Predictions of Advection

1
Predictions of Advection Dispersion in the
Tidal Hudson River A High-Resolution SF6 Tracer
Study
Peter Schlosser1,2,3, David T. Ho2,3, Paul
Schmieder2,3 Presenter Inna Sokolyanskaya
2,4 1Dept. of Earth Environmental Engineering,
Columbia University 2Lamont-Doherty Earth
Observatory, Columbia University 3Dept. of Earth
Environmental Sciences, Columbia
University 4Dept. of Environmental Sciences,
Barnard College
Fig. 2. High-Resolution, Automated, Real-Time
Measurement System.
Abstract The design of environmental
policy for the proper management of the Hudson
River is dependant upon a good understanding of
water motion dynamics. Ongoing experimentation
with the physical processes of advection and
dispersion is thus crucial to determine the
patterns by which natural and anthropogenic
contaminants introduced into the water will
spread and mix. For the past
twelve years geochemists from Columbia
Universitys Lamont-Doherty Earth Observatory
have been conducting research on the tidal Hudson
to create a realistic visualization of the
rivers hydrodynamics. The latest large-scale
tracer experiment was conducted during the summer
of 2004. Two others had been previously conducted
on different sections of the river in the summers
of 2001 and 2003, and the results of these
studies were subsequently analyzed.
After injection, the tracer was monitored from a
boat using an automated measurement system for
periods of 14, 11 and 11 days, in 2001, 2003, and
2004, respectively. From previous analyses of the
earlier experiments, it was determined that
longitudinal dispersion was more rapid in 2001
(70.1 4.3 m2/sec) than in 2003 (days 1-4
24.66.6 m2/sec days 4-11 48.0 1.9 m2/sec).
Mean advection was initially greater in 2003
(days 1-4 3.5 0.2 km/day) than in 2001 (0.8
0.1 km/day), but then decreased to a value
similar to that of 2001 (days 4-11 0.9 0.1
km/day). 1 Since the dimensions of
the Hudson are not uniform throughout the length
of the channel and the river flow changes daily,
both the river geometry and the discharge are
taken into account so that the net advection and
the longitudinal dispersion of the tracer may be
predicted for the 2004 data. The mean
cross-sectional areas of the experimental
sections of the river were 17598 m2 during 2001
and 12135 m2 (first four days) and 14220 m2
(final seven days) during the 2003 experiments.
The relationship between the cross-sectional area
and the longitudinal dispersion was direct. The
mean river discharges were 121 m3/sec in 2001,
and 285 m3/sec (first four days) and 181 m3/sec
(final seven days) during the 2003 experiments.
The relationship between the river discharge and
net advection was direct. During the
experiment in 2004, the mean cross-sectional area
of that particular section of the Hudson was 5400
m2 and the mean daily discharge was 40 m3/sec.
Since that cross-section is less than half of
that during the 2003 experiment, it is likely
that the longitudinal dispersion will be even
smaller than that in 2003. In addition, the river
flow in 2004 was one third of the mean in 2001.
Thus, we can expect an even smaller net advection
of the tracer than had been that year. Finally,
Fig. 1. Map of the Hudson River.
The section in red indicates the stretch of the
tidal Hudson River where the 2004 experiment was
conducted. Sections in blue and green show where
the tracer release experiments were conducted in
2001 and 2003, respectively. The section in
orange indicates where the two earlier
experiments overlapped.
c. Water is pumped through the membrane contactor
of the automated measuring system, where gases
are stripped via counter flow of N2.
a. The boat (Riverkeeper) in New Baltimore, NY.
The sampling pump attached to the bow is tilted
up for storage.
b. Gas is bubbled into the water column via a
perforated hose.
Abstract (contd) since the tracer was injected
at slack before ebb during 2004, we can expect
its upstream location (relative to injection
point) to be similar to that of 2003, when the
SF6 was released at the same time in the tidal
cycle. 1 Ho, D.T., P. Schlosser, F. Hellweger,
T. Caplow, Factors controlling net advection and
longitudinal dispersion in the tidal Hudson
River Results from SF6 tracer release
experiments, New York, Eos Trans. AGU, 84(46),
Fall Meet. Suppl., Abstract H41D-1041, 2003.

Fig. 3. Time-Series Images from
Earlier Experiments. (Ho and Schlosser, 2003)
Daily tidally-corrected longitudinal SF6
distribution from tracer release experiment in
2001 (top) and 2003 (bottom). The vertical line
in each plot denotes the injection point.
Distributions from the first day of injection for
both years are not shown, and distribution from
July 31, 2001 is incomplete and has been omitted.
Results from the 2004 experiment are not yet
ready in presentable form.

Fig. 4. Hudson River Discharge. Tidal river
discharge measured at Green Island, NY, in 2001
(left), 2003 (center) and 2004 (right) during the
period of the experiment.
Method Sulfur hexafluoride (SF6) is
used extensively in field experiments due to its
non-toxicity, detection limit over a large
concentration range, inert nature and
inexpensiveness. During the three SF6 studies, an
automated, high-resolution tracer measurement
system mounted on a boat. The system monitored
the tracer continuously via a submersed pump that
was attached to the front of the boat, which
transferred water through a PVC hose that
connected to a gas extraction unit. Onboard, the
gas was analyzed using a gas chromatograph (See
Fig. 2.) A laptop computer controlled all the
functions, collected data, and displayed results
in real time. During each
experiment, ca. 4.3 moles of the inert gas SF6
were injected into the Hudson River, several
meters above the bottom, near Newburgh, NY in
2001, Hyde Park, NY, in 2003, and New Baltimore,
NY in 2004. After injection, the tracer was
monitored using the automated measurement system
for periods of 14, 11 and 11 days, in 2001, 2003,
and 2004, respectively. The sampling interval
was two minutes and the detection limit was 1 x
10-14 mol L-1. The spread of the tracer plume
during 2001 and 2003 are shown in Fig. 3.
Fig. 7. Predictions for 2004
Based on Previously Analyzed Data (Ho and
Schlosser, 2003)
Fig. 6. Hypothetical Advection of
Tracer, 2004. The net advection
is calculated by dividing the river discharge by
the cross-sectional area.
Fig. 5. The cross-sectional area
of the Hudson River. The area decreases steeply
with movement upstream.
Results and Discussion Predictions for
the 2004 data are based on the relationships
observed in the previous experiments, as well as
Hudson River geometry and discharge. Net
advection is a major mechanism that affects the
movement of dissolved pollutants, and is thus
crucial in understanding how far downstream a
contaminant will travel. Advection is mainly
dependant upon the river discharge, the magnitude
of which is noted in Fig. 4 for the different
years of the experiment. The flow in 2003,
particularly at the beginning of the experiment,
has been greatest. A subsequent observation of
Fig. 3 reveals that the net advection of the
tracer peaks had differed from day to day in 2001
and 2003, but had been greatest during the first
four days of the 2003 experiment (when the flow
was highest). On average, the river discharges
were 121 m3/sec in 2001 and 285 m3/sec (first
four days) and 181 m3/sec (final seven days)
during the 2003 experiments. In 2004, it was 40
m3/sec. It is thus concluded that in 2004 the
mean net advection will be the smallest of the
three years. A visualization of
longitudinal dispersion provides an understanding
of how far upstream (and downstream) a
hypothetical contaminant will spread. Dispersion
is heavily dependant upon river geometry,
including sharp turns and breaks in a channel,
but it particularly relies on the cross-sectional
area of the channel. The cross-sectional area of
the Hudson decreases with movement upstream, as
can be seen in Fig 5. Relationships from the
2001 and 2003 experiments indicate that
decreasing cross-sectional area results in
decreasing dispersion. It is thus concluded that
dispersion will be smallest in the upper part of
the Hudson, where the 2004 experiment was
conducted. Finally, it is
important to know the location of a contaminant
relative to the injection point. This location
is dependant on when during the tidal cycle
pollutant was released. Release at slack before
flood will cause contaminant to be moved upstream
by the tide. In 2004, the tracers release at
slack before ebb will cause the relationship
between the final location and the injection
point to be similar to that of 2003, also
conducted at slack before ebb. For all results,
see Fig. 7.