Soils from all areas of the US were considered, and those having a wide range of textural, geomorphic, and chemical properites were selected (Figure 1). Slopes were in the range of four to six percent. Priority was given to soils which were of a dominant type, or of agricultural importance for a given region, and also to those soils known to be highly erodible (Alberts et al., 1987). A summary of the soil series and locations is given. Figure 2 shows the position of each of the soils studied on a textural triangle. The number associated with each soil represents the order in which it was studied.
Both Figures 1 and 2 include three soils which were eroded in 1986, identifed by letters. Discussion of the results from these soils can be found in Brenneman (1988) and Elliot et al. (1989a).
Water for each experiment was obtained locally from wells, municipal water supplies, irrigation canals, or neaby ponds or reservoirs. Where necessary, local cooperators or contractors hauled water to the site. Two portable 50 m3 steel-framed pvc folding water tanks held the water until it was pumped to the simulator. The temperature of the eroding water was noted, and a sample was collected for quality analysis.
During the season prior to erosion study, the site crop was corn, wheat, some other row crop or fallow. Excess surface residue was removed, and the sites were deep-tilled and lightly disked three to twelve months prior to erosion studies (Laflen et al., 1987).
Upon arrival, the direction of the rills was determined by orienting the rills to give a desired slope of three to six percent. On slopes greater than six percent, or on soils believed to be highly erodible, slopes were decreased by offsetting the direction of the rills from the main slope direction. In 1987, slopes of the rills were determined by photogrammetric analysis for each individual rill. In 1988, the elevations of the top and bottom of each nine meter long rill plot were measured with a level to calculate individual rill slopes.
A ridging tool mounted on a small tractor formed six rill plots, as well as ridging areas where interrill plots would be installed. Rill plots, each nine meters long, spaced at 460 mm were established with a metal sheet across the top of each plot. Metal collectors and pvc collector pipes to carry the out flow to collecting pits were installed at the down stream end of the rills (Figures 3 and 4).
Some ridges were flattened to construct 500 mm wide and 750 mm long infiltration plots with a slope similar to the natural hill slope. Interrill erosion plots, 500 mm wide and 750 mm long, with sides sloping approximately 50 percent toward a central collecting trough were installed (Figure 4).
Infiltration and interrill plots were established on the outside of the rill plots (Figure 3). On the 1987 soils, two infiltration plots were uncovered, and two were covered to prevent crusting (Liebenow et al., 1989). In 1988, all four infiltration plots were covered. All of the plots were within the wetted diameter of a rotation boom rainfall simulator (Swanson, 1965).
The rainfall simulator was equipped with Veejet 80100 nozzles which were operated at a pressure to give an intensity of about 62 mm hr-1. Rain gages were installed immediately uphill from each infiltration and interrill plot. The total amount of water used was recorded by a water meter.
During the set-up procedure, samples for bulk densities were taken from each rill plot and near two infiltration plots using the compliant cavity technique (barnes et al, 1971). Additional samples of the top soil were collected for future analysis. Soil moisture samples from the surface layer, and a second layer 250 to 300 mm below, were collected near the infiltration plots.
The erosion data collection was divided into three periods:
During the first period, runoff samples were collected from the interrill and infiltration plots every five to ten minutes. Rill runoff observations were made every five minutes from the onset of runoff until runoff equilibrium was reached, when there was no apparent increase in runoff for two subsequent observations. The time required to reach flow equilibrium varied from 30 to 90 minutes.
Once rill flow equilibrium had been reached, maximum velocities of the flow in each rill were measured by injecting fluorescent dye at an upstream marker, and using a stopwatch to note the time required to reach a marker six meters downstream. The rainfall was stopped and rill cross-sectional shapes were recorded with a rillmeter. Soil strength measurements were made.
During period two, rainfall was resumed and the outflow was measured until a flow similar to that observed at the end of period one was reached. Flows were then added at the top of each rill from a pvc manifold with in-line flow restrictors at pressures set to give a desired flow rate. Nominal added flow rates 7, 14, 21, 28, and 35 liters per minute were used. Flow rates were measured by weighing the outflow from each rill, collected in a bucket over a fixed time period. For each rill at each flow addition, two flow rate observations and one velocity measurement were made, and two sediment concentration samples were collected.
During 1987, upon completion of the second period, fall cone penetrometer strengths and observations on rill headcuts were made (Kohl et al., 1988b). For both years, rill shape observations were made using a rill meter. Soil moistrure samples were collected at two depths from selected interrill plots.
For the third period, a final set of flows were added, but this time with no rainfall. The same observations on flows, velocities, and sediment concentrations were made as for period two.
For each flow rate during periods two and three, a set of black and white stereo photographs was taken by two air-photo cameras mounted 10 m apart, 15 m above the site.
Following period three, final strength measurements and rill meter measurements were made. Bulk density measurements were carried out for each rill, and for two infiltration plots. In 1987, the fall cone was not used, but details on rill headcuts were noted.
Outside the wetted circle of the simulator, the soil surface was wetted by filling a 750 mm diameter ring with about 100 mm of water. The water was allowed to infiltrate, and soil strength measurements were carried out on the external plot of varying times once excess water had infiltrated.
A field laboratory, equipped with balances, drying ovens, and a sink, accompanied the field experiments. The runoff and sediment concentrations from the interrill plots, and the sediment concentrations from the rills were found gravimetrically from the one-liter samples in the laboratory. Soil moisture contents were also determined gravimetrically.
All results from field and laboratory measurements were entered into portable computer spreadsheets. Subsequent analyses were done with spreadsheets and Basic programs on personal computers, and on a main frame computer.
It was noted that the erosion rates from interrill erosion varied with time, with the initial erosion rates at low runoffs being near zero. The erosion rates would then increase to either and equilibrium value on clays and sands, or maximum value followed by decreasing rates on many of the silt soils. The mean of the last four erosion rates was selected as the erosion rate for a given plot.
If there was an outlier value within the last four erosion rates, then it was not considered, and the four most consistent of the last five values were used. The plot mean was divided by the slope factor (equation {3}) and the square of the plot rainfall intensity in m2s-2, to calculate Ki for the plot. The means of the eight uncovered plots in 1987, and means of the six ridged plots in 1988 were used to calculate Ki.
To calculate Kr, it was first necessary to calculate the hydraulic radius for each level of rill flow. To find the hydraulic radius, the rill flow "maximum" velocity was observed using a fluorescent dye and recording the length of time required for the front of a slug of dye to travel six meters down the rill. An influorometer was used on four sites to measure the concentration of dye in the rill outflow. The time required from the injection of dye six meters uprill, until the peak concentration of dye was observed by the influorometer, was used to calculate an average velocity. The ratios of maximum velocity to average velocity were found to be 0.551 for sandy soils with relatively wide, shallow rills, and 0.687 for silt soils with narrow deep rills, and clay soils with trapezoidal shaped rills. The measured rill flow rate was then divided by the average velocity to determine an average cross-sectional area of flow.
Cross-sectional shapes of rills were recorded using a rill meter. In 1987, the rill meter was photographed and the location of the pin subsequently noted manually from 35 mm slides. In 1988, the rill meter pin locations were located using an image analyzer. On photographs taken by the air photogrammetric cameras, rill channel flow width were measured to determine the average width of channel for each rill, for each flow rate.
The cross-sectional area and the rill meter pin locations were read by a Basic computer program on PC, and the location of the water surface necessary to give the calculated area was found using a Secant Search Technique. The program then calculated a width of flow and a hydraulic radius (Kohl et al., 1988a). If the width necessary to give the desired area differed more than 10 mm from that observed on the photographs, the rill meter pin locations were modified to effectively alter the rill width, and the program was run again. This procedure was iterated until the width of the rill was the same as that found from the photograph, and the hydraulic radius for a given flow rate was found (Elliot et al, 1989b).
The transport capacity of the flow in the rill was estimated from the relationship (Foster and Meyer, 1972a):
To find the transport coefficient (B), transport coefficients were first estimated for each category in the aggregate size distribution of the eroded sediment and for the range of observed water temperature using Yalin's bedload and transport equation (Yalin, 1963). A weighted average based on the aggregate size distribution and water temperature for each soil was then carried out (Elliot, 1988). The transport coefficients used for each soil are noted on the heading of each soil's respective rill data analysis worksheet.
The transport capacity (Tc), the rill width (wr), and the observed rate of sediment leaving each respective rill (Qs), were then used in Equation {9} calculate the detachment capacity for each flow rate. From this set of detachments and respective hydraulic shears, values for Kr and Tauc were calculated by linear regression. Dc' will then be calculated from Equation {4} and substituted back into Equation {9}. After several iterations, values for Kr and Tauc for each rill for the rain with added flow, and the added flow only conditions were found.
