United States
Environmental Protection
Agency
Environmental Sciences Research
Laboratory
Research Triangle Park NC 27711
Research and Development
EPA-600/S3-84-001 Jan. 1984
&EPA Project Summary
Natural Hydrocarbon Emission
Rate Measurements from
Selected Forest Sites
Brian K. Lamb, Hal H. Westberg, Timothy Quarles, and Donald L. Flyckt
Studies of biogenic hydrocarbon emis-
sions were conducted in a hardwood
forest in Pennsylvania during 1979 and
in a coniferous forest in Washington
during 1980. The principal objective of
the studies was to compare a branch
enclosure method with a micrometeor-
ological gradient technique for measur-
ing biogenic hydrocarbon fluxes for
forested areas. A second important
objective for the Pennsylvania work
was to develop a regional natural hydro-
carbon emission inventory for use in the
Northeast Regional Oxidant Study.
Isoprene emission fluxes determined
by the gradient profile procedure in the
deciduous forest agreed reasonably
well with those measured using the
enclosure technique. The isoprene flux
from the gradient profile data was 10%
higher than the enclosure flux at 30°C,
but was approximately three times
greater than the enclosure flux at 20° C.
The differences at the low temperatures
possibly were caused by the lack of
profile data at the lower temperatures.
The combined enclosure and gradient
profile data were correlated with ambi-
ent temperature to the same degree as
the correlation of each data set alone. In
the Washington study, the alpha-pinene
flux as measured by the gradient profile
method ranged from 76 to 1,320
//g/mz-hr whereas, the range deter-
mined using the enclosure method was
9 to 700 /ug/m2-hr. The mean fluxes
from the two methods were within the
estimated limits of uncertainty. Alpha-
pinene fluxes determined with the gradi-
ent profile method increased exponen-
tially with increasing relative humidity.
Emission fluxes calculated from the
branch enclosure samples were not
correlated with ambient relative humid-
ity, but were strongly correlated with
temperature when wet and dry branches
were considered separately.
Total biogenic hydrocarbon emissions
from the state of Pennsylvania were
calculated to be 3400 tons/day during
August. Approximately 75% of these
emissions were from forested lands,
and the remainder were from agricul-
tural crops, primarily corn. The forest
emissions were approximately evenly
divided among isoprene-emitting and
non-isoprene-emitting hardwoods and
softwood trees.
This Project Summary was developed
by £PA's Environmental Sciences Re-
search Laboratory, Research Triangle
Park, NC, to announce key findings of
the research project that is fully docu-
mented in a separate report of the same
title (see Project Report ordering in-
formation at back).
Introduction
The significance of naturally emitted
hydrocarbons in rural atmospheric chem-
istry remains uncertain. This uncertainty
arises in part from a lack of information
concerning the magnitude of sources and
the distribution of hydrocarbon species in
the atmosphere. Current estimates of
natural hydrocarbon emission rates have
been obtained using a vegetation enclo-
sure technique, a micrometeorological
gradient profile method, and an energy
balance/Bowen ratio approach. Confi-
dence in these methods has been limited
by uncertainties about the effects of
enclosing vegetation and in measure-
ments of small vertical gradients of
-------
temperature, wind speed, and hydro-
carbon concentrations above a forest.
These experimental problems have pre-
vented widespread agreement upon actu-
al hydrocarbon emission rates.
In this report, field measurements of
biogenic hydrocarbon emission rates are
reported for the predominant species of
trees and some agricultural crops in the
northeastern U.S. and for a coniferous
forest in the Pacific Northwest. These
data were obtained using a vegetation
enclosure method and, for the forests, a
micrometeorological gradient profile tech-
nique. The major objectives of the investi-
gation were (1) to obtain natural hydro-
carbon emission rates from vegetation in
the northeastern U.S. as a basis for
modeling natural hydrocarbon fluxes in
the Northeast Regional Oxidant Study
(NEROS), (2) to compare emission fluxes
determined by a vegetation enclosure
technique and a micrometeorological
gradient profile method, and (3) to com-
pare the chemical composition of essen-
tial plant oils with the composition of gas
emissions from selected vegetation.
Procedure
In the forest studies, the flux of a
particular hydrocarbon was calculated
from surface layer theory, based upon
measurements of vertical wind speed,
temperature, and hydrocarbon concentra-
tion profiles collected along a tower rising
above the forest canopy. For the hard-
wood forest, 30-min average air samples
were collected at six levels on the tower
using Teflon sampling pumps and 6-L
Tedlar bags. For the coniferous forest, 30-
min average air samples were collected
at five levels on a tower in stainless steel
tubes packed with Tenax-GC adsorbent.
Vegetation enclosure samples in each of
the study areas were obtained using a
branch enclosure method. This involved
enclosing a branch in a 100-LTedlar bag,
collecting a background sample from the
bag, filling the bag with hydrocarbon-free
air, and, after a measured length of time,
collecting a second sample of the bag air..
When used to develop an area emission
inventory, the individual branch emission
rates (yug/g-hr) were multiplied by an
appropriate biomass factor (g/m2) to give
the area emission flux (//g/m2-hr). All
hydrocarbon samples were analyzed with
Perkin-Elmer and Hewlett-Packard gas
chromatographs equipped with subambi-
ent temperature programming capabil-
ities.
The hardwood forest was located in
gently sloping terrain approximately 10
km northeast of York, Pennsylvania. A
vegetation inventory of the forest indicat-
ed that the woodland was a second
growth oak and chestnut forest. The
biomass factor for isoprene-emitting spe-
cies was determined to equal 379 g/m2
from the vegetation inventory and an
empirical relationship between the diam-
eter of the tree at breast height (DBH) and
the biomass. The biomass factor for non-
isoprene emitting species was 206 g/m2.
The average canopy height was 20 m.
The coniferous forest was located 36
km southeast of Seattle, Washington, in
the A. E. Thompson Research Forest
(University of Washington). The forest
consisted almost entirely of Douglas fir,
with red alder, western hemlock, and
western red cedar present in small
amounts. The age of the stand was 52
years and the average canopy height was
31 m. The biomass factor was estimated
to equal 830 g/m2 from a relationship
between biomass and the age of the
stand developed specifically for the Thomp-
son forest.
Results
Northeastern Biogenic
Hydrocarbon Emissions
Results from the Pennsylvania study
indicated that volatile hydrocarbon emis-
sions can be divided into four categories
based on similarities m composition and
emission source. The first two categories
are represented by hardwood trees that
(1) emit isoprene and (2) do not emit
isoprene. The two remaining groups are
(3) monoterpene emitting softwood trees
and (4) agricultural crops.
Included in the group of isoprene
emitting hardwood trees measured were
oak, black locust, and sycamore. Isoprene,
which is only emitted during daylight
hours, accounted for 78% of the total
volatile hydrocarbon emissions from this
group. As shown in Figure 1, the emission
rates were directly related to temperature
in an exponential manner, with 90% of
the variation in emission rates associated
with temperature variations. The regres-
sion relationship between temperature
and emission rate predicts a total hydro-
carbon emission rate of 6.1 /ug/g-hr at
25°C.
The non-isoprene-emitting hardwood
trees measured included black gum,
sassafras, tulip tree, red maple, dogwood,
red hickory, black cherry, beech, silver
maple, and birch. The hydrocarbon emis-
sions from these species were also
related to temperature in a exponential
manner, with 91% of the variation in
emission rates associated with temper- .
ature fluctuations. The regression rela-
tionship between temperature and emis-
sion rates shown in Figure 1 predicts an
emission rate of 3.4 /jg/g-hr at 25°C.
Major identified components of the vola-
tile emissions were the monoterpenes a-
pinene, sabinene, /3-pinene, myrcene,
camphene, /3-phellandrene, A3-carene,
and linalool.
The monoterpene-emitting softwood
species measured included eastern white
pine, Virginia pine, eastern hemlock, and
pitch pine. The volatile emissions from
this group consisted of a-pinene, sabi-
nene, camphene, /3-pinene, myrcene, fl-
phellandrene, limonene, and A3-carene.
These compounds comprised 50-95% of
the volatile emissions. Correlation be-
tween emission rate and temperature
was not observed for this group; thus, the
release of the monoterpenes must be
controlled by some temperature-inde-
pendent process.
The agricultural crop category meas-
ured included tobacco, corn, clover, alfal-
fa, and pasture. This group is poorly
defined at present and may require
further divisions. The members emitted
compounds eluting near the monoter-
penes and very small amounts of iso-
prene. There appeared to be no relation-
ship between emisison rate and temper-
ature.
The emission rates established in this
study can be combined with vegetation
surveys and biomass relationships to
provide estimates of the magnitude of
biogenic emissions. Total biogenic hydro-
carbon emissions from the state of Penn-
sylvania were calculated to be 3400
tons/day during August. Approximately
74% of these emissions were from forest-
ed lands. The balance was estimated to
result primarily from corn (25%) and other
agricultural crops and pasture. The forest
emissions were approximately equally
divided among isoprene-emitting hard-
woods (890 tons/day), non-isoprene-emit-
ting hardwoods (710 tons/day), and
conifer species (910 tons/day).
In order to gain insight into the pro-
cesses that control hydrocarbon emis-
sions from vegetation, extracts of essen-
tial oils from leaf and branch samples
were obtained at the Pennsylvania study
site. For the monoterpene-emitting soft-
woods, the composition of the oils in the
needles and branches match the range of
compounds identified in the emission
samples quite closely. This agreement
suggests that volatilization of oils is the
major source of emissions from these *
species. For isoprene-emitting hard-
-------
100.0
60.0
20.0
10.0
6.0
o
^ 2.0
1.0 -
0.6 ~
0.4
10
15
30
35
Figure 1.
20 25
Temperature °C
Total nonmethane hydrocarbon emission rate as a function of temperature for
isoprene-emittinghardwoods (PA A 1-—)andnon-isoprene-emitting hardwoods
(PA • 2—/
woods, the extraction results showed
little similarity to the emission samples
and no isoprene was measured in the
extraction samples.
Comparison of Vegetation
Enclosure and
Micrometeorological Gradient
Techniques
In Pennsylvania, isoprene fluxes meas-
ured by both techniques were observed to
vary exponentially with temperature. The
isoprene flux estimated from the enclo-
sure data at 20°C ambient was 890 /j-
g/m2-hr and the flux calculated from the
gradient profile data was 2510 Ai-g/m2-
hr. At 30°C ambient, the difference
between estimates obtained using the
two methods was approximately 10%
(7300/^-g/m2-hrfrom the enclosure data
and 8000 /u-g/m2-hr from the gradient
profile method). The differences at 30°C
were within the estimated experimental
uncertainty of the two methods of meas-
urement, but the results at the lower
temperature were not within the error
limits. However, very few gradient profile
experiments were conducted at low tem-
peratures. Furthermore, as shown in
Figure 2, the combined data from the
enclosure and gradient profile measure-
ments were correlated with ambient
temperature to the same degree as either
data set alone. Thus, the data obtained in
Pennsylvania indicate that the two inde-
pendent methodsfor measuring isoprene
flux yield results that are in reasonable
agreement.
In Washington, alpha-pmene flux meas-
ured via the gradient profile technique
increased exponentially with increasing
relative humidity at relative humidities
greater than approximately 60%. The
calculated flux was also positively corre-
lated with the ratio of relative humidity to
wind speed as shown in Figure 3. Alpha-
pinene fluxes were only weakly corre-
lated with ambient temperature. The
mean flux of alpha-pinene measured in
seven gradient profile experiments was
440 /jg/m2-hr (median = 230 A/g/m2-hr),
with a range of 76 to 1320 Aig/m2-hr.
The mean alpha-pinene flux estimated
from 13 enclosure samples was 150
mg/m2-hr (median = 46 //g/m2-hr) with a
range of 9 to 700 //g/m2-hr. No relation
between ambient relative humidity and
alpha-pinene flux was observed in the
enclosure data. However, when the sam-
ples were grouped according to wet and
dry branches, a distinct exponential rela-
tionship with increasing ambient temper-
ature was evident, as shown in Figure 4.
No single environmental parameter
was correlated with the fluxes determined
by both methods. As a result, it was not
possible to compare directly fluxes pre-
dicted for a common reference point, as
was done for isoprene as a function of
temperature. The mean flux of alpha-
pinene for the gradient profile experi-
ments and the mean flux predicted by the
enclosure samples were within experi-
mental uncertainties estimated for the
two methods. The wide range of fluxes
observed was not a result of random
experimental errors. Rather, changes in
environmental conditions such as temper-
ature and humidity were observed to
affect the emission rates dramatically. In
view of the strong dependence of flux
upon humidity, the lowfluxesdetermined
with the enclosure method suggested
that humidity inside the bag was generally
lower than ambient.
The error limits estimated for the two
techniques were based on standard error
analysis procedures. Data from typical
experiments were combined with esti-
mated or measured uncertainties in the
independent variables to calculate the
overall error and degree of reproducibility
in each flux method. In Pennsylvania, the
error in the gradient profile method was
estimated to equal ± 35%; in Washington,
the error was approximately ± 55%. The
major sources of error were uncertainties
in the zero plane displacement height and
the hydrocarbon concentration gradient.
The gradient profile method is estimated
to be reproducible to within less than ±
17% for test conducted at a specific site.
The overall error estimated for a typical
enclosure sample was ± 40%, and the
technique was estimated to be reproduc-
ible to within ± 20% for samples collected
in a particular forest.
Summary and Conclusions
For a typical northeastern deciduous
forest, isoprene emissions accounted for
approximately 78% of the total non-
-------
10.000
1.000
0)
100
Figure 2.
Gradient
r = 0.65
Gradient +
Enclosure
r = 0.63
Enclosure (T*mKJ
r = 0.60
Enclosure (T
r = 0.85
10
20
Temperature °C
30
Isoprene flux as a function of ambient temperature measured at 6m above the
canopy by the gradient profile method (line 1). the gradient profile and enclosure
method (line 2). and only the enclosure method (line 3) for Pennsylvania. Line 4 is
from the Pennsylvania enclosure data as a function of enclosure temperature.
ed here show that the enclosure method
yields emission fluxes quite similar to
those obtained with a totally independent
technique.
The enclosure system is portable, easily
operated by one person and not limited to
idealized sites. However, care is required
to use the enclosure method. Enclosure
conditions must be closely monitored in
order to relate the data to ambient
conditions. Biomass factors must be
developed from site-specific inventories
and representative biomass relationships.
In comparison the gradient profile method
involves considerable effort and instru-
mentation, the site requirements are
restrictive to the point of being impracti-
cal, and the uncertainties in specifying
the zero plane displacement height and
property gradients reduce the applicability
of the technique to the kind of compar-
ative studies described in this report.
methane hydrocarbon emission rate.
Isoprene concentrations and isoprene emis-
sion fluxes increased exponentially with
ambient temperature. At 30°C, the
emission flux of isoprene was approxi-
mately 8000 jug/m2-hr. Results from the
vegetation enclosure method and the
micrometeorological technique generally
agreed well.
The emission rates of alpha-pinene
measured by the gradient profile method
in a northwestern Douglas fir forest were
closely correlated with relative humidity.
Emission rates measured by the enclo-
sure method were not correlated with
ambient relative humidity. However, emis-
sion rates from wet branches were an
order of magnitude higher than those for
dry branches. Alpha-pinene emission
rates from both wet and dry branches
increased exponentially with increasing
ambient temperature. During the autumn
sampling period, the emission flux of
alpha-pinene was less than 1800/ug/m2-
hr and ambient concentrations above the
canopy were less than 1 jug/m3. In this
case, the range of fluxes observed via the
enclosure and gradient methods over-
lapped, but the median values did not
agree within the estimated uncertainty of
the methods.
The comparisons between the two
measurement techniques in Pennsylvania
and in Washington have better defined
the reliability of emission flux estimates.
Although the usefulness of data collected
with the enclosure method has been
disparaged previously, the results report-
-------
1,000
= 0.77
•c
\
<0
i
700
40
20
Figure 3.
40
60
RH/uz (%/m/s)
80
100
120
Alpha-pinene flux as a function of relative humidity/wind speed measured by the
gradient profile method.
-------
1000
500
200
100
(0
I. 50
20
10
Dry Branches
= 092
10 12 14
Temperature (°Cj
16
18
Figure 4.
Alpha-pinene flux as a function of ambient temperature measured by the enclosure
method.
------- |