GEOPHYSICAL RESEARCH LETTERS, VOL. 18, NO. 7, PAGES 1225-1228, JULY 1991
PRELIMINARY MEASUREMENTS OF CO2 IN MELTING SNOW
R.
A.
U.S.D.A.
Sommerfeld,
Forest
R.
Service,
C.
Abstract.
Measurements
snow-soil
interface
concentrations
Concentrations
R.
the
observed
0.45
soil
interface.
m above
the
in
the
snow
the
meltwater
from
the
the
snow-soil
soil
[Bales
base
of
interface
at
with stainless
of approximately
cylinders
had one end closed
stainless
steel
could
be
The
snowpacks
at the
[Coyne and Kelley,
after
about
and
soil
of
mm.
Three
50
undisturbed
their
mm and
450
above
layer
centers
of
Four
their
ground.
One was
center
at a depth
were
snow from
mm above
in
snow had
soil-snow
interface.
were
installed
with
the
its
wall;
installed
(5-mm)
to
with
about
in the
tubing
m of
was
a thin
a
NPT
collectors.
0.6
axes parallel
in the soil
into
onset
of snowmelt
1974;
Solomon,
with
the
were
The
freezing,
of CO•
soils
with
A 1/16
to
collectors
accumulated.
Introduction
in
plate.
attached
ice
at the
collectors
concentrations
steel
mesh with openings
50 •m.
One of the
Swagelok fitting
was inserted
wall of each cylinder
so that
the
CO2.
observed
with
1990].
The gas collectors
consisted
of
stainless
steel
cylinders
100 mm in
diameter
and 10 mm high (Figure
1).
The ends of the cylinders
were covered
the
atmospheric
Elevated
al.,
Methods
January
been
et
did not
interchange
in CO2
start
of melt could have important
effects
on meltwater
chemistry,
especially
since streams at this time
flow under a covering
of snow
preventing
equilibration
with
have
Station
Service
coincided
snowpack were consistent
with the
measured CO• levels.
Decreases in pH at
constant
alkalinity
of up to 0.8 units
were associated
with the excess CO•.
The origin
of the excess CO• is
uncertain
but may be related
to litter
decomposition.
Elevated
levels
of CO•
near
Research
the
with the beginning
of melt.
Measurements
of the pH and alkalinity
of
Experiment
narrow lysimeters
that
interfere
with gaseous
snow-
The increase
concentrations
Reuss
the snowpack about 200 m from the
collection
site
using a set of long,
elevated
up to 2120 ppmv.
greater
than 1700 ppmv
were
O.
Mosier
Agricultural
of CO2 near
showed
and J.
Rocky Mountain
A.
U.S.D.A.
Musselman,
were
the
inserted
a snow pit
at
50 mm,
soil
surface.
150
1987].
Significant
increases,
if
common, would have important
effects
on
the chemistry
of meltwater,
overland
flow,
and soil
water.
Such effects
would be especially
significant
at the
time of melt initiation
in the spring,
because
blanket
for
the
of
streams
snow
that
several
stream
months.
water
with
at
of
base
the
We designed
system
to
be
the
close
the
to
interstitial
air
snowpack.
a gas collection
observe
concentrations
under
a
accumulated
Therefore,
should
equilibrium
the
flow
has
in
the
CO•
the
soil
and
in
the
snow near the snow-soil
boundary.
During
the 1990 melt season,
we
collected
samples from one location
at
the Glacier
Lakes Ecosystems Experiment
Site
(GLEES) at an elevation
of 3280 m.
The
site
Wyoming,
collected
is
55 km west
in
the
Snowy Range.
of
Laramie,
meltwater
from
the
We also
base
of
This paper is not subject to U.S. copyright.
Published in 1991 by the American Geophysical Union.
Paper
Fig.
number 91GL01502
1.
Sketch
of the
CO2 collector.
1225
This file was created by scanning the printed publication.
Errors identified by the software have been corrected;
however, some errors may remain.
1226
A
Sommerfeld
collector
placed
soil
with
with
one
its
surface
closed
end
et
al.:
COz in Snow
was
1.90
mesh end down on the
in
contact
with
a
litter
1.75
layer
approximately
7 mm thick.
A 3-m
length
of 1.5-mm (1/16-in.)
OD teflon
tubing
was attached
to each collector,
with
the
above
other
the
snow pit
end
held
expected
on
a
snow level.
was carefully
1.60
standard
The
backfilled
1.45
with
snow, and the collectors
were left
undisturbed
for the snowpack to develop
for
the
rest
of
the
season.
Gas sample collection
began on
April
6, just
prior
to snowmelt when
the snowpack was approximately
1.9 m
thick.
Snowmelt began on approximately
April
10 as evidenced
by melt-freeze
metamorphism
observed
in the surface
layers.
Significant
meltwater
began to
flow in snow lysimeters
on April
23.
We drew
samples
from
the
disposable
with plastic
10 ml were drawn
and discarded
to completely
purge the
tubing,
whose internal
volume was
approximately
1.5 ml.
We obtained
the
first
sample set using a hypodermic
needle that
closely
fit
the inside
diameter
of the tubing.
However,
we
could
not
be
certain
that
the fit
was airtight,
when taking
subsequent
samples
we used Swaglok
attachments
to insure
an airtight
fit.
All
samples
were collected
between
1100
and 1200 local
time.
In addition,
we
obtained
samples from the air over the
snow
and
from
containing
a
tank
at
the
approximately
site
2000 ppmv COz
in
dry air.
Syringes
were put into Zip-Lok
plastic
bags for transport
within
1 hr
to a field
laboratory
at 2585 m
elevation
and
a
distance
km.
At the
field
were
filled
with
lab
of
the
deionized
about
12
plastic
water
bags
to
set
calibrated
of
samples
tank
was
taken
of COz in air
from
at
a
343
ppmv; approximately
ambient
concentration.
Samples transported
in
water
showed significantly
less scatter
than those tranported
in air.
The syringes
were then taken to
Fort
Collins,
Colorado
for sample
analysis
by gas chromatography
(GC).
An ultrasonic
detector
and Porapak
Q
column system were used for the GC
analyses
[Mosier and Mack, 1980].
detection
limit
of the GC system
near 10 ppmv with a precision
of
Carbon
dioxide
upon calibrated
Scott Analytical.
concentrations
gas
\
\
MAY
28\
0
..Q
0.85
•
0.70
standards
0.40APR6
0.25
O.lO
/
where
measured
concentrations
ambient
are
•'•,
'
/
17
JUN
6
APR 18
/
I
\
-o.o5
o
2000
lOOO
NR
P CO2 , ppmv
Fig.
2.
CO• concentrations
heights
The
above
the
at different
snow-soil
concentration
interface.
labeled
AIR
is
350
ppmv; approximately
the world average.
Extra points
on May 17 were taken from
tubing
inserted
into
the
snow.
height
of the snow-air
surface
for each
sample day.
Each data point
represents
a single
sample.
During most of the
studied,
CO• was elevated
above
atmospheric
concentrations
in the
bottom
45 cm of the snowpack by factors
of 4 to more than 6.
On April
25, May
9, and May 28 the concentrations
in the
air
above the snow were 398, 415, and
390 ppmv, respectively,
indicating
a
possible
enrichment
of the atmospheric
concentration
from
Samples
the
taken
showed that
based
from
standards.
the
GLEES
snow.
from
the
CO• was lost
Samples
or
from
consistently
Tank samples
tank
at
the
from the
transported
the
field
from
lab
were
lower
taken
than this
value.
at the GLEES and at
the field
lab averaged
1483 ppmv (S.D.
136) with no significant
difference
between
the samples
taken
at the two
locations.
On June
are
-
syringes.
The tank averaged
1845 ppmv,
(SD 8.6)
at the Fort Collins
laboratory,
based on our calibration
The
is
1%.
are
(D 0.55
site
Results
The measurements
APR
25
klAY
9 '•
(D 1.00
period
preclude
diffusion
of any gas between
the surrounding
air
and the syringes.
A test
ß
gas
collectors
using 12-ml
plastic
syringes
fitted
valves.
Approximately
because
E 1.30
shown in
Figure
air
plotted
at
the
2
the
GLEES
transported
two samples
6 tank
and
the
samples
field
in two ways.
was in water
lab
taken
at
were
One set
directly
of
from
Sommerfeld
the
deionizer;
the
whose COz content
bubbling
for 1/2
was
with
ambient
those
COz averaged
in
the
1560 ppmv COz.
diffused
from
water,
water
the
1377 ppmv
set
averaged
indicates
plastic
from
that
syringes
the
water,
COz
into
through
the Zip-Lok
bags,
into the atmosphere.
On June 18 we collected
parallel
samples using the disposable
plastic
syringes
and glass syringes
with
plungers
sealed with mineral
oil.
Four
samples
each were collected
from -50,
0, and 50 mm and the tank.
The samples
in the plastic
syringes
were about 10%
lower
in COz than those
in the glass
of
lysimeters
water
from
taken
snowmelt
during
the
[1987].
relative
described
-5
above
concentrations
to
45
cm
show
that
the
to April
June
of
.15
18-25
6 are
collector
described
possible
and .25
errors.
The
m from April
6
and from April
much larger
experimental
errors.
above
18-25
than
the
Also,
the
tests
that
the
errors
showed
to
were systematic
and proportional
concentration
with the highest
to the
collector
low.
measurements
Thus the
probably
spread
larger
about
indicate.
While
the
measurements
consistent
generation
in the
measurements
June 6 indicate
originated
examination
near
the
with
litter
on Apr.
the
25 and
in the snow. Microscppic
of a sample of snow taken
collectors
filamentous
comm.].
identification
Apr.
some COz may have
revealed
presence
of significant
fungi,
bacteria,
alga
pers.
are
COz
layer,
6,
that
of COz
active
that
were measured, generation
1.0 g m-z COz day -•.
of COz
0.5
to
This estimate
is
based on a simple Fick's
Law
calculation
assuming a solid
fraction
of
0.5
and
COz in air
a diffusion
of 13.9
consistent
with
presented
coefficient
mm
z s -•.
the
for
It
is
measurements
in Soloman and Cerling
[1987].
The effect
on meltwater
is
of
The equilibrium
of the reaction
of COz with
(H+) (HCO3-) / P
(1)
pressure
a function
alkalinity.
expression
water
of COz partial
pH is
c0•.
= K
denote
activity
for the ions and Pco
2 is the partial
pressure
generally
for the gaseous CO2. Log K is
given as -7.81.
Equation (1)
combines the reaction of CO2 with water
and the dissociation
of H2CO
3 to form H+
and HCO3-, so that
includes
•he
constant
the Henry's
the constant
law constant
for dissociation
(average
samples),
the
numbers of
and diatoms,
with
predominating
.73
and
of H2CO
3.
ppm for
alkalinity
[R. Dufford
The more
specific
of these
organisms
100 snow
may be defined
as,
alk = [HCO3-] + [CO32-]
+ [OH-] -
[H+]
(2)
where
the
[Reuss
brackets
denote
and Johnson,
concentration
1985].
In
these
systems [OH-] and [CO32-] are
insignificant
and may be neglected,
may the
distinction
as
between
concentration
and activity
strength = 10-4 to 10-5).
(ionic
With these
combining (1) and (2)
simplifications,
gives,
H+)2 + (alk)(H +) - K P
c02
20%
in the values
was
than the measurements
generally
Generation
indicate
would have to be approximately
measured on any
concentrations
on May 17 and May 28 may
have been essentially
identical
(Figure
2) although
their
differences
represent
the extreme
changes at
would
respiration
by these organisms, which
would in turn imply an unidentified
substrate.
To maintain
the gradients
carbon
day can be quantitatively
compared but
that the comparisons among different
days are less reliable.
For example,
the
snow
For water draining
from the snowpack
with very little
dissolved organic
Discussion
tests
the
where the parentheses
with approximately
the same snow depth.
The alkalinity
was determined
by means
of Gran titration.
The pH was
generally
measured within
24 hr on
samples stored
in sealed plastic
bottles.
Our laboratory
and quality
control
proceedures
are described
in
The
1227
same period
as the COz samples at a site
approximately
200 m from the COz site
EPA
in
meltwater
syringes
which in turn were about 10%
lower than the tank which was analyzed
in Fort Collins
for comparison.
The points
in Figure
3 are from
analyses
CO•. in Snow
remains to be done.
by
through
it
in water
second
This
and
in
had been increased
gas from the tank
hour.
The syringes
while
the
other
et al.:
= 0
(3)
Thus, for any combination of
and CO2 partial
pressure,
the
found from the real positive
equation
(3). as shown by the
Figure
alkalinity
pH may be
root of
curves in
3.
The estimate
of 1620 ppmv CO2 was
obtained by fitting
the pH-alkalinity
relationship
derived from Eq. (3) using
a nonlinear
regression
procedure.
The
R2 is 0.87, with most of the lack of
fit attributable
to the single outlier.
This estimate is well within the range
of the
measured
independent
sampling
values,
and provides
verification
of the gas
measurements.
The
300
and
an
1228
Sommerfeld et al.:
8.0
-
7.5
-
7.0
-
6.5
-
COz in Snow
snow
300
is
least
3000 ppmv
atmosphere
snow
equilibrium
and
-
-so
s'o
Alkolinity eq/I
Fig. 3. pH versus Alkalinity:
theoretical
curves for different
concentrations
lysimeter
data.
(ppmv)
COz
of
which
suggests
some contamination
of the snowmelt
surface
water
[Bales
et al.,
1990].
However,
given the nature
of the
between
to
establish
are useful
the
probable
The presence
of
coupled with
negative
clearly
pH values
shows that
point
in the melting
acids
CO•
are
being
eluted
alkalinity
from the
the
levels
the
these
pH of
layer.
CO• content
solution
P.
the
increase
in
J.
carbon
J.,
dioxide
across
of
soil
solution
acid-
neutalinzing
capacity
by addition
of
dissolved
inorganic
carbon.,
Envir.
Sci. Tech..
2__3.1021-1024,
1989.
Handbook
of
Methods
Studies,
for
for
Acid
Laboratory
Surface
Water
Chemistry,
U.S.
Protection
Agency, Washington,
Sci.
1121-1123,
Soc.
Environmental
DC,
Amer.
J..
44.
1980.
Reuss, J. O. & Johnson,
D. W.,
of soil
processes
on the
acidification
deposition,
591-595,
Reuss,
J.
of soils
J. Envir.
Effect
by acid
Qual.. 1__2.
1985.
O.,
& Johnson,
Deposition,
Soils
Verlag,
D. W.,
Acid
and Waters,
Studies
Springer
Series
N__.Y.,.
#59.
109 pp.,
1986.
Solomon, D. K. & Cerling,
annual carbon dioxide
we
of CO• of
ambient
were
the
GLEES.
The
at
the
time
of
snowmelt.
However,
& Kelley,
Atmos.
1990.
soil:
2257-2265,
T. E., The
cycle in a
observations,
and implications
for
Water Resources Res.
2__3.
1987.
are:
concentrations
to
I.
Ecoloqical
1985;
and
conclusion
measurements
Sommerfeld,
R. A., and
G., Ionic
tracer
movement
Variations
Soil
The highest
concentrations
occurred
at the snow-soil
boundary,
indicating
a source in the observed
litter
soil
a Wyoming snowpack,
24A(11).
2749-2758,
modeling,
weathering,
and
6 times
normal
in the
snow at
of
Coyne,
montane
obtained
began
onset
the high
Mosier,
A. R., & Mack, L.,
Gas
chromatographic
system for precise,
rapid
analysis
of Nitrous
Oxide,
Conclusions
more
than
measured
of
1987.
1989].
Elevated
through
Envir..
Analyses
some
strong
atmosphere
[Reuss and Johnson,
Reuss and Johnson,
1986; David
from
time
the
C.,
D.
at
process
drainage increases after equilibration
with the lower CO• in the open
drew
the
at
snowpack.
Deposition
replaces
base cations
on the exchange
complexes.
Thus, while equilibrium
with elevated
CO• decreases the pH of
the soil
solution,
the alkalinity
is
actually
increased.
As a result
of the
results
the
Bales,
R.
Kebler,
EPA,
below
resulting HzCO
3 is dissociated into H+
and HCO3-. Much of the resulting H+
The
with
affect
alkalinity
snowpack.
The percolation
of waters
containing
elevated
levels of CO• into
and through the soil also has
implications
for soil solution and
drainage water chemistry.
The
Vance,
at
cover
Generation
in
pressure.
increased
snow,
an arctic
snowpack during spring.
J. Geoph¾s. Res..
7__9. 799-802,
1974.
David, M. B., & Vance, G. F.,
by
pH, alkalinity,
and CO•, these points
values,
the
References
100 •eq/1,
relationship
can
from
CO• on the relationship.
A few of the
alkalinity
values shown in Figure 3 are
5.0,
over
and snow
3000 ppm lines in Figure 3 are included
to provide perspective
on the effect
of
helping
the
chemistry.
The pH of the snowmelt
water
was
depressed
by the elevated
CO•.
However,
presence
of negative
alkalinities
and pH values
below 5.0 in
some samples indicates
that at some
times strong acid pulses are eluted
4.5
excess
of
locally.
The
5.0
in
some
initial
snowmelt prevents
the water
from coming into equilibrium
with the
atmosphere,
so that the meltwater
is in
-
4.0
with
CO• evolved
from the snow appears
to have increased
the COz concentration
1620
in the
5.5
consistent
measurements.
a source
in the
R.A. Sommerfeld,
R.C. Musselman,
andJ.O. Reuss,
U.S.D.A. ForestServide,RockyMountainExperiment
Station,Fort Collins,Colorado80526-2098.
A.R. Mosier,U.S.D.A. Agricultural
Research
Service,
Fort Collins, Colorado80526.
Received:
April
Accepted:
May
18,
13,
1991
1991