Morris & Fan. Reservoir Sedimentation Handbook
Подождите немного. Документ загружается.
FLUVIAL MORPHOLOGY AND SEDIMENT SAMPLING 8.27
heavily instrumented research sites help define the characteristics of bed load transport
processes.
8.7.4 Grain Tracking
The downstream motion of individual stones along a streambed can be traced by marking
the stones in one section of the stream and relocating them repeatedly over time to
determine transport distances and rates. A problem with this method is the difficulty of
finding the stones after they have been moved downstream, and simple marking methods
such as painting makes it impossible to trace particles that become buried. More
sophisticated tracer methods include embedding magnets in stones, which allows
recovery despite burial to as much as 0.6 m with commercial metal detection equipment
(Schick et al., 1987). Smaller material such as sands can be labeled, released, and
recovered at downstream points using a bed load sampler. These methods do not measure
total bed load and provide only semi-quantitative information on the motion of material
in the size classes marked and subsequently recovered. However, observing the
movement of painted stones is a useful method for determining the condition of initiation
of motion in different areas of a streambed.
8.8 SAMPLING OF COARSE BED MATERIAL
The transport rate for bed material is often determined from the bed material grain size
distribution and transport formula, and river sedimentation models require data on bed
material as the basis for transport and armoring computations. The composition of the
bed material is also required for many type of sediment studies relating to bed stability,
initiation of motion, scour, streambed degradation below dams and environmental
impact. This section outlines methods to sample the coarse bed material in sand- and
gravel-bed rivers.
8.8.1 Sampling Sand Beds
Sand beds are sampled by obtaining bulk samples which are returned to a laboratory, split
as required, and sieved in accordance with procedures outlined in Chap. 5. Sediments in
sand bed streams can be sampled by hand at low water when bars are exposed or at
shallow depth. In relatively shallow and low-velocity rivers (less than 0.5 m/s) with little
or no gravel, a Ponar-type dredge or similar device may be used from a boat. The Ponar
dredge (Fig. 8.10b) is similar to an Eckman dredge, but is heavier and much better suited
to collecting surface sediments sand size or smaller. Where the current is appreciable a
bed material sampler based on the rotating bucket principle (Fig. 8.106) may be used,
which will scoop a 7-cm-wide sample to a maximum depth of about 5 cm.
8.8.2 The Sampling Problem in Gravel-Bed Streams
The sampling procedures used in sand-bed streams are generally unsuitable for gravels
and larger material. Conditions in gravel-bed rivers present special sampling difficulties,
which have been reviewed by Church et al. (1987). There is no sampling equipment
FLUVIAL MORPHOLOGY AND SEDIMENT SAMPLING 8.28
for the collection of gravel or coarser material beyond wading depth (other than use of
heavy construction equipment), so samples are normally collected from exposed bars
during periods of low water. It is often not practical to bulk sample cobbles or larger bed
material because the stones are too large to be carried back to the laboratory, or to be
sieved and weighed in the field. Stones larger than 180 mm are difficult to handle and
weigh.
Gravel beds generally have wide size ranges, often exceeding 10 phi () units [Eq.
(5.1)]. The size distribution may he bimodal, characterized by separate peaks in sand and
gravel ranges separated by few grains in the 1- to 4-mm size range (Fig. 8.5). Gravel-bed
rivers can also exhibit large spatial variations in grain size. The coarsest material occurs
along the leading edge of bars, in riffles, and along the line of the main current; smaller
material accumulates toward the trailing edge of bars, in pools, and at progressively
higher elevations on point bars. River beds containing significant gravels are typically
armored, resulting in a surface layer about one grain thick which is much coarser than the
remainder of the deposit, or the subarmor. Sand may he absent from the bed surface
while constituting over half the material beneath the armor layer.
Because of the large variation in grain size, a single method cannot be used for sizing
bed material in gravel-bed rivers. Fines (smaller than 0.062 mm), if present, are
determined by settling tests. Sands and gravels up to about 16 mm arc normally sized
with laboratory selves. Stones can be measured with specially constructed sieves or a
hand-held sizing template, but more frequently their diameter is measured along the
intermediate b axis. Axis measurement is the only practical method for sizing stones
larger than 180 mm. Different grain size distributions are produced by the different
methods for selecting and determining the grain size frequency of the sampled stones.
Conversion factors to render equivalent the samples collected and analyzed by different
techniques are presented in Sec. 9 of this chapter.
8.8.3 Selection of Sampling Areas
Because the grain size in gravel-bed rivers exhibits large variations from one point to
another, selection of appropriate sampling sites is a critical aspect of a sampling program.
There are three dimensions to the problem of sample site selection. Depth: is the armor or
the subarmor to be sampled? Longitudinal: sample in riffles or pools. at bends or
crossings? Lateral: sample at midchannel or closer to the banks?
Sampling location depends in part on the purpose of the sampling. If information is
needed on the initiation of motion or surface roughness. the armor layer should be
sampled. If bed material transport is to be computed for discharges large enough to
mobilize the armor layer, or if a grain size must be specified in a sediment transport
model, then both armor and subarmor should be sampled to define the grain size subject
to transport. For environmental and fisheries studies, the finer portion of the bed is
typically of great interest, especially with respect to the suitability of spawning gravels
because even a relatively small increase in material smaller than 0.85 mm can have a
large impact on gravel permeability and the oxygenation of eggs. To analyze impacts of
releases for gravel-flushing or channel maintenance, it may he important to have
sampling locations representative of different aquatic habitats (e.g., both riffles and
pools).
Two patterns will normally he apparent in the longitudinal profile of a gravel-bed
river. First, the profile can usually be divided into a series of steps characterized by
shallow fast-flowing riffles separated by deeper pools of slower-moving water. Finer
sediment tends to accumulate within the pools and is not representative of the bed as a
whole or the larger grains that control streambed morphology. Sampling should typically
FLUVIAL MORPHOLOGY AND SEDIMENT SAMPLING 8.29
be focused in riffles, not pools. In bars, the upstream edge will normally contain coarser
FLUVIAL MORPHOLOGY AND SEDIMENT SAMPLING 8.30
material than the downstream edge. Lateral variation in bed material grain size can also
be pronounced, especially on point bars. Material toward the center of the stream can be
much coarser than material closer to the shoreline. Sampling should normally focus on
the area of coarser sediment, avoiding the areas where finer material is deposited because
of a localized reduction in hydraulic energy. Consistent criteria must be used for selection
of sample locations along the entire reach sampled. Otherwise, the sampled grain size
will reflect variations introduced by inconsistent sampling criteria rather than the true
longitudinal variation along the stream caused by variations in hydraulic conditions and
sources of sediment supply.
Bed material sampling should be timed to coincide with low water so that the bed of
the stream will be exposed to the greatest extent possible. If sampling is performed
during periods of higher water, the sample sites will be limited to the margins of the
stream where the bed material is finer, and sample results may not be representative.
There is no suitable method to sample gravels in water more than about knee-deep, since
portable sampling equipment will not penetrate the stone bed.
8.8.4 Selection of Sampled Stones
Once the sample areas have been selected, there are three alternatives for selecting the
group of stones to be analyzed.
1. Bulk sampling. Bulk sampling involves collection of a volume of bed material
using a cylindrical sampler (Fig. 18.19b), a shovel, backhoe, etc. Samples from
submerged deposits should be collected in an enclosed container to prevent fines from
washing out.
2. Grid sampling. The grid or transect method uses some variant of the pebble count
technique described by Wolman (1954). The selection of individual stones from within a
sampling area for inclusion in the size distribution is randomized so that the selection
probability depends on the exposed surface area of the stone. The grid method includes
the selection of each stone at the intersection of a square grid set out using string or wire,
linear transects with sampling stations at regular intervals, or a system to randomly select
each sample stone such as walking and sampling the grain at the point of each toe, at the
point of a tossed stick, etc. Normally the sampled points should be at least two stone
diameters apart. However, if the grid is small enough that two grid points fall on the same
stone, that stone should be counted twice. If a grid point falls on fines (e.g., smaller than
8 mm), it should be noted as "fines," and a point falling exactly in the crevice between
two stones should be discarded. One variant of this method, which has the advantage of
also working in shallow water, is to bend over at each sampling point, eyes closed or
averted, and select the first stone touched (Leopold, 1970).
3. Areal sampling. This method is based on the selection of all surface stones within
a perimeter (e.g., 1 m
2
). Another areal-based selection method involves spray painting a
portion of the exposed bar and selecting all the painted stones. A third areal sampling
method involves sizing of stones by using vertical photographs of the bed, including a
scale in each photograph.
Bulk sampling can be used to sample the entire bed including both armor and
subarmor, or it can be used to sample the subarmor only after the armor layer has been
removed by hand. Bulk sampling is not well-suited for sampling armor because the
irregular surface layer one grain thick does not represent a well-defined volume. The
remaining techniques are suitable for sampling the surface layer only. All techniques are
oriented to randomizing sample collection within a preselected sampling area. However,
FLUVIAL MORPHOLOGY AND SEDIMENT SAMPLING 8.31
the sampling area itself cannot be selected randomly; selection must be based on the
stream morphology to ensure that sampling areas are representative.
A variety of more specialized techniques are available for sampling the bed for
specific needs, such as studies examining spawning gravels. For instance, bulk samples
which preserve the bedding patterns can be obtained by freeze coring, injecting liquid
carbon dioxide or liquid nitrogen into the bed through one or more standpipes and
recovering the solid block of frozen bed material that forms around the standpipe
(Walkotten, 1976).
8.8.5 Measurement of Stone Sizes
Bulk samplers are normally size-classed by sieving. However, when a sample contains
grains too large to sieve, the oversize stones can be measured individually. Each stone
may be oriented so that it has three axis: a = long, b = intermediate, and c = short. One
common size classification method is to measure the length of the intermediate or b axis
of each stone with a ruler or calipers, treating this as the nominal or sieve diameter for
assigning each stone to a ½ phi (½) size class. As an alternative, a square mesh or sizing
template cut into metal may be used to size stones by ½ size classes, similar to sieving.
Either method yields similar results. In the case of photographs, because stones usually
lay flat with their shortest axis oriented vertically, the b axis corresponds to the shorter of
the two visible dimensions.
8.8.6 Frequency by Size Class
Once sorted by size the percentage distribution of stones within each size class can be
determined by either counting or weighing the stones:
Count. The number of stones in each size class is counted. This represents the most
rapid way to tabulate and record field data.
Weight. The weight of the stones in each size class is determined with a scale in the
field or laboratory, or by computing weights from the b-axis dimension by assuming
that the b axis is equivalent to the nominal diameter and the volume V of the
equivalent sphere is computed by V =
4
/
3
r
3
The volumes are converted to weight
by multiplying by specific weight (e.g., 2.65 g/cm
3
). Although there are variations
between individual stones, over a sample population use of the b axis to estimate the
weight of stones gives results very similar to direct weighing (Church et al., 1987).
In areas having unusual (e.g., flat or elongated) gravel shapes, it is recommended that
this conversion be checked by using local samples. The size to weight conversion
can also be computed by multiplying the count within each class interval by the
mean stone weight in each class interval. However, stones in the largest class should
he weighed individually.
The number of stones counted within a size class interval must be multiplied by the
average weight of stones for that class interval to obtain the total weights by class. Table
8.2 gives
½
. size classes and the corresponding diameters and volumes for a sphere
representing the mean weight in each class. The diameter of the stone representing the
average weight of all stones in the class interval can be estimated very closely by
FLUVIAL MORPHOLOGY AND SEDIMENT SAMPLING 8.32
Mean diameter
d
2
u
d
2
L
2
1/2
(8.4)
FLUVIAL MORPHOLOGY AND SEDIMENT SAMPLING 8.33
TABLE 8.2 Mean Grain Sizes for ½ phi (½) Size Classes, Based on Weight.
Class limits
Mm
Mean diameter,
mm
Mean volume
cm
3
181 to 256 -7.5 to -8.0 220.6 5621
128 to 181 -7.0 to -7.5 156.0 1988
90.5 to 128 -6.5 to -7.0 110.3 703
64 to 90.5 -6.0 to -6.5 78.0 248
45.2 to 64 -5.5 to -6.0 55.1 27.6
32 to 45.2 -5.0 to -5.5 39.0 31.0
22.6 to 32 -4.5 to -5.0 27.6 11.0
16 to 22.6 -4.0 to -4.5 19.5 3.88
11.3 to 16 -3.5 to -4.0 13.8 1.37
8 to 11.3 -3.0 to -3.5 9.74 0.484
Note:
= -log
2
dia (mm) = -3.3219 log
10
dia (mm).
where d
u
and d
L
, are the upper and lower grain diameters in the class interval. If stone
diameters or weights are uniformly distributed across a class interval, this equation
provides a better estimate of the mean stone volume (or weight) than use of either the
arithmetic or geometric mean of grain diameters d
u
and d
L.
8.8.7 Number of Stones Sampled
How many stones must be included in a sample from a gravel-bed stream for it to be
representative? Wolman (1954) recommended that at least 100 grains be included, but
indicated that a minimum of 60 also gave reliable results. More recent research reviewed
by Church et al. (1987) suggests that as few as 50 may be adequate. In the areal method
in which all stones are counted, the sampling area should be adjusted for the size of the
material to be sampled to obtain the requisite number of stones without excessively large
counts. Because the size of bed material varies from point to point within a stream, given
a limited sampling effort a better overall picture of the bed material will probably be
obtained by collecting smaller samples from more locations as opposed to larger samples
from fewer locations.
8.8.8 Presentation of Grain Size Results
Either count or weight frequency data may be arranged into histograms or, more
commonly, cumulative distribution functions. Both types are illustrated in Fig. 8.5.
8.8.9 Truncated Samples
Samples from gravel beds may be truncated through either accident or design. A tenfold
change in diameter produces a thousandfold change in the weight of individual grains,
and, in gravel beds having a wide range of grain sizes, the smaller particles may
constitute an insignificant part of the sample by weight, especially in the surface layer.
FLUVIAL MORPHOLOGY AND SEDIMENT SAMPLING 8.34
Gravel beds may also have a few stones significantly larger than others, which may not
be included in normal sampling. Thus, the sample may be truncated at either or both
ends. This truncation may be unimportant for modeling bed material transport; the largest
stones are infrequent and the smaller may act as wash load and insignificantly affect bed
configuration. However, truncation limits and the presence of unsampled material,
especially fines, should be indicated in field notes. Although sand or fines may form an
insignificant part of the bed from the standpoint of transport hydraulics, their presence
may be critical from the standpoint of environmental analysis (e.g., spawning gravels,
transport of contaminants associated with fines), and the quantification and modeling of
their behavior will be critical in some applications.
8.8.10 Use of Photographs
Vertical photographs of bed material which include a scale are invaluable in documenting
the condition of the bed, and methods for quantifying grain size from photographs have
been described by Adams (1979). Vertical photographs can be taken using a handheld 35-
mm camera and placing a scale bar with ½ increments in each photo. This method has
the advantage of allowing many sites to be sampled rapidly with highly portable
equipment, and provides a permanent visual record of the data collection sites.
Photographs may be taken at many sites for comparison or archival purposes, and grain
size computed for only a limited number of these sites. The photographs to be analyzed
are enlarged to 200 x 250 mm (8 × 10 in) and overlaid with a transparent grid which is
used to select stones to construct a distribution curve by the grid-count method described
in the next section. Stones are sized by using the sampling bar contained in the photo to
measure the shortest visible axis (the b axis, since the stone will normally lie with the
shortest c axis oriented vertically). As a disadvantage, the photographic method does not
measure the true b-axis length because the b axis will not always be parallel to the bed
(imbrication angle) and because of partial hiding of the stone and shadow effects. Church
et. al. (1987) concluded that photographic methods for estimating grain size may he
adequate for gravel and large material where high accuracy of absolute lengths is not
required for size classification. However, photographic measurements should be checked
by conventional field techniques in each study area, since packing and imbrication angle
can change from one environment to another. Adams (1979) reported that the sizes
measured by the photographic method are biased, being about 0.1 , smaller than the
equivalent sieve size. The grain size frequency distribution curve for sieve and
photographic sizing are of similar shape, but displaced by the amount of bias.
8.9 BED MATERIAL GRAIN SIZE CONVERSION FACTORS
Hydraulic transport formulas have been based largely on laboratory work in which bulk
samples are divided into size classes by sieving and weighing. It is desirable to have
gravel-bed samples reported in a manner that is consistent with the bulk-sieve-weight
procedure used in laboratory flumes and for smaller grain sizes. Different sample
collection and frequency methods will produce different frequency distributions
(Leopold, 1970). A system of correction factors that can be applied to different sampling
techniques was presented by Kellerhals and Bray (1971) and was subjected to rigorous
scrutiny by Church et al. (1987) with both laboratory and field techniques. The latter
concluded that the Kellerhals-Bray corrections "yield reasonable results" and
recommended their use.
FLUVIAL MORPHOLOGY AND SEDIMENT SAMPLING 8.35
FIGURE 8.16 Cube model for developing sampling conversion factors (after Kellerhals an
d
B
ray, 1971).
The Kellerhals-Bray correction scheme was constructed by visualizing the bed as filled
with a series of cubes. The cubical model was used only to simplify computations: the
method is independent of the exact particle shape. The sample cube is illustrated in Fig.
8.16 along with a tabulation of key geometric features. The conceptual model contains
three sizes of stones, and the total weight within each size class is the same (as shown in
column e of the figure) resulting in 33.3 percent of the total sample weight in each size
class.
The conversion factors were developed for three types of sampling methods (bulk.
grid, and areal) and two types of frequency measurements (weight and count), all
described in the previous section. If a grid is placed over the surface, or a vertical
photograph of the surface, and each stone falling beneath a grid intersection is assigned to
a size class, the resulting frequency count should be directly proportional to the surface
area of the exposed stones (col. g in Fig. 8.16). This grid-count method will produce a
grain size distribution identical to a bulk-sieve-weight sample. All other methods require
conversion factors to produce a size distribution similar to that which would have been
obtained by using the bulk-sieve-weight technique. The conversions as a function of
grain (or class) diameter are given in Table 8.3 for the sampling method (bulk, grid,
areal) and the frequency measurement (weight or count). Two examples will he given to
illustrate use of the conversion factor.
FLUVIAL MORPHOLOGY AND SEDIMENT SAMPLING 8.36
TABLE 8.3 Conversion Factors for Surface Samples
Bulk-sieve-
weight
Grid-count Gird-weight Areal-count Areal-weight
Bulk-sieve-weight - 1
d
3
1/d
2
d
Grid count 1 -
d
3
1/d
2
d
Grid weight
1/d
3
1/d
3
-
1/d
5
1/d
3
Areal count
d
2
d
2
d
5
-
d
3
Areal weight 1/d 1/d
d
2
1/d
3
-
Source: KeIlerhals and Bray (1971)
8.9.1 Areal Count to Bulk Sieve Conversion
For this example consider the cube introduced in Fig. 8.16 and assume that all the surface
stones within a perimeter have been sampled and size-classed, and the number within
each size class has been counted (areal-count method). In Table 8.4, the first two columns
represent the raw field data: counts within each size class. The counts are classified by
the smallest limit of the class interval, equivalent to reporting sieved grain sizes by the
"retained on
mesh size. From these data the values in column 3 are computed, but these
percentages will not produce a frequency distribution equivalent to the bulk-sieve-weight
method. It is necessary to apply a correction factor of d
2
from Table 8.3. the value of
which is computed in column 4. Since the grain size distribution is based on a percentage
of the total sample weight or mass, it is not necessary to determine the weight of the
stones, only the conversion factors. Applying the column 4 correction factor to the
percentages in column 3 yields the corrected numerical values in column 5, from which
the percentage distribution in column 6 and the cumulative distribution in column 7 can
be computed. In this conversion, the length d refers to the diameter corresponding to the
mean weight of that class interval [Eq. (8.4)].
A second example involves a weight conversion using one of the Kellerhals and Bray
field datasets, as summarized in Table 8.5. In this case, all stones within a perimeter were
sampled and weighed (areal-weight method). To convert this to either the bulk weight or
the grid count format, which are identical, the conversion factor is 1/d. The first two
columns in Table 8.5 are obtained from a grain size distribution curve, from which
columns 3 and 4 are computed. The weight (column 3) is multiplied by the l/d conversion
factor (using the diameter in column 4) to produce the corrected values of W/d in column
5. The numerical values in column 5 are used to compute the percentage distribution by
TABLE 8.4 Areal Count to Bulk-Sieve-Weight Conversion Based On Fig. 8.16
Site d
Areal
count C
Percent
of count
d
2
Converted
counts Cd
2
Percent
by class
Percent
finer
(1) (2) (3) (4) (5) (6) (7)
8 100.0
4 12 4.8 16 0.76 33.3 66.7
2
48 19.0 4 0.76 33.3 33.3
1 192 76.2 1 0.76 33.3 0.0
Sum 252 2.29