92
JOURNAL OF APPLIED METEOROLOGY
VOLUME 39
Wintertime Ozone Fluxes and Profiles above a Subalpine Spruce–Fir Forest
KARL ZELLER
USDA Forest Service, Fort Collins, Colorado
(Manuscript received 1 August 1998, in final form 23 March 1999)
ABSTRACT
High rural concentrations of ozone (O 3 ) are thought to be stratospheric in origin, advected from upwind urban
sources, or photochemically generated locally by natural trace gas emissions. Ozone is known to be transported
vertically downward from the above-canopy atmospheric surface layer and destroyed within stomata or on other
biological and mineral surfaces. However, here the authors report midwinter eddy correlation measurements of
upward vertical O 3 flux of 0.2 mg m22 s21 (5.6 kg km22 day21 ) above a subalpine canopy of Picea engelmannii
and Abies lasiocarpa in the Snowy Range Mountains of Wyoming. Simultaneous below-canopy upward fluxes
reached 0.1 mg m22 s21 . These results corroborate similar late winter (presnowmelt) upward O 3 fluxes of 0.5
mg m22 s21 (19 kg km22 day21 ) taken at the same site in 1992. Profile results show sustained ‘‘countergradient’’
fluxes below the canopy and sustained ‘‘with gradient’’ fluxes above the canopy. Ozone concentrations that
decrease for several hours to several days correspond to simultaneously increasing positive (upward) O 3 fluxes
and vice versa. These phenomena, in addition to above- and below-canopy reversed gradient patterns, suggest
that O 3 may be stored temporarily in either the snow base or the tree stand itself.
1. Introduction
Forest ecosystems play a role in the uptake and destruction of tropospheric O 3 . This role and the tropospheric O 3 budget in remote forested ecosystems are
uncertain (Chameides and Lodge 1992). Ozone deposition, which is rapid during the growing season and
slower during winter months (Wesely 1983), is retarded
further by surface snow cover (Stocker et al. 1995; Zeller and Hehn 1995, 1996). Zeller and Hehn (1996)
showed an apparent but unexpected effect of snow cover
on O 3 fluxes above a subalpine spruce–fir forest. In the
presence of below-canopy surface snow, O 3 fluxes reverse direction from negative (downward) to positive
(upward) (Zeller and Hehn 1994, 1996). Positive O 3
fluxes attributed to vertical entrainment of clean air from
aloft have been measured by aircraft (Lenschow et al.
1982) and modeled (Gao and Wesely 1994) in the upperatmospheric boundary layer. Measurements of negative
vertical O 3 profiles above forested canopies have led to
‘‘countergradient’’ O 3 flux claims, with the assumption
that O 3 only deposits toward the earth’s surface (Fontan
et al. 1992; Enders 1992; Denmead and Bradley 1985;
Kelly and McTaggart-Cowen 1968). Galbally and Allison (1972) reported wintertime upward ozone fluxes
(1.6 mg m22 s21 ) over fresh snow based on gradient
Corresponding author address: Karl Zeller, USDA Forest Service,
240 W. Prospect, Ft. Collins, CO 80526-2098.
E-mail: kzeller@lamar.colostate.edu
measurements at 1807 m in southeast Australia. The
observation of positive O 3 flux above forests is also not
unique. Table 1 gives a summary of O 3 flux measurements taken above forest canopies. Two grass sites are
included in Table 1 because those data were collected
above snow near the site studied here. In Table 1 both
positive flux or negative deposition velocity V d indicate
upward O 3 flux.
a. Ozone profiles and fluxes
An inverse trend (with time) between above-canopy,
ambient O 3 concentrations and O 3 fluxes was observed
by Zeller and Hehn (1996) during the presence of snow
cover. These observations, and laboratory-determined
ozone-to-ice adsorption sticking coefficients, led to a
hypothesis that the positive O 3 fluxes discussed above
were caused by the snow base acting as a leaky capacitor
(Zeller and Hehn 1996). The ‘‘capacitor’’ snow base
might store and release O 3 modulated by turbulent air
interactions with the porous snow surface (Massman et
al. 1997). Zeller and Hehn (1996) calculated that a 1-mdeep snow base could sustain the apparent O 3 emission
of 0.5 mg m22 s21 for 0.5–45 days. Ozone profile and
flux measurements were made below and above the canopy to explore further the positive O 3 phenomena. This
paper presents the results of O 3 profile (vertical gradient)
and O 3 flux measurements taken during January 1993
at the same Snowy Range Mountains site in Wyoming.
JANUARY 2000
93
ZELLER
TABLE 1. Summary of representative forest ozone flux measurements.
Author(s)
Technique*
Forest type
Flux
[mg m22 s21 ; up (1);
down (2)]
Mean
ozone
(ppb)
Vd
[cm s21 ; up (2);
down (1)]
Enders (1992)
EC
Spruce
—
5–50
21.0 to 1.8
Enders et al. (1989)
EC
Spruce
—
10–40
28.9 to 1.8
Fontan et al. (1992)
PR
Pine
20.11 max
18–38
0.04 to 0.8
Matt and Womak (1989)
EC
Spruce–fir
5–50
0.1 to 0.8
Lopez et al. (1993)
PR
Pine
—
20
0.2 to 0.5
Padro (1993)
EC
Deciduous
—
25–100
0.3 to 1.0
0.1 to 0.3
Rondon et al. (1993)
CH
Spruce
5–70
0 to 0.5
Wesely (1983)
EC
Deciduous
Zeller and Hehn (1994)
Zeller and Hehn (1995)
EC
EC
Spruce–fir
Subalpine meadow
Stocker et al. (1995)
EC
Dry grassland
Zeller and Hehn (1996)
EC
Spruce–fir
0 to 20.4
0 to 20.2
—
—
0.01 to 0.37
1.0 to 20.6
20.01 to 20.35
23–70
29–48
21.5 to 0.4
0.01 to 0.07
0.09 to 20.35
15–59
—
0.6 to 20.6
33–70
—
Time/location
Jun 1989
Germany
Jun 1987
Germany
Sep 1985
France
Sep 1987
Maine
Sep 1984
France
Aug 1988
Apr 1990
Canada
Jun 1990
Sweden
Jan 1981
N. Carolina
Jan 1992
May 1994
Wyoming
Feb 1989
Colorado
Apr–May 1992
Wyoming
* Eddy correlation: EC, gradient profile: PR, chamber: CH; — is no report.
b. Ozone description
Snowy Range hourly O 3 concentrations average 45–
60 ppb all year (Wooldridge et al. 1997) and are typical
of clean, high-altitude rural sites (Wunderli and Gehrig
1990). The above-canopy diurnal O 3 concentration at
this location does not exhibit the large day/night maximum/minimum pattern typical of urban, photochemically dominated air masses at any time during the year
(Wooldridge et al. 1997). The wintertime, 3-m (near
surface) below-canopy diurnal O 3 concentration also
does not exhibit a day/night maximum/minimum pattern
(Wooldridge et al. 1997). However, the 3-m, belowcanopy summertime O 3 concentrations have a predominant day/night pattern in which minimum values can
drop to 10 ppb at night because of deposition (Wooldridge et al. 1997).
2. Methods
a. Site
Data were collected at the Brooklyn Lake tower site
situated in an Engelmann spruce–subalpine fir forest
opening approximately 30 m in diameter. The site is
within the USDA Forest Service’s Glacier Lakes Ecosystem Experiment Site (GLEES) area in the Snowy
Range of the Medicine Bow National Forest, Wyoming.
The GLEES complex is described by Musselman
(1994). The 29-m Brooklyn Lake tower (base elevation
3186 m, 418229N, 106814.59W) is approximately 3 km
southeast of the Snowy Range ridge (3460-m average
elevation). The average forest stand height H is 17 m
with representative displacement height d of 11.7 m and
roughness length z 0 of 1.7 m. The terrain within 1 km
of the tower slopes 12.5% from west to east and 29.7%
from north to south. This site is relatively complex for
eddy flux experiments; however, concurrent measurements of momentum and sensible heat fluxes provided
reasonable values. Fitzjarrald and Moore (1992) found
that scalar flux measurements in nonhomogeneous regions were robust and representative of the upwind footprint. Upwind terrain in the predominant wind direction,
southwest–northwest, is forested with a flat 12.2%
slope for at least 1 km.
The tower itself is about 12-m distance downwind of
the upwind edge. The GLEES forest is not dense and
Brooklyn site clearing is 1.8H. A dense forest clearing,
diameter 2.5H, might cause a negative vertical wind
above 0.5H, a reverse-flow eddy below 0.5H, and an
overall wind measurement error of 10% (Miller et al.
1991). The first two affects were not observed: vertical
winds were positive; at 9 m (0.53H), wind directions
were always identical to those at 23 and 29 m and wind
speeds averaged a third lighter. Based on these measurements, on previous transseasonal O 3 flux magnitude
and direction results (Zeller and Hehn 1996), and on
the energy balance results that included sensible heat
determined by eddy correlation, the O 3 fluxes reported
here are taken as representative of those within and
above the canopy.
94
JOURNAL OF APPLIED METEOROLOGY
b. Flux and profile measurements
Meteorological, O 3 concentration, profile, and flux
data presented here were collected 15–26 January 1993
with a 1.3-m snow cover. The eddy correlation system
(Zeller et al. 1993; Massman et al. 1990; Zeller and
Hehn 1996) was employed to measure O 3 , sensible heat,
and momentum fluxes; O 3 concentrations; temperature;
and wind speed at 9 m (8 m below canopy top) and 23
m (6 m above canopy) on the Brooklyn tower. Standard
meteorological sensors are permanently mounted at 10
and 29 m for routine GLEES measurements (Musselman
1994). The essential sensors for O 3 flux measurements
are the Gill UVW anemometer and the chemiluminesence ambient air monitor (CAAM) (Ray et al. 1986).
The CAAMs were continuously calibrated using
TECO49 commercial UV adsorption instruments. Ambient air was sampled from heights of 9, 23, and 29 m
through 1.6-cm-diameter Teflon tubes using high-volume (155 lpm) pumps. The intake system lag time t l
was typically 2.4 s. Gill UVW anemometers were used
in place of sonic anemometers for their greater durability in harsh alpine weather. The 17-Hz O 3 eddy deviations, c9 5 c 2 c (c: 200-s recursive filter average
concentration); temperature T9; and turbulent wind components u9, y 9, and w9 were multiplied and then averaged
over half-hour sampling periods (McMillen 1988) to
obtain the vertical flux F c [mg m22 s21 , Eq. (1)] after a
vector coordinate rotation for the w 5 y 5 0 streamline
FC 5 w9(t 2 t l )c9(t).
(1)
Here, negative F c indicates downward flux. Sensible
heat and momentum flux were obtained likewise. Coordinate rotations do not affect scalar flux sign and have
a very small effect on measured flux magnitudes. The
optimal choice for the recursive filter (mean removal
time) period is a function of sampling height and atmospheric stability. The 200-s mean removal time used
in this study to determine the running means (c, w, etc.)
is shorter than some might recommend for forest eddy
flux applications. It was chosen to maximize capture of
turbulent eddies while minimizing nonstationarity effects. As an example, tests using a longer mean removal
time at the same site and elevation during the summer
season indicate that O 3 flux magnitudes do not change
much for filters greater than 150 s but can be as high
as 5% larger using 600 s (versus 200 s).
Average wind streamline tilt (6) at the Brooklyn tower tends to follow the terrain and is wind direction sensitive (Zeller and Hehn 1996). Fluxes with associated
streamlines within 698 (9 m) and 658 (23 m) of horizontal account for 98% of the data presented here. Wind
directions during this study were consistently southwest
to northwest, hence the streamline tilt was almost always
positive due to the 12.5% slope east of the tower. Gill
UVW anemometer data were corrected in real time for
the inherent cosine response problem (Massman and
Zeller 1988). The fast-response temperature sensor was
VOLUME 39
routinely damaged by harsh weather, so sensible (and
latent) heat fluxes were calculated using the gradient
method (Table 2) to provide for the 23-m directional
and intensity comparisons with the measured sensible
heat and O 3 fluxes. Latent heat fluxes were also calculated for the energy balance. The energy fluxes were
calculated as rC p K(]T/]z) [and rLK(]q/]z)] using measured eddy diffusivities (K) for momentum assuming
similarity with heat (and vapor) diffusivities. Given the
known inaccuracies of using the gradient method for
forest applications (Raupach 1989), the calculated 23-m
latent heat values in Table 2 are larger than expected
for a wintertime scenario. These high values are a result
of using the measured momentum diffusivity. Figure 1
demonstrates the daytime energy balance result without
accounting for forest heat storage. The energy balance
was accomplished using measured sensible heat and soil
heat flux (,12 W m22 ), modeled latent heat, and estimated net radiation. Net radiation was calculated based
on a regression analysis of simultaneous R n and total
solar radiation R T December 1997 measurements: [R n
5 a(R T 2 11) 2 1 0.58, where a 5 20.03 from 0900
to 1100 MST, and a 5 20.01 from 1130 to 1500 MST].
Businger’s (1986) list of eddy correlation measurement concerns were used for data evaluation and editing.
Ozone flux corrections [Eq. (11), Luening and Judd
1996] for vapor effects were on the order of 10.007
mg m22 s21 (about 10% of the daytime measurement).
The O 3 flux data were not corrected for vapor effects.
Data below the scale height (h t , |w9c̄9/(]c/]t)|) of 9
and 23 m were culled. A significant percentage of the
23-m O 3 flux (w9c̄9) data were culled as a result of this
procedure. Nonetheless, the culled data mostly followed
the daily positive flux pattern. A digital Butterworth
filter was applied in real time to account for aliasing.
Instrument response and instrument separation corrections would typically increase flux magnitudes by 20%–
80% (Zeller et al. 1993). The latter do not affect the
main result (O 3 flux direction), hence the flux data are
presented here without those corrections as they would
typically increase the magnitude of the upward O 3 fluxes
slightly more than the downward. Order of magnitude
and flux direction are not affected by the lack of these
corrections and make the reported upward flux values
conservative.
Neutral stability micrometeorological statistics for
wind, temperature, and O 3 were fairly consistent during
this study and equivalent to results from other field studies. For example s w /u* at 23 m was consistently 1.5 6
0.5 compared to 1.3 for flat terrain. Dimensionless wind
shear, f m 5 kz/u* (]u/]z), was consistently 1.5 6 0.3
where u* 5 (2w9u9)1/2 was calculated from the measured momentum flux. There are no aberrant values in
Table 2 to indicate a sampling problem with either the
sensors, the sampling system, or the tower configuration.
Ozone profile (dc/dz) measurements were made from
9 to 23 m with two O 3 analyzers using the half-hour
6
10
4
2
6
5
12
26
4
8
—
280
280
269
230
295
305
283
265
305
295
386
0.55/0.28
0.45/0.24
0.18/0.05
0.28/0.03
0.05/0.08
0.29/0.26
0.46/0.24
0.35/0.16
0.00/0.17
0.37/0.23
—/0.25
23.1
21.5
20.4
20.3
21.4
24.1
22.7
21.9
20.2
21.8
—
* Estimated by gradient method, g; regression estimate, r; variable, v; midday average, †; —, missing data or no measurement taken.
9.5
7.5
3.3
3.0
3.7
11.2
8.0
8.9
9.1
10.4
10.8
400/230r
420/240r
490/282r
430/245r
285/163r
485/278r
190/113r
225/132r
435/252r
515/293r
475/272r
137 g
152 g
217 g
148 g
125 g
—
58 g
—
125 g
205 g
141 g
290 g
160 g
450 g
350 g
220 g
275 g
120 g
90 g
75 g
150 g
20 g
—/90
—/75
—/120
—/65
95/70
—/160
—/—
—/—
—/—
—/—
—/—
—/87
95/90
95/70
95/73
94/46
94/73
94/89
92/62
89/36
86/30
92/70
—/29
25/28
24/27
23/27
23/210
24/211
24/28
24/212
212/217
211/215
22/211
—/0.07
—/0.05
0.09/0.03
—/—
—/0.08
0.19/0.10
0.22/0.06
0.13/0.06
0.17/0.05
—/0.07
0.08/0.09
—/—
4.1/1.5
—/—
—/—
3.0/1.8
5.6/3.5
3.0/—
1.7/1.6
—/1.2
—/2.6
—/3.5
45/—
41/47
43/40
43/38
48/40
46/41
47/40
47/39
36/36
46/32
43/38
15
16
17
18
19
20
21
22
23
24
25
Yearday
1.8
1.2
0.6
0.5
1.2
2.0
1.6
1.4
0.4
1.3
—
w
(m s21 )
23/10 m
†
u
u9w9
u*
(m s21 ) (m 2 s22 ) (m s21 )
23 m
23 m
23 m
†
†
†
Radiation
max
(W m22 )
R t /R n
Heat flux
max
Heat flux
½h
est.
Latent
(W m22) (W m22 ) heat flux
23/10 m
23 m (W m22 )
RH max/
min
(%)
29 m
Temperature
max/min
(8C)
29 m
O 3 flux max
O 3 max/ O 3 flux total
½h
min
(mg m-2 d-1) (mg m22 s21)
(ppb)
23/10 m
23/10 m
TABLE 2. Daily ozone, ozone flux, and meteorological values.*
95
—
128
—
—
—
—
—
—
—
—
123e
ZELLER
u
Precip(deg) itation Snow
23 m (mm) cover
†
daily
(cm)
JANUARY 2000
FIG. 1. Energy balance at Brooklyn tower. Measured sensible heat
(H) and modeled latent heat (LE) plotted against estimated net radiation (R n ) and measured soil heat flux (G ) for the period 15–20
Jan 1993.
switching intake system described by Zeller et al.
(1993). Measurements for 23–29 m employed a third,
frequently intercalibrated O 3 analyzer on a separate intake system. Accuracy for dc/dz values for 9–23 m is
60.005 ppb m21 and for 23–29 m is 60.15 ppb m21 .
Temperature profiles between 9 and 23 m were made
with a copper–constantan thermocouple, between 10
and 29 m with standard temperature sensors, and between 23 and 29 m derived from the difference. Hence,
dT/dz values for 9–23 m (60.038C m21 ) are more accurate than those for 23–29 m (60.18C m21 ).
3. Results
The eddy correlation measurements show consistent
upward daytime O 3 fluxes at 9 and 23 m during the
January 1993 snow-covered measurement period. The
23-m flux was about one-third of the rate measured in
April 1992 (Zeller and Hehn 1996). The upward flux at
9 m is again about half of that simultaneously measured
at 23 m. These results are in contrast to the expected
downward fluxes measured at the same site during the
growing season in the absence of snow cover (Zeller
and Hehn 1996). Peak upward 9- and 23-m O 3 fluxes
(i.e., half-hour values) reached 0.1 and 0.2 mg m22 s21
(integrated over 24 hr: 3.5 and 5.6 kg km22 day21 ) on
20–21 January 1993. Sometimes nighttime O 3 flux did
not cease at either level as observed in most O 3 flux
experiments. Nighttime positive fluxes above snow cover were previously observed at this site in 1992 (Zeller
and Hehn 1996). Figures 2 and 3 show O 3 concentra-
96
JOURNAL OF APPLIED METEOROLOGY
FIG. 2. Half-hour average O 3 concentration at 23 m (dashed) and at NDDN site 169 (dots) in
parts per billion (ppb), O 3 flux (open circle–solid lines) and vertically integrated time rate of O 3
change (1) in mg m22 s21 at 23 m for the period 15–26 Jan 1993.
FIG. 3. Half-hour average O 3 concentration (dashed) in parts per billion (ppb), O 3 flux (open
circle–solid lines) and vertically integrated time rate of O 3 change (1) in mg m22 s21 at 9 m for
the period 15–26 Jan 1993.
VOLUME 39
JANUARY 2000
ZELLER
97
FIG. 4. Wind speed (m s21 ) and Reynolds stress (m 2 s22 ) at 23
(solid lines) and 9 m (dashed) for the period 15–26 Jan 1993.
FIG. 5. Ozone gradient (ppb m21 ) for 9–23 m (solid line) and 23–
29 m (dashed) for the period 15–26 Jan 1993.
tions, O 3 fluxes, and vertically integrated time rate of
O 3 change for the 11-day measurement period listed in
Table 2.
Table 2 summarizes the meteorological and O 3 data.
Midday meteorological values are representative averages between 1000 and 1500 MST, the diurnal hours of
greatest O 3 flux activity. The maximum half-hour values
for O 3 concentration, O 3 flux, and heat fluxes are the
maximum values measured (or estimated) for the 24-h
period from midnight to midnight. Total daily O 3 flux
was obtained by integrating half-hour values commencing midnight each day. The maximum daily flux
value of 15.6 mg m22 on yearday (JD) 20 compares in
magnitude to 223 mg m22 on JD 190 in 1992 during
a growing season. This JD 20 maximum flux also compares to the maximum upward flux, 119 mg m22 , on
JD 113 in 1992 above the 1.1-m snow cover (Zeller and
Hehn 1996). In addition to the presence of snow, Zeller
and Hehn (1996) noted that ambient temperature appeared to affect O 3 flux direction. The maximum daytime temperatures reported in Table 2 never exceeded
08C as they did during the 1992 measurements. Ozone
flux direction is neither affected by vertical wind speed
or direction nor horizontal wind direction; both would
indicate a terrain-induced bias. During the study, two
weather systems associated with wind shifts and pressure drops passed through the site area at JD 18.6 and
JD 22.5. Wind directions veered from northwest to
southwest on JDs 15–18.6, abruptly backed to the northwest on JD 18.6, then again veered from southwest to
northwest on JDs 18.6–22.5 when they backed sharply
to northwest again. Wind directions remained northwest
from JD 22.5 through JD 25.
The stronger multihour surges of upward O 3 flux are
usually associated with simultaneous drops of several
parts per billion in ambient O 3 concentrations (Zeller
and Hehn 1996). These phenomena are easy to see in
Fig. 3 on JDs 20, 24, and 25. Wind speeds and Reynolds
stresses for 9 and 23 m shown in Fig. 4 were also strong
during these periods. The 23-m vertical wind sensor
malfunctioned on JD 25, so no flux data are reported at
23 m for that day. Note the period JDs 23.5–24.5: wind
speeds remained high during the night while O 3 concentrations rose 11 ppb, and the O 3 fluxes remained
positive but slowly dropped in magnitude.
The above-canopy daytime O 3 gradient from 23 to
29 m was consistently negative (fluxes were with the
gradient), as predicted by Zeller and Hehn (1996). This
result, as seen in Fig. 5 (average 20.3 ppb m21 ), is not
typical of surface-layer O 3 gradients and would have
been unexpected prior to the observed positive fluxes
at the same site. The through-canopy gradient (8 m below to 6 m above the 17-m canopy) was consistently
positive (average 0.05 ppb m21 ) and counter to the gradient. This magnitude is about four times smaller when
compared with surface-layer values above an active
growing season canopy (Zeller et al. 1993). Note the
gradients peak sharply in opposite directions at midnight
on JD 19. Ozone concentrations at 23 m were about 7
ppb higher (see Figs. 2 and 3) at this time. Wind speeds
and vertical mixing (Fig. 4) were very low at the same
time. In contrast to the midnight JD 19 observations,
the midday, JD 22, O 3 gradients above and below canopy were close to 0.0, while wind speeds and vertical
mixing were strong.
Surface layer temperature gradients are typically adiabatic (negative) during daylight when turbulent mixing
is occurring and stable (positive) at night when abovesurface air decouples from air adjacent to the surface,
causing reduced turbulence and little mixing. Ozone
98
JOURNAL OF APPLIED METEOROLOGY
FIG. 6. Temperature gradient (8C m21 ) for 9–23 m (solid line) and
23–29 m (dashed) for the period 15–26 Jan 1993.
concentrations are also known to be somewhat temperature sensitive. Note that, in Fig. 6, the above- and below-canopy temperature gradients followed the O 3 gradient on JD 19 (during minimum mixing) but did not
on the night of JD 24 when wind speeds were higher.
4. Discussion
This study extends the observed phenomenon of upward O 3 flux above a subalpine spruce–fir forest during
early (October) and late (April–May) winter to midwinter (January). The new observations of consistent
negative O 3 gradients above the canopy and consistent
countergradient flux below the canopy suggest that the
forest–snow system may be producing or somehow
modulating large amounts of O 3 .
In addition to surface or canopy emissions of O 3 ,
upward fluxes in complex terrain suggest the possibility
of horizontal and/or vertical advection of O 3 . The potential contributions of local O 3 production or destruction (R) and advection below 29 m can be roughly estimated from the measured data by vertically integrating
the equation for O 3 conservation from the surface to
measurement height. Using the average wind streamline
(w 5 y 5 0), assuming w9c90 5 0, and neglecting molecular diffusion gives Eq. 2:
E
0
29 m
]c
dz 1
]t
52
E
E
29 m
u
0
2
0
E
29 m
0
29 m
]w9c9
dz
]z
]c
dz 2
]x
E
E
]y 9c9
dz 1
]y
29 m
0
0
]u9c9
dz
]x
29 m
R d z.
(2)
VOLUME 39
As seen in Figs. 2–3, the vertical flux [second term,
Eq. (2)] is at least an order of magnitude greater than
the local time rate of change (first term). Therefore, the
local time rate of change does not significantly contribute to the vertical flux. If horizontal advection and turbulent transport (third–fifth terms) were insignificant,
the positive vertical fluxes would be due either to locally
generated O 3 (last term) or to the vertical advection
term, which was canceled to obtain Eq. (2). Vertical
advection does not seem a likely explanation given the
lack of correlation between the vertical flux movement
and 6w noted by Zeller and Hehn (1996). However,
based on the current study results, typical magnitudes
for w(]c/]z) are 20.13 mg m23 s21 above 23 m and
10.02 mg m23 s21 from 9 to 23 m. If the snow surface
O 3 concentration were at or near 0.0 ppb, then w(]c/
]z) could be as high as 10.6 mg m23 s21 below 9 m.
Without knowing the O 3 gradient below 9 m, it is difficult to complete this assessment. It is clear that vertical
advection is negative above 23 m and positive below.
It is also likely that some of the observed positive O 3
flux is a result of ‘‘cleaner’’ air with lower O 3 concentrations mixing downward from above. However, this
does not account for the observations of sustained positive flux and of above-canopy sustained negative gradient.
Atmospheric boundary layer O 3 usually requires nitric oxide (NO), nonmethane hydrocarbons, and ultraviolet energy to drive its photochemical production (Olszyna et al. 1994). Forests are sources of natural biogenic nonmethane hydrocarbons, which are known precursors for O 3 production and a possible cause of higher
rural O 3 concentrations. Concentrations of nitrous oxide
(N 2O) above typical ambient levels have been measured
under and above the snow cover at GLEES (Sommerfeld
et al. 1993). Since the same microorganisms that generate N 2O also generate NO (Hutchinson and Davidson
1993), the possibility of an NO source during winter
months exists, but the likely concentration is very low.
Unfortunately, the chemical precursors of ozone associated with forest trace gas emissions, NO x [i.e., nitrogen dioxide (NO2) 1 NO] and biogenic nonmethane
hydrocarbons were not measured for this study. At a
remote Swedish site, Rondon et al. (1993) found that
during June and July (growing season) NO 2 was emitted
from conifers only when concentrations fell below 0.5–
0.7 ppb. There are no corresponding wintertime measurements. At Niwot Ridge, Colorado, which is a similar
forested site at 3050 m elevation and located in the same
Rocky Mountain range 140 km south-southeast of
Snowy Range, Wyoming, Parrish et al. (1986) reported
typical NO x concentrations of 0.25–1.0 ppb during summer and 0.05–0.4 ppb during winter. Urban air from the
Denver metropolitan area occasionally reaches the Niwot site. For this discussion, we assume NO x concentrations at the Brooklyn tower would have been similar
or lower than those measured at Niwot. Given the presumed lack of high O 3 precursor concentrations in this
JANUARY 2000
ZELLER
wintertime scenario, it is unlikely O 3 is generated locally.
A horizontal O 3 gradient is necessary to produce either horizontal advection [Eq. (2) third term] or horizontal turbulent flux [Eq. (2) fourth and fifth terms].
Ozone concentrations appear to be horizontally uniform
in the surrounding area: simultaneously daytime O 3 concentrations near Centennial, Wyoming, 8 km southeast,
during 1990 were within 1–2 ppb of those measured at
3-m height and values measured at the open meadow
U.S. EPA National Dry Deposition Site (NDDN No.
169) 10-m height, 100 m southwest of the Brooklyn
Lake tower site during 1993 were typically within 1 ppb
(Wooldridge et al. 1997). Figure 3 shows that the independently measured 1-h NDDN No. 169 O 3 concentrations during this study are congruent with those at
23 m on the tower up to JD 18.6 when the first weather
system passed through. After JD 18.6 the NDDN No.
169 O 3 values drop 2 ppb below those at 23 m but are
congruent with those measured at 29 m (not shown).
The area surrounding NDDN No. 169 was also snow
covered. Based on the measurement comparisons among
the Brooklyn site, the NDDN site, and the Centennial,
Wyoming, site that suggest the horizontal O 3 gradients
are very small, horizontal advection is an unlikely contributor.
The last term in Eq. (2), R, remains a possible contributor indicating either w9c90 is not zero as assumed
(indicating a snow source) and/or the tree canopy is
contributing O 3 in some way. Galbally and Allison
(1972) speculated O 3 might be absorbed on fresh snow
without total destruction; however, Zeller and Hehn
(1996) showed that adsorption was more likely and
could account for the storage of 2.4 3 10 20 O 3 molecules
m23 of snow base and upon release during turbulent
conditions provide for the observed positive fluxes. The
observed O 3 profile, positive below and negative above
canopy, now leads to the added speculation that the trees
may also be storing O 3 . Ambient temperatures are below
zero, so chemical reactions between biological surfaces
and O 3 may be either significantly retarded or nonexistent. Using 40 ppb O 3 (7.2 3 1011 molecules cm23 ),
an O 3-to-tree sticking coefficient of 0.001 (unknown for
conifers, therefore the value for ice was used), and 3 3
10 4 cm s21 for O 3 molecular velocity, the estimated rate
of O 3 adsorption to trees would be 2.2 3 1013 molecules
cm22 s21 . Given an estimated 2.8 m 2 m22 one-sided leaf
area index (LAI) (8.4 two-sided LAI for conifers) of
the surrounding 17-m forest, plus 1.0 m 2 m22 for trunks
and branches, gives 5.5 3 10 3 cm 2 m23 as the withinforest specific surface. Assuming O 3 saturation within
1 s, there might be 2 3 1018 O 3 molecules m23 in the
surrounding 17-m forest. This highly speculative value,
about 1% of the estimate for snow (Zeller and Hehn
1996), seems inconsequential, but it is worth further
investigation given the rough assumptions made and the
observed O 3 profile. Conifers are known to respire during winter (Prentice et al. 1992). Again, because of the
99
below-freezing air temperatures involved, it is possible
that O 3 is also stored within leaf stomata chambers. If
this were the case, it would increase the above estimated
potential for canopy O 3 storage.
During summer, growth season scenarios, larger
downward O 3 fluxes are typically associated with higher
O 3 concentrations (Zeller and Hehn 1996; Zeller et al.
1993). However, this study and the 1992 study (Zeller
and Hehn 1996) showed wintertime O 3 concentrations
decreasing with increasing positive flux and vice versa.
In the current study the longest sustained period was
from JD 24 to 26 (see Fig. 3) when 9-m O 3 dropped 9
ppb and O 3 flux peaked at 0.7 and 0.9 mg m22 s21 on
JD 24 and JD 25, respectively. In 1992 this phenomenon
was observed over several days of decreasing concentrations and increasing flux followed by several days of
increasing concentrations and decreasing positive flux.
Sudden drops in O 3 concentrations were also associated
with sharp increases in upward O 3 flux. This short-term
burst behavior is less discernible in the current data.
Given an equilibrium between ambient O 3 concentrations and ice surface saturation and possibly tree surface
saturation, this inverse behavior adds credence to the
possibility that O 3 is temporarily stored by snow and
trees and released by turbulent air interactions.
Zeller and Hehn (1996) showed that the delineating
factor for upward versus downward O 3 fluxes appeared
to be snow cover and (to a lesser extent) ambient temperature. Both are environmental factors that potentially
affect stomatal function. The measured fluxes are usually diurnal in nature, reflecting higher daytime turbulent O 3 mass transfer. Some measurements show continued but weaker upward O 3 fluxes during nights with
strong winds. If snow storage were the only source of
O 3 , the only explanation for daytime peak fluxes might
be sublimation of ice resulting in O 3 release. However,
given the shape of the current-study O 3 profile, tree
physiology also appears to play a role. The minimum
ambient temperature for conifer growth is 58C (Prentice
et al. 1992). Ozone fluxes are always negative during
daylight hours when tree canopies are actively growing.
5. Conclusions
The O 3 data results measured by eddy correlation
during January 1993 at a subalpine spruce–fir forest site,
Snowy Range, Wyoming, show positive (upward) wintertime fluxes below canopy (peak 0.1 mg m22 s21 ) that
are counter to the gradient (average 10.05 ppb m21 )
and positive fluxes above canopy (peak 0.2 mg m22 s21 )
that are with the gradient (20.3 ppb m21 ). The upward
flux magnitudes are one-third lower than measured during late winter 1992 but equivalent to those measured
in October 1992. The shape of the 9–23–29-m O 3 profile
suggests there is more ozone available between 9 and
29 m than below 9 or above 29 m. The source of the
extra O 3 is unknown and the explanation for the upward
O 3 fluxes at 9 m is unknown, but the negative 23–29-m
100
JOURNAL OF APPLIED METEOROLOGY
above-canopy O 3 gradient does explain the observed
upward flux at 23 m. The rapid increase in opposite
directions of the above- and below-canopy O 3 gradient
at midnight JD 19 during the cessation of wind demonstrated a significant build up in O 3 concentration near
the 23-m height that implies an O 3 storage (or emission)
at or near that elevation.
The January 1993 measurements presented here do
not disprove the snow storage hypothesis, but they demonstrate a sustained countergradient flux. The forest canopy itself is implicated as an additional, though lesser,
O 3 storage location. Possible future studies include (i)
laboratory experiments with various snow types and
ages to study the quantities and sticking ratios of O 3 to
ice; (ii) laboratory experiments with living conifers under wintertime conditions to study O 3 uptake, storage,
and release (if any); (iii) field experiments at an alternative, flatter subalpine site having a similar forest; (iv)
the addition of at least two (total four) more sampling
heights for O 3 and NO x at the current study site: 1 m
above the snow surface, 96 m, 17 m (tree height), and
296 m; and (v) the addition of horizontally spaced O 3
and NO x concentration measurements surrounding the
current study site.
Acknowledgments. The micrometeorological data analyzed in this paper could not have been collected without the daily in-field technical support of Ted Hehn,
electronic and atmospheric science specialist, USDA
Forest Service, Ft. Collins, Colorado, currently with the
University of Wyoming Department of Renewable Resources.
REFERENCES
Businger, J. A., 1986: Evaluation of the accuracy with which dry
deposition can be measured with current micrometeorological
techniques. J. Climate Appl. Meteor., 25, 1100–1124.
Chameides, W. L., and J. P. Lodge, 1992: Tropospheric ozone: Formation and fate. Surface Level Ozone Exposures and their Effects
on Vegetation, A. S. Lefohn, Ed., Lewis Publishers, 5–30.
Denmead, O. T., and E. F. Bradley, 1985: Flux–gradient relationships
in a forest canopy. The Forest–Atmosphere Interaction, B. A.
Hutchison and B. B. Hicks, Eds., D. Reidel, 421–442.
Enders, G., 1992: Deposition of ozone to a mature spruce forest:
Measurements and comparison to models. Environ. Pollut., 75,
61–67.
, U. Teichmann, and G. Kramm, 1989: Profiles of ozone and
surface layer parameters over a mature spruce stand. Mechanism
and Effects of Pollutant-Transfer into Forests, H. W. Georgii,
Ed., Kluwer Academic Publishers, 21–35.
Fitzjarrald, D. R., and K. E. Moore, 1992: Turbulent transports over
tundra. J. Geophys. Res., 97 (D15), 16 717–16 729.
Fontan, J., A. Minga, A. Lopez, and A. Druilhet, 1992: Vertical ozone
profiles in a pine forest. Atmos. Environ., 26A, 863–869.
Galbally, I., and I. Allison, 1972: Ozone fluxes over snow surfaces.
J. Geophys. Res., 77, 3946–3949.
Gao, W., and M. L. Wesely, 1994: Numerical modeling of turbulent
fluxes of chemically reactive trace gases in the atmospheric
boundary layer. J. Appl. Meteor., 33, 835–847.
Hutchinson, G. L., and E. A. Davidson, 1993: Processes for production and consumption of gaseous nitrogen oxides in soil. Agricultural Ecosystem Effects on Trace Gases and Global Climate
VOLUME 39
Change, Special Publ. 55, American Society of Agronomy, 79–
93.
Kelly, J. J., Jr., and J. D. McTaggart-Cowen, 1968: Vertical gradient
of net oxidant near the ground surface at Barrow, Alaska. J.
Geophys. Res., 73, 3328–3330.
Lenschow, D. H., R. Pearson Jr., and B. B. Stankov, 1982: Measurements of ozone vertical flux to ocean and forest. J. Geophys.
Res., 87 (C11), 8833–8837.
Leuning, R., and M. J. Judd, 1996: The relative merits of open- and
closed-path analysers for measurement of eddy fluxes. Global
Change Bio., 2, 241–253.
Lopez, A., J. Fontan, and A. Minga, 1993: Analysis of atmospheric
ozone measurements over a pine forest. Atmos. Environ., 27A,
555–564.
Massman, W. J., and K. F. Zeller, 1988: Rapid method for correcting
the non-cosine response errors of the Gill propeller anemometer.
J. Atmos. Oceanic Technol., 5, 862–869.
, D. G. Fox, and K. F. Zeller, 1990: Verifying eddy correlation
measurements of dry deposition: A study of energy balance components of the Pawnee Grasslands. Research Paper RM-288,
USDA Forest Service RMFRES, Ft. Collins, CO.
, R. A. Sommerfeld, A. R. Mosier, K. F. Zeller, T. J. Hehn, and
S. G. Rochelle, 1997: A model investigation of turbulence-driven
pressure-pumping effects on the rate of diffusion of CO 2 , N 2O,
and CH 4 through layered snowpacks. J. Geophys. Res, 102
(D15), 18 851–18 863.
Matt, D. R., and J. D. Womak, 1989: Atmospheric surface exchange
of ozone to a spruce forest: Measured and inferred. Ozone in
the Atmosphere, R. D. Bojkov and P. Fabian, Eds., Deepak Publishing, 490–493.
McMillen, R. T., 1988: An eddy correlation technique with extended
applicability to non-simple terrain. Bound.-Layer Meteor., 43,
231–245.
Miller, D. R., J. D. Lin, and Z. N. Lu, 1991: Air flow across an alpine
forest clearing: A model and field measurements. Agric. For.
Meteor., 56, 209–225.
Musselman, R. C., Ed., 1994: The Glacier Lakes ecosystem experiments site (GLEES): An alpine global change research study
area. Tech. Rep. RM-249, USDA Forest Service RMFRES, Ft.
Collins, CO, 94 pp.
Olszyna, K. J., E. M. Bailey, R. Simonaitis, and J. F. Meagher, 1994:
O 3 and NO y relationships at a rural site. J. Geophys. Res., 99
(D7), 14 557–14 563.
Padro, J., 1993: Seasonal contrasts in modelled and observed dry
deposition velocities of O 3 , SO 2 , and NO 2 over three surfaces.
Atmos. Environ., 27A, 807–814.
Parrish, D. D., and Coauthors, 1986: Measurement of the NO x–O 3
photostationary state at Niwot Ridge, Colorado. J. Geophys.
Res., 91 (D5), 5361–5370.
Prentice, I. C., C. Wolfgang, S. P. Harrison, R. Leemans, R. A. Monserud, and A. M. Solomon, 1992: A global biome model based
on plant physiology and dominance, soil properties and climate.
J. Biogeogr., 19, 117–134.
Raupach, M. R., 1989: Stand overstory processes. Philos. Trans. Roy.
Soc. London B, 324, 175–190.
Ray, J. D., D. H. Stedman, and G. J. Wendel, 1986: Fast chemiluminescent method for measurement of ambient ozone. Anal.
Chem., 58, 598–600.
Rondon, A., C. Johansson, and L. Granat, 1993: Dry deposition of
nitrogen dioxide and ozone to coniferous forests. J. Geophys.
Res., 98 (D3), 5159–5172.
Sommerfeld, R. A., A. R. Mosier, and R. C. Musselman, 1993: CO 2 ,
CH 4 , and N 2O flux through a Wyoming snowpack and implications for global budgets. Nature, 361, 140–142.
Stocker, D., K. Zeller, and D. Stedman, 1995: O 3 and NO 2 fluxes over
snow measured by eddy correlation. Atmos. Environ., 29, 1299–
1305.
Wesely, M. L., 1983: Turbulent transport of ozone to surfaces common in the eastern half of the United States. Trace Atmospheric
JANUARY 2000
ZELLER
Constituents: Properties, Transformations, and Fates, S. E.
Schwartz, Ed., John Wiley and Sons, 345–370.
Wooldridge, G., K. Zeller, and R. Musselman, 1997: Ozone concentration characteristics in and over a high-elevation forest site.
Theor. Appl. Climatol., 56, 153–164.
Wunderli, S., and R. Gehrig, 1990: Surface ozone in rural, urban and
alpine regions of Switzerland. Atmos. Environ., 24A, 2641–2646.
Zeller, K., and T. Hehn, 1994: Wintertime anomalies in ozone deposition above a subalpine spruce–fir forest. Proc. Fourth USDA
Forest Service, Research and Applications of Chemical Sciences
in Forestry, USDA, 131–138.
101
, and
, 1995: Ozone deposition in a snow-covered subalpine
spruce–fir environment. Proc. Boulder Symp., Biogeochemistry
of Seasonally Snow-Covered Catchments, IAHS Publ. 228, 17–
22.
, and
, 1996: Measurements of upward turbulent ozone fluxes above a subalpine spruce–fir forest. Geophys. Res. Lett., 23,
841–844.
, and Coauthors, 1993: Initial results from the Pawnee eddy
correlation system for dry acid deposition research. Research
Paper RM-282, USDA Forest Service RMFRES, Ft. Collins, CO,
44 pp.