Prior to each experiment the FT bin was encased with 5-cm-thick insulation board and a freeze
plate was placed on the soil surface to initiate freezing from the top, as in nature. The FT bin was frozen
once to its full depth, and then thawed. Both the C and FT bins were sealed to minimize changes in soil
moisture while the FT cycle progressed. The FT and C bins were then placed side-by-side in the CRREL
soil-erosion simulator and elevated to the same slope. Equal incoming tap-water flows dropped about 3
cm from the flow inlet to the rill surface of each bin, flowed down the rills, and exited the bins via broad
V-shaped weirs. At the start of each experiment the applied overland flow distributed uniformly across
the width of these initial rills. The experimental setup is pictured in Figure 1.
Experiments were conducted in three series, each having a specific soil moisture range, 1618%
(low), 2730% (mid), or 3740% (high, saturated) by volume. Each series included six experiments at
three nominal flow rates: 0.4, 1.2, and 2.4 L/min, and two slopes, 8 and 15. The soil weight (kg) and
pretest soil moisture (% by volume) of each bin are given in Table 1 by test series as mean standard
deviation. Comparisons between FT and C bin data reveal similar means and standard deviations for all
series, with soil weight and moisture content increasing together, reflected in the mean bulk densities of
each series. Freezing times given in Table 1 increased significantly with soil moisture because of the
greater heat content. Thaw times were typically just over 24 hours.
We measured many parameters prior to, during, and following each experiment to quantify the
erosion occurring in each bin. Included in these measurements were the cross-sectional geometry of the
imposed rectangular rill at two locations before an experiment, sediment losses and flow discharge
through time during, and eroded rill cross sections after. Sediment load gave an integrated measure of
erosion through time, while cross-sectional data provided measures of erosion at specific locations. Both
types of data were obtained to provide independent evidence, allowing us to assess the concurrence of
trends, and yielding a stronger basis for conclusions. We measured rill cross sections and maximum rill
widths and depths using a sliding pin meter to obtain horizontal distance from the bin wall and depth from
the bin top to each point. Enough points were measured to adequately define rill cross-sectional shape.
Figure 2 gives sample pre- and post-experiment cross sections for both bins. Maximum rill widths and
depths were acquired at nine longitudinal locations spaced 0.1 of the bin length L apart.
We measured sediment losses and flow discharge by collecting timed samples throughout the
experiment. Samples of 250 mL were collected at 1, 2, 3, 5, 7, and 10 minutes from the start of an
experiment, and 500-mL samples followed at 12, 15, 20, and every 5 minutes thereafter until the end of
the experiment. The sampling intervals increased as the experiment progressed and the rate of erosion
generally slowed. The fill times of the known volume provided discharge measurements, and filtered
samples quantified total sediment in transport. We used a cm-rule scaled in 10ths of a cm to measure
groundwater depths in each well before, many times during, and again after each experiment. Pins were
inserted at nine locations spaced along the rill to mark the waterline at the end of each experiment. We
then measured the location and elevation of each pin, and calculated the overall water surface slope.
During each experiment we visually observed and described the FT and C rill channels as they developed,
and recorded these observations. In addition, photographs documented rill development at a two-minute
interval from a position 1.6 m directly above the bins.
4
Previous Page
