Abstract
Accurate measurements of suspended-sediment
concentration require suspended-sediment samplers to
operate isokinetically, within an intake-efficiency range of
1.0 ± 0.10, where intake efficiency is defined as the ratio
of the velocity of the water through the sampler intake to
the local ambient stream velocity. Local ambient stream
velocity is defined as the velocity of the water in the river at
the location of the nozzle, unaffected by the presence of the
sampler. Results from Federal Interagency Sedimentation
Project (FISP) laboratory experiments published in the early
1940s show that when the intake efficiency is less than 1.0,
suspended-sediment samplers tend to oversample sediment
relative to water, leading to potentially large positive biases
in suspended-sediment concentration that are positively
correlated with grain size. Conversely, these experiments
show that, when the intake efficiency is greater than 1.0,
suspended‑sediment samplers tend to undersample sediment
relative to water, leading to smaller negative biases in
suspended-sediment concentration that become slightly more
negative as grain size increases.
The majority of FISP sampler development and testing
since the early 1990s has been conducted under highly
uniform flow conditions via flume and slack-water tow
tests, with relatively little work conducted under the greater
levels of turbulence that exist in actual rivers. Additionally,
all of this recent work has been focused on the hydraulic
characteristics and intake efficiencies of these samplers,
with no field investigations conducted on the accuracy of the
suspended-sediment data collected with these samplers. When
depth-integrating suspended-sediment samplers are deployed
under the more nonuniform and turbulent conditions that
exist in rivers, multiple factors may contribute to departures
from isokinetic sampling, thus introducing errors into the
suspended-sediment data collected by these samplers that may
not be predictable on the basis of flume and tow tests alone.
This study has three interrelated goals. First, the
intake efficiencies of the older US D-77 bag-type and
newer, FISP-approved US D-96-type1 depth-integrating
suspended‑sediment samplers are evaluated at multiple
cross‑sections under a range of actual-river conditions. The
intake efficiencies measured in these actual-river tests are
then compared to those previously measured in flume and tow
tests. Second, other physical effects, mainly water temperature
and the duration of sampling at a vertical, are examined to
determine whether these effects can help explain observed
differences in intake efficiency both between the two types
of samplers and between the laboratory and field tests. Third,
the signs and magnitudes of the likely errors in suspendedsand
concentration in measurements made with both types of
samplers are predicted based the intake efficiencies of these
two types of depth-integrating samplers. Using the relative
difference in isokinetic sampling observed between the
US D-77 bag-type and D-96-type samplers during river tests,
measured differences in suspended-sediment concentration
in a variety of size classes were evaluated between paired
equal-discharge-increment (EDI) and equal-width-increment
(EWI) measurements made with these two types of samplers
to determine whether these differences in concentration are
consistent with the differences in concentrations expected
on the basis of the 1940s FISP laboratory experiments. In
addition, sequential single-vertical depth-integrated samples
were collected (concurrent with velocity measurements)
with the US D-96-type bag sampler and two different rigidcontainer
samplers to evaluate whether the predicted errors in
suspended-sand concentrations measured with the US D-96-
type sampler are consistent with those expected on the basis of
the 1940s FISP laboratory experiments.
Results from our study indicate that the intake efficiency
of the US D-96-type sampler is superior to that of the US D-77
bag-type sampler under actual-river conditions, with overall
performance of the US D-96-type sampler being closer to, yet
still typically below, the FISP-acceptable range of isokinetic
operation. These results are in contrast to the results from
FISP-conducted flume tests that showed that both the US D-77
bag-type and US D-96-type samplers sampled isokinetically
in the laboratory. Results from our study indicate that
the single largest problem with the behavior of both the
US D-77 bag-type and the US D-96-type samplers under
actual‑river conditions is that both samplers are prone to large
time‑dependent decreases in intake efficiency as sampling
duration increases. In the case of the US D-96-type sampler,
this problem may be at least partially overcome by shortening
the duration of sampling (or, instead, perhaps by a simple
design improvement); in the case of the US D-77 bag-type
sampler, although shortening the sampling duration improves
the intake efficiency, it does not bring it into agreement with
the FISP‑accepted range of isokinetic operation.
The predicted errors in suspended-sand concentration
in EDI or EWI measurements made with the US-96-type
sampler are much smaller than those associated with EDI
or EWI measurements made with the US D-77 bag-type
sampler, especially when the results are corrected for the
effects of water temperature and sampling duration. The bias
in the concentration in each size class measured using the
US D-77 bag-type relative to the concentration measured
using the US D-96-type sampler behaves in a manner
consistent with that expected on the basis of the observed
differences in intake efficiency between the two samplers in
conjunction with the results from the 1940s FISP laboratory
experiments. In addition, the bias in the concentration in
each size class measured using the US D-96‑type sampler
relative to the concentration measured using the truly
isokinetic rigid-container samplers is in excellent agreement
with that predicted on the basis of the 1940s FISP laboratory
experiments. Because suspended-sediment samplers can
respond differently between laboratory and field conditions,
actual-river tests such as those in this study should be
conducted when models of suspended-sediment samplers
are changed from one type to another during the course
of long-term monitoring programs. Otherwise, potential
large differences in the suspended-sediment data collected
by different types of samplers would lead to large step
changes in sediment loads that may be misinterpreted as
real, when, in fact, they are associated with the change in
suspended‑sediment sampling equipment.
1For the purpose of this study, both the US D-96 and the US D-96-A1
sampler (Davis, 2005) are herein referred to as the US D-96-type sampler.