube, B. C., Fan, Pielke, R. A., Baron, J., Chase, T., Copeland, J., Kittel,
S.-M., Sutton, D. J., Bazzaz, A. and Munger, J. W. 1998. T. G. F., Lee, T. J., Walko, R. and Zeng, X. 1996. Use
Sensitivity of boreal forest carbon balance to soil thaw. of mesoscale models for simulation of seasonal weather
Science 279, 214–217. and climate change for the Rocky Mountain States. In:
Gower, S. T., Kucharik, C. J. and Norman, J. M. 1999. GIS and environmental modeling: progress and research
Direct and indirect estimation of leaf area index, f APAR , issues (eds. M. F. Goodchild, L. T. Steyaert, B. O. Parks,
and net primary production of terrestrial ecosystems. Re- C. Johnston, D. Maidment, M. Crane and S. Glendinning).
mote Sens. Environ. 70, 29–51. GIS World, Inc., Ft. Collins, CO, 99–103.
Grier, C. C. and Running, S. W. 1977. Leaf area of mature Potter, C. S., Randerson, J. T., Field, C. B., Matson, P. A.,
northwestern coniferous forests: relation to site water bal- Vitousek, P. M., Mooney, H. A. and Klooster, S. A. 1993.
ance. Ecology 58, 893–899. Terrestrial ecosystem production, a process model based
Hanson, P. J., Edwards, N. T., Garten, C. T. Jr. and An- on global satellite and surface data. Glob. Biogeochem.
drews, J. A. 2000. Separating root and microbial contribu- Cycles 7, 811–841.
tions to soil respiration: A review of methods and results. Running, S. W. and Hunt, E. R. Jr. 1993. Generalization
Biogeochem. 48, 115–146. of a forest ecosystem process model for other biomes,
Harmon, M. E., Ferrell, W. K. and Franklin, J. F. 1990. Effects BIOME-BGC and an application for global scale mod-
on carbon storage of conversion of old-growth forests to els. In: Scaling physiological processes: leaf to globe (eds.
young forests. Science 247, 699–7002. J. R. Ehleringer and C. Fields). Academic Press, Orlando
Harmon, M. E., Harmon, J. M., Ferrell, W. K. and Brooks, D. FL, 141–158.
1996. Modeling carbon stores in Oregon and Washington Ryan, M. G., Hunt, E. R. Jr., McMurtrie, R. E. et al. 1996.
forest products: 1900–1992. Clim. Change 33, 521–550. Comparing models of ecosystem function for temperate
HJA. 2002. H.J. Andrews Long Term Ecological Research conifer forests. I. Model description and validation. In:
site web page. [Link] Global climate change: effects on coniferous forests and
Ito, A. and Oikawa, T. 2000. A model analysis of the grasslands (eds. A. I. Breymeyer, D. O. Hall, J. M. Mellilo
relationship between climate perturbations and carbon and G. I. Agren). John Wiley, New York, NY, 313–361.
budget anomalies in global terrestrial ecosystems: 1970– Schulze, E.-D., Wirth, C. and Heimann, M. 2000. Managing
1997. Clim. Res. 15, 161–183. forests after Kyoto. Science 289, 2058–2059.
Kern, J. S., Turner, D. P. and Dodson, R. F. 1997. Spatial pat- Song, C., Woodcock, C. E., Seto, K. C., Lenney, M. P. and
terns in soil organic carbon pool size in the Northwestern Macomber, S. A. 2001. Classification and change detection

Tellus 55B (2003), 2

668 D. P. TURNER ET AL.

using Landsat TM data: when and how to correct at- Turner, D. P., Dodson, R. and Marks, D. 1996. Comparison
mospheric effects? Remote Sensing Environ. 75, 230– of alternative spatial resolutions in the application of a
244. spatially distributed biogeochemistry model over complex
Sprugel, D. G. 1985. Natural disturbances and ecosystem terrain. Ecol. Model. 90, 53–67.
energetics. In: The ecology of natural disturbances and Turner, D. P., Cohen, W. B., Kennedy, R. E., Fassnacht, K. S.
patch dynamics (eds. S. T. A. Pickett and P. S. White). and Briggs, J. M. 1999. Relationships between leaf area
Academic Press, New York, NY, 335–352. index and TM spectral vegetation indices across three tem-
Thomas, S. C. and Winner, W. E. 2000. Leaf area index of an perate zone sites. Remote Sensing Environ. 70, 52–68.
old-growth Douglas-fir forest estimated from direct struc- Turner, D. P., Cohen, W. B. and Kennedy, R. E. 2000. Al-
tural measurements in the canopy. Can. J. For. Res. 30, ternative spatial resolutions and estimation of carbon flux
1–7. over a managed forest landscape in western Oregon. Land.
Thornton, P. E. 1998. Regional ecosystem simulation: com- Ecol. 15, 441–452.
bining surface- and satellite-based observations to study USDA. 1992. Forest Service resource inventories: an
linkages between terrestrial energy and mass budgets. PhD overview. USDA Forest Service, Forest Inventory, Eco-
Dissertation. University of Montana, 280 pp. nomics and Recreation Research, Washington, D.C. USA.
Thornton, P. E., Running, S. W. and White, M. A. 1997. Vogelmann, J., Sohl, T. and Howard, S. 1998. Regional Char-
Generating surfaces of daily meteorological variables over acterization of Land Cover Using Multiple Sources of
large regions of complex terrain. J. Hydrol. 190, 214– Data. Photogram. Eng. Remote Sensing 64, 45–57.
251. Waddell, K. L. 2002. Sampling coarse woody debris for
Thornton, P. E. and Running, S. W. 1999. An improved al- multiple attributes in extensive resource inventories. Ecol.
gorithm for estimating incident daily solar radiation from Indicators (in press).
measurements of temperature, humidity, and precipitation. Wallin, D. O., Swanson, F. J., Marks, B., Cissel, J. H. and
Agric. For. Meteorol. 93, 211–228. Kertis, J. 1996. Comparison of managed and pre-
Thornton, P. E., Hasenauer, H. and White, M. A. 2000. Simul- settlement landscape dynamics in forests of the Pacific
taneous estimation of daily solar radiation and humidity Northwest, USA. For. Ecol. Manag. 85, 291–309.
from observed temperature and precipitation: an applica- White, J. D. and Running, S. W. 1994. Testing scale de-
tion over complex terrain in Austria. Agric. For. Meteorol. pendent assumptions in regional ecosystem simulations.
104, 255–271. J. Veg. Sci. 5, 687–702.
Thornton, P. E., Law, B. E., Gholz, H. L., Clark, K. L., Falge, White, M. A., Thornton, P. E., Running, S. W. and Nemani,
E., Ellsworth, D. S., Goldstein, A. H., Monson, R. K., R. R. 2000. Parameterization and sensitivity analysis of
Hollinger, D., Falk, M., Chen, J. and Sparks, J. P. 2002. the BIOME-BGC terrestrial ecosystem model: net primary
Modeling and measuring the effects of disturbance his- production controls. Earth Interactions 4, 1–85.
tory and climate on carbon and water budgets in evergreen Woodward, F. I., Lomas, M. R. and Lee, S. E. 2001. Predicting
needleleaf forests. Agric. For. Meteorol., in press. the future productivity and distribution of global terrestrial
Turner, D. P., Koerper, G. J., Harmon, M. E. and Lee, J. J. vegetation. In: Terrestrial global productivity (eds. J. Roy,
1995. A carbon budget for forests of the conterminous B. Saugier and H. A. Mooney), Academic Press, San Diego
United States. Ecol. Appl. 5, 421–436. CA, 521–541.

Tellus 55B (2003), 2