Dear Doc Hydro: I just ran across an EPA field
manual for surveying wadeable streams that instructs observers to
visually estimate substrate particles sizes and which uses larger
size classes than the phi or 1/2 phi Wentworth classes commonly
used during Wolman pebble counts. I've always had my field crews
measure particles and tally them by size class using the more than
20 size classes of the Wentworth scale recommended by the American
Geophysical Union. Visual estimates and fewer size classes would
certainly speed up field work. What do you think of this idea?
Doc Hydro recognizes that visual estimates, or "ocular assessments,"
have been widely used in the past by biologists to estimate the
sizes of substrate particles. Typically, these estimates placed
particles into broad size classes, for example, "1-to-3-inch
gravel" thereby making it difficult to directly compare this
data with more precise particle size distribution typically collected
by geomorphologists and engineers based on the Wentworth scale.
Despite the continued use of visual estimates, I'm unaware of any
comprehensive published studies demonstrating that visual estimates
are reproducible among different observers.
Since there's no substitute for real data, Doc Hydro decided to
perform a little experiment comparing visual and measurement techniques
on a 100 particle sample of river rocks. Although there are only
5 replications of each technique and statistically this is less
rigorous than one might wish, the results provide valuable insight
into differences among the various techniques.
The 100 particles were sampled along a grid located on a large
river bar using the Wolman method. The round river rocks were placed
into buckets and brought to the lab for easy measurement. In the
lab, Doc Hydro carefully identified the intermediate axis of each
rock and measured the diameter as precisely as possible using a
caliper. The size distribution from this measurement is considered
to represent the "true" size distribution of the particles.
The caliper measured d16, d50, and d84 sizes were 20, 42, and 83
mm, respectively.
Five different observers, ranging from very experienced to inexperienced,
were asked to visually estimate the size of each particle and tally
them into one of the EPA EMAP size classes (table 1). Observers
were provided with a marble and a tennis ball to help them calibrate
their eyes.
Next each observer measured all of the particles using a FISP SA-97
Hand-held Particle-size Analyzer, commonly called a gravel-O-meter
or gravel template, and tallied particles by standard 1/2 phi Wentworth
size classes. The procedure was repeated a third time using a ruler
and particles were again tallied by Wentworth size classes.
Figure 1 shows the plotted particle size distributions for all
of the observers for each of the measurement techniques. Table 2
shows the average d16, d50, and d84 values of the 5 observers obtained
from the plotted particles size distribution curves.
It's apparent from the plots that all of the observers had very
similar results using the gravel-O-meter and rulers since the particle
size distribution curves plot almost directly on top of each other.
In contrast, distribution curves from the visual estimates show
considerable variability among observers.
Table 2 confirms this result with the gravel-O-meter and the ruler
having almost identical d16 and d50 values and little variability
(coefficients of variation zero to 8 percent). In contrast, visual
estimated sizes are smaller and have coefficients of variation ranging
from 18 to 34 percent indicating that different observers obtained
a wide range of different size distribution statistics. Using a
chi-square statistic to compare d16, d50, and d84 values to the
caliper measured true values, only the visually estimated d16 size
of 9.2 mm is statistically different from the others at the 95%
confidence level.
Doc Hydro believes that the major expenditure of any field effort
is getting to the field site. Once there, do the best technical
job possible and take the time needed to do a quality job. Therefore,
pick up the particles and measure them as accurately as you can
and tally them using the standard Wentworth size classes so that
the data are comparable among disciplines. Since different observers
will most likely be collecting your data each year, it's especially
important to reduce observer variability to the maximum extent possible.
Doc Hydro recommends use of a gravel-O-meter because these measurements
are less prone to observer error in identifying the intermediate
axis and produce data that is comparable to sieve measurements.
Table 1. Visual estimate size classes and descriptors. From: Peck,
D.V., J.M. Lazorchak, and D.J. Klemm (editors). Unpublished draft.
Environmental Monitoring and Assessment Program-Surface Waters:
Western Pilot Study Field Operations Manual for Wadeable Streams.
EPA/XXX/X-XX/XXXX. U.S. Environmental
Figure 1. Plots of the particle size distributions for each of
the 5 observers using the gravel-O-meter, a ruler measurement, and
visual estimates. The size distributions from gravel-O-meter and
ruler measurements are almost identical and tend to plot very tightly
on top of each other. The particle size distributions for the visual
estimates show considerably more scatter than the direct measurement
techniques. For example, d50 values of visual estimates range from
31 to 44 mm for the 5 observers while the observers using the ruler
and the gravel-O-meter consistently measured the d50 as 41 and 38
mm, respectively.
Table 2. Average d16, d50, and d84 values from 5 observers using
3 different measurement techniques. Values in parenthesis are the
coefficients of variation of the 5 observers.
|
