spot aij is selected at point P on a gray-level image
determining vector fields of water-surface veloci-
at time t0. A second spot bij is located away from
ties and ice velocities requires the same overall pro-
point P on a subsequent image taken at time t = t0
cedure as used for measuring various laboratory
+ dt. A cross-correlation analysis is conducted to
flows, except for two factors. It does not require a
find the maximum value of cross-correlation
laser beam or sheet, and it is applied over a much
coefficient Rab. Estimation of sub-pixel movement
greater area than is typical of prior laboratory use
yields the most probable displacement of aij dur-
of PIV. The technique uses an interrogation algo-
ing the period between the images. Once the dis-
rithm developed by Fujita and Komura (1994) and
placement is known, a velocity vector is deter-
extended by Aya et al. (1995). They showed that,
mined. Assembly of velocity information from
for larger scale flow fields--typical of the surficial
interrogation spots enables a composite flow field
flow distributions in rivers conveying floating
to be formed.
debris or ice--debris and foam at areal concentra-
During recent years, PIV has been used as a con-
tions common in rivers were adequate tracers for
veniently fast method for mapping relatively
application of PIV. They also showed that natural
small-scale velocity fields, or subareas of flow
lighting can be adequate for obtaining video
fields, in a variety of small-scale experimental flow
images of suitable quality over the period needed
facilities. It has mainly been used for determining
to determine the velocities.
flow distributions in small areas (nominally 20
The extension of PIV to ice movement raises
20 mm). Extension of PIV for larger flow fields
several questions about the limits of the PIV tech-
poses problems with flow-seeding and illumina-
nique for determining ice velocities: Is the tech-
tion. Recently, however, Fujita and Komura (1994)
nique suitable for all sizes of ice pieces? Can it
and Aya et al. (1995) demonstrated that PIV is
handle mixtures of ice-piece sizes? What areal con-
adaptable to determine distributions of surface
centrations of ice are required? Are there ice speed
velocities for flow in a river channel. Surface-flow
limitations? What frequencies of velocity fluctua-
patterns in most river sites are sufficiently large
tion can be analyzed using PIV? In many respects,
in dimension and low in frequency as to be ame-
the collective answer to most of these questions is
nable to PIV applied in a scaled-up format not
that the field of view contained by an image, and
requiring the high concentrations of seed or tracer
the period between images, must be commensu-
particles needed when PIV is used to determine
rate with the requirements for statistical analysis
intricate flow structures at laboratory scale.
of the interrogation spots. Ice piece size is not an
Using video images from a camera sited on a tower
issue, provided the ice piece is clearly discernible
overlooking the river, Fujita and Komura (1994)
in a video, photograph, or radar image. Mixtures
and Aya et al. (1995) found that drifting debris
of ice piece sizes also are not an issue, because the
(small pieces of wood, weeds, etc.) and foam (from
image-processing method can be modified to suit
naturally occurring humic acids in rivers and
particle densities. The technique is suitable for all
lakes) on water surfaces served as tracers, enabling
areal concentrations of ice. However, if ice is used
PIV interpretation to obtain velocity vectors. Their
as a seed tracer for determining flow patterns, then
successful adaptation of PIV for determining
ice concentration must be such that, at any instant,
velocities of surface currents and drifting debris
each interrogation spot contains at least one ice
prompted the present consideration of PIV for
piece. The issue of ice-speed limitation is best ad-
determining ice velocities.
dressed in terms of the period between images
The main problems that must be faced in
analyzed. Faster ice pieces require shorter inter-
applying the technique to larger areas of flow, as
vals between images, especially if the ice piece tra-
usually required for hydraulic models of river sites
jectory is changing.
and for field application, are the amount and uni-
The PIV technique described here uses and
formity of particle seeding and adequate lighting
extends the PIV technique as adapted by Fujita
needed to create a useable set of gray-level
and Komura (1994) and Aya et al. (1995) for deter-
images. Mapping of a flow field requires an
mining velocities of objects moving on water sur-
adequately large amount of seeding material to
faces. It was used to obtain whole-field velocities
ensure that, at any instant, each interrogation spot
of water and model-ice movement in a hydraulic
in the flow field contains at least one particle. Rea-
model simulation of the confluence of the Missis-
sonably constant illumination is needed to ensure
sippi and Missouri rivers. PIV had not been used
stationarity of the conditions under which video
before to determine velocities in an experimental
images are interrogated. Application of PIV for
setup of the magnitude used in the present study,
10
Previous Page
