INTERRILL ERODIBILITY: COLLECTION AND ...

2 downloads 0 Views 1MB Size Report
Meyer concluded that for soils with a high clay content, this relationship does not ... (1981) and Meyer and Harmon (1984) ignored the effect of ..... St. Joseph, MI:.
INTERRILL ERODIBILITY: COLLECTION AND ANALYSIS OF DATA FROM CROPLAND SOILS A. M. Liebenow, W.J.Elliot, J. M. Laflen, MEMBER

MEMBER

Assoc. MEMBER

ASAE

ASAE

ASAE

ASAE

ABSTRACT Interrill erodibility was measured on eighteen cropland soils in the western half of the United States, using a rotating boom rainfall simulator. There were significant differences in infiltration rates, runoff rates, and erosion rates among the soils. A slope factor, based on plot slope, to account for differences observed in interrill erosion rates, is proposed. The data show that this slope factor may also be a function of soil properties. An interrill erodibility coefficient is proposed for each soil. INTRODUCTION ncluded in provisions of the Food Security Act of 1985 were mandates for the implementation of soil conservation practices by farmers as an eligibility requirement for federal farm subsidies. These mandates necessitate the use of an equitable method of predicting soil erosion on a farm. For over a quarter of a century, the Universal Soil Loss Equation (USLE) has been used for predicting soil erosion by water. The empirically based USLE has limitations (Watson and Laflen, 1986) which can be overcome by recent advances in computer technology (Lane et al., 1987). The potential existed to model the hydrologic and erosion processes, and to more accurately predict soil erosion under a much broader range of conditions. The USDAWater Erosion Prediction Project (WEPP) was initiated to exploit this potential. Its purpose was "to develop new generation water erosion prediction technology for use by . . . organizations involved in soil and water conservation and environmental planning and assessment" (Foster, 1987). This technology is manifested in a computer program predicting soil loss rates. The program will be distributed nationally for use in county and state Soil

I

Article was submitted for publication in May 1990; reviewed and approved for publication by the Soil and Water Div. of ASAE in August 1990. Contribution from the Ohio Agricultural Research and Development Center and the USDA-Agricultural Research Service, National Soil Erosion Research Laboratory. This research was planned, funded, and carried out by the USDA-Agricultural Research Service from Iowa State University. This article was originally submitted for the ASAE National Student Paper Competition. The authors are A. M. Liebenow, Industrial Representative, Interstate Power Co., Mason City, lA (former Research Associate, The Ohio State University, Columbus); V^^. J. Elliot, Assistant Professor, The Ohio State University, Columbus; J. M. Laflen, Research Leader, USDAAgricultural Research Service, National Soil Erosion Research Laboratory, Purdue University, West Lafayette, IN; and K. D. Kohl, Soil and Water Engineering Specialist, North West Area Office, Iowa State

University Cooperative Extension Service, Ames. 1882

K.D.Kohl

Assoc. MEMBER

Conservation Service offices and in federal and state research programs. The WEPP computer model uses topographic details, climatological statistics, crop growth parameters, hydrologic simulation, and a database of soil properties to predict erosion rates. In addition, algorithms relating infiltration and rill and interrill erodibility to soil properties are required (Lane et al., 1987). One element of the WEPP research focused on measuring soil erodibility and infiltration by using simulated rainfall on field test plots. The observed erodibility and infiltration could then be related to measurable soil properties by the regression equations (Foster, 1987). The overall goal of the USDA-ARS WEPP cropland research program was to measure the erodibility of a broad range of cropland soils, and relate that erodibility to measurable soil properties. This article reports the results of the interrill erodibility analysis on 18 of those soils and presents a method for accounting for slope in the interrill erosion process.

REVffiw OF LITERATURE Soil erosion is "a process of detachment and transportation of soil material by erosive agents" (Ellison, 1947). As early as 1917, scientists noted that erosion by flowing water could be divided into two major components. Lehmann (1917) referred to these two components as "sheet washing" and "gullying". As late as 1932, references to sheet washing and gullying as the two major components of erosion by flowing water could still be found (Jones, 1932). Within the last 20 years, these two forms, detachment and transportation by flowing water and detachment and transportation by raindrop impact, have become known as "rill" and "interrill" erosion, respectively (Foster and Meyer, 1972). Not until the late 1960s, however, was the computer technology developed to treat these components separately. Interrill erosion is principally caused by raindrop impact (Meyer and Harmon, 1984). When a raindrop hits the soil, it dissipates the energy it possesses to the soil particles it contacts, causing the soil particles to detach, and freeing them to be carried away by surface runoff. The mass of particles eroded depends upon rainfall properties, surface properties, and soil properties (Watson and Laflen, 1986). Intensity, a rainfall property, has been determined to affect the interrill erosion rate by approximately a power of two (Meyer, 1981; Watson and Laflen, 1986). Meyer concluded that for soils with a high clay content, this relationship does not hold (Meyer, 1981). However, Watson and Laflen (1986) refuted this claim and attributed

© 1990 American Society of Agricultural Engineers 0001-2351 / 9 0 / 3306-1882

TRANSACnONS OF THE ASAE

the variation in intensity exponents to factors other than clay content. Slope, a surface property, may also affect the interrill erosion rate. The Danish Hydraulic Institute (1986) has determined that for a slope of less than 5%, the effect of slope is not important to interrill erosion rate. Meyer (1981) and Meyer and Harmon (1984) ignored the effect of slope in computing interrill erosion rate during experiments to study effects of rainfall intensity on erosion. Lattanzi et al. (1974) found that an expected 20-fold increase in soil loss from interrill areas when slope was increased from 2% to 20%, was only about doubled. They also noted that runoff was independent of slope. The different physical, chemical, and mineralogical properties of soils affect the interrill erosion rate (Foster, 1982). The interrill erodibility of a soil is the susceptibility of that soil to interrill erosion. Soil properties which affect erodibility include primary particle size distribution, organic matter content, soil structure, content of iron and aluminum oxides, electro-chemical bonds, initial moisture content, and aging. Several forms of the equation to describe interrill erosion have been proposed. Three forms are: D. = KjfS^ (Meyer, 1981)

(1)

D. = K. ^ (Foster, 1982; Alberts et al., 1988) (2) TPOI D. = K.rS^ (Watson and Laflen, 1986)

(3)

where the interrill erosion rate, the interrill erodibility; p and q are regression coefficients, I = the intensity of rainfall, Sf = a slope factor, and S = the slope. In equation 3, the slope exponent, q, ranged in value from 0.16 to 0.47 for the three soils examined (Watson and Laflen, 1986).

EXPERIMENTAL METHOD The experiment involved applying simulated rainfall to small ridged and flat plots on a wide variety of soils located in the western United States (Table 1). As part of the USDA-ARS WEPP project, these soils were selected from a nationwide soil set that represented the broad range of soils expected on U.S. cropland (Alberts et al., 1987). Plots were located on land that had previously been in com, sorghum, or a small grain. After harvest, all surface residue was removed by raking or buming. The plot was moldboard plowed to a depth of 0.2 m (or to the bottom of the A horizon if it was less than 0.2 m in depth). Shortly before simulation, the plot was disked. Then, ridges 0.5 m on center, were formed using a tractor-mounted ridging tool. Ten small interrill plots, each measuring 0.75 m x 0.5 m, were formed (fig. 1). Six of these interrill plots were ridged plots (fig. 2), the remaining four plots, formed by smoothing ridged plots, were referred to as flat plots (fig. 3). Plot borders were galvanized steel sheets driven into the soil to prevent water movement to and from the plot. Soil outside the plot was sloped to prevent surface water accumulation. VOL. 33(6): NOVEMBER-DECEMBER 1990

TABLE 1. Name, classification, type, and location of soils studied No Soil name 1 2 3 4 5 6 7 8 9 10

Sharpsburg Hersh Keith Amaiillo Woodward Heiden Whitney Academy Los Banos Portneuf

11 12 13 14 15 16 17 18

Nansene Palouse Zahl Pierre Williams Barnes Sverdrup Barnes

Location

Classification

Soil type

Typic argiudoll Typic ustochrept Aridic argiustoll Aridic paleustalf Typic ustochrept Udic chromustert MoUic haploxeralf MoUic haploxeralf Typic haploxeroU DurixeroUic calciorthid Pachic haploxeroll Ultic haploxeroll Entic haploboroll Typic torrert Typic argiboroll Pachic argiboroll Udic haploboroll Udic haploboroll

Silty clay Lincoln, NE Sandy loam Ord,NE Silt loam Albin,WY Loamy sand Big Spring, TX Silt loam Buffalo, OK Clay Waco, TX Sandy loam Fresno, CA Loam Fresno, CA Clay Los Banos, CA Silt loam Twin Falls, ID Silt loam Silt loam Loam Clay Loam Loam Sandy loam Loam

Pullman, WA Pullman, WA Bainville, MT WaU,SD McClusky, ND Goodrich, ND Morris, MN Morris, MN

The soil in the ridged plots was left undisturbed after the plot borders were installed. Two flat plots were covered with a porous covering (burlap or fumace filter) to prevent crusting while the other two flat plots were left exposed. The results of the exposed plots only are considered in this article. Rocks over 50 mm in diameter and vegetation were removed from the plots during construction. A rotating boom rainfall simulator applied water from wells, municipal water supplies, or open reservoirs, at a rate of approximately 63 mm/h for about 1 h. Rain gauges at the upper end of each interrill plot were used to measure rainfall amount. Initial runoff sample collection bottles were put in place as soon as the ground became damp. Ten to twelve samples were collected from each plot at five to seven minute intervals. The times of the start and finish of each collection period were recorded. The sample bottles were weighed, dried, and reweighed to determine the mass of sediment and water in the samples. Water temperature was measured at each site, and a sample of water was collected for further chemical analysis (Elliot et al., 1988). Flow rates, erosion rates, and runoff rates were computed based on the mass of sediment or water, the time

Rill Collection Pits

Figure 1-Plan view of erosion experiment site. 1883

Rain Gauge

RESULTS AND DISCUSSION RAINFALL INTENSITY

0.75 m

INFILTRATION

Plan

Figure 2~Diagrain of ridged plot.

over which they were collected, and the contributing area. Infiltration was calculated as the difference between rainfall intensity and runoff rate for each sample. Runoff and erosion rates started at zero, and increased to an equilibrium value, or to a maximum erosion rate followed by decreasing erosion rates. The interrill erosion rate and equilibrium infiltration rate for a plot were assumed to be the means of the last four samples, except where one sample differed markedly from the others. The raw data are reported in Elliot et al. (1989b). The ridge side slope was used as the slope in the analysis for ridged plots, while the slope of an interrill flat plot was assumed to be the same as that measured along an adjacent rill bottom (figs. 2 and 3). For 12 soils, photogrammetric analysis provided cross-sections for calculating the ridge-side slope on a ridged plot on each side of the simulator. It was assumed that otherridgedplots on the same side of the simulator had the same slopes. On the six soils that had no photogrammetric data,ridgedplots were assumed to have slopes similar to those measured on other soils with similar classifications. Rat plots had slopes of 3 to 6%, andridgedplots around 50%. The results were analyzed using analyses of variance (SAS Institute, 1982) to see if the effects of soil type, side of the simulator and type of plot on rainfall intensity, infiltration, interrill erosion rate, and interrill erodibility were significant. A difference was considered significant if the significance probability (Pr > F) was 0.05 or less. Standard errors for treatments and means were calculated (Cochran and Cox, 1957). Rain Gauge

Slope 0.75 m

0.5 m

Isometric

Plan

Figure 3-Diagram offlatplot. 1884

Table 2 gives the means of the rainfall intensity for the left andrightside of the simulator, the flat andridgedplots, and the overall mean for each soil. The variation in rainfall intensity from site to site was significant, likely because of adjustments to the simulator, wind direction, and precipitation (soil 4) during the experiment. There were no significant differences in mean rainfall intensity between the left and right sides of the simulator, or between plot types. Mean infiltration rates are shown in Table 3. There were significant differences among soils. These differences can be used to calculate infiltration parameters based on soil properties (Rawls et al., 1983). Plot position apparently had little effect on infiltration rate. The infiltration rate on flat plots was significantly greater than onridgedplots, but there was also a significant interaction between slope and soil type. Most soils with higher infiltration on flat plots than on ridged plots had lower interrill erodibility (fig. 4). It may be that for highly erodible soils, sediment partially filled the macropores on flat plots and reduced infiltration; but, on ridged plots, the sediment was transported from the plot. On soils with low interrill erodibility, however, less sediment was detached, so less was deposited on the flat plots. With less deposition, macropores would be similar on flat andridgedplots, so as suggested by the Philip infiltration equation (Skaggs et al., 1969), infiltration would be greater on flat plots than on ridged plots because of the greater depth of overland flow. Further hydrologic study is necessary to clarify this matter.

TABLE 2. Mean rainfall intensity (mm h-^) Soil No. of plots

Left* Right

Hatt Ridged Mean$

3

5

2

6

8

58 65 60 64 60 57 64 59 53 56 66 59 64 63 68 75 61 78

62 58 74 59 68 63 62 61 65 57 62 59 61 63 57 74 62 72

65 64 63 58 63 63 67 62 58 64 65 58 67 68 54 79 61 75

58 63 66 64 63 58 62 59 58 54 64 59 61 62 67 74 61 76

60 63 65 62 63 59 63 60 58 56 64 59 63 63' 63 75 61 76

Mean 63 63 Mean S.E. 1.08

64

63

63

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

1.13

Plots located on left or right side of rotating boom rainfall simulator, S.E. = 4.01. Flat plots with eroding slopes similar to site, 46%;ridgedplots with eroding side slopes of 50-60%, S.E. = 4.81. S.E.=2.74. TRANSACTIQNS OF THE ASAE

TABLE 3. Mean infiltration rate (mm ti-^) Hatt Ridged Meant

Soil No. of plots

Left*

Right

3

5

2

6

8

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

19.96 22.61 10.15 21.44 23.83 6.52 28.69 15.34 15.92 25.75 24.66 16.17 14.22 10.35 20.22 31.40 24.88 32.15

19.21 21.48 15.10 19.69 21.72 17.43 19.07 13.65 14.39 22.50 21.67 8.95 10.42 17.15 10.41 25.91 24.13 26.67

38.10 14.54 10.15 20.53 17.25 15.42 22.14 14.13 17.84 33.64 24.66 12.46 20.37 21.91 14.55 32.29 23.92 25.85

13.54 24.92 12.63 20.87 24.97 9.50 17.73 14.84 14.51 21.49 23.16 13.79 10.27 9.90 17.20 28.35 24.82 31.66

19.68 22.32 12.01 20.78 23.04 11.19 18.83 14.71 15.34 24.53 23.54 13.46 12.79 12.90 16.54 29.34 24.59 30.21

21.11 18.65 1.16

19.21

Mean 19.86 18.26 MeanS.E. 1.19

* Plots located on left or right side of rotating boom rainfall simulator, S.E. = 5.01. t Flat plots with eroding slopes similar to site, 4-6%; ridged plots with eroding side slopes of 50-60%, S.E. = 4.92. i S.E. = 3.01. EROSION RATE

Calculated mean interrill erosion rates are shown in Table 4. There were significant differences in mean interrill erosion rates among soils. The difference in rates between the ridged and flat plots was also significant. As was predicted by Foster (1982), significantly higher interrill erosion rates were observed on the ridged plots on TABLE 4. Mean interrill erosion rate (g m-2 m-*) Soil No. of plots

Left*

Right

Hatt Ridged Meanij: Ridged Flat 2 6 8

3

5

21.51 51.26 48.26 53.34 49.08 18.18 41.02 36.59 27.15 13.84 45.64 49.05 40.43 34.81 44.84 35.19 33.16 30.30

29.40 83.67 58.11 59.47 84.27 17.21 31.39 36.34 33.45 13.00 50.72 61.53 48.57 27.17 33.57 34.58 25.71 37.30

9.76 29.37 24.72 70.91 19.47 61.23 16.76 85.08 16.76 77.45 15.63 18.62 13.46 45.39 15.09 43.63 14.59 34.49 6.00 16.03 22.98 55.73 30.42 61.49 25.26 49.56 15.01 37.60 17.44 48.34 12.96 42.29 7.53 37.98 20.16 37.18

24.47 59.36 51.95 55.64 62.28 17.76 37.41 36.50 29.51 13.52 47.54 53.73 43.48 31.95 40.61 34.96 30.37 32.92

3.01 2.87 3.14 5.08 4.62 1.19 3.37 2.89 2.36 2.67 2.43 2.02 1.96 2.50 2.77 3.26 5.04 1.84

Mean 37.79 41.75 MeanS.E. 1.99

16.90 46.78 2.23

39.10

2.94

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

Plots located on left or right side of rotating boom rainfall simulator, S.E. = 8.44. Flat plots with eroding slopes similar to site, 4-6%; ridged plots with eroding side slopes of 50-60%, S.E. 9.44. S.E. = 5.78.

VOL. 33(6): NOVEMBER-DECEMBER 1990

all soils. There was a significant interaction between soil type and slope, which means that the effect of slope on erosion rate was different for different soils. A comparison of the ratio of the erosion rates from a ridged plot to that of a flat plot reveals that on highly erodible soils, the ratio is greater. The higher ratio is probably because the overland flow transport mechanism is not adequate to carry away the large amounts of sediment detached by raindrop impact on the flatter plots (Foster, 1982). This difference is particularly evident on two sandy soils, 4 and 17, where the high interrill erosion rates of sand grains led to a 5:1 ratio between the ridged and flat plot erosion rates. Soil 6, an aggregated clay, had a small erosion rate which resulted in a ratio of approximately one between the ridged and flat plot erosion rates. As this soil eroded, it formed spherically shaped aggregates which, with the smooth shape and low specific gravity, would easily be transported by overland flow, even on low slopes. The effect of the slope of the plot and its interaction with soil type on interrill erosion merit further analysis. There were significant differences in erosion rates between the left and right side of the simulator even though there were no significant differences in rainfall, runoff, or infiltration because of the side on which the plots were located. This side-to-side difference can possibly be explained by the effect of the metal plot borders in "collecting" raindrops on the right side of the simulator where the booms were swinging upslope and shielding the top edge of the plot from raindrops on the left side of the simulator where the booms were sweeping downslope (figs. 1, 2, and 3). On several soils, a dry surface about 20 mm wide was observed on the side of a plot border shielded from the horizontal motion of raindrops leaving the simulator. The edge effect could be made negligible in future rotating-boom research if there were the same number of plots to the left and right of the simulator, or if larger plots were used. The interaction between soil and side was not significant. Differences in interrill erosion rates due to rainfall intensity were significant. This finding was in keeping with those reported in the review of literature. Since there was little variation in intensities in this experiment, no further analysis of the relationship between intensity and observed erosion rates was made. SLOPE FACTOR

Because the observed erosion rates varied with the type of the plot on most soils, a study of plot slope effects in these data, as well as those published by Lattanzi et al. (1974), Watson and Laflen (1986), and Meyer and Harmon (1987), was carried out to see if the effects of slope could be eliminated by a simple slope factor. When plotting all of the erosion rates versus their respective slopes, a curve similar in shape to figure 5 was obtained. This curve is the same basic shape as the theoretical curve suggested by Foster (1982). Various forms of the slope factor function were analyzed using a nonlinear regression package (Allen, 1987) and the following form provided the largest coefficient of determination (r^) for the above data sets: { - g sin() }

Sf = d - fexp

(4)

1885

-%

o Ridged Data



• Flat Data

^ 6H

— Ridged Regression r 2= 0.07 Flat Regression r 2= 0.49

altered until the value that gave the highest r^ was found. The resulting coefficients are presented in Table 5. There was little change in the r^ if g varied within a range of + 10% of that shown in Table 5, so fixing g at the median value of 4 for all soils provided a better comparison of the other coefficients. Table 6 gives the values of all the coefficients in equation 4 by fixing g at a value of 4. The regression analyses resulted in an equation in the form:

5i = d'

o u

0

10

20

30

40

50

Infiltration Rate, i, mm/hr

{ - g sin((|>) }

f exp

(6)

where d ' = Kid, f = Kif,and d and f are as in equation 4. From equation 5, Kj Sf was substituted for I^ in equation 6 to give:

Figure 4-Interrill erodibility vs. infiltration rate.

KjSf = d' - r e x p

{ - g sin((|)) }

(7)

where Sf = the slope factor, d = a constant, f = a constant, g = a constant, and (|) = the slope angle. It was assumed in the analysis that when the slope was greater than or equal to 100% or 45°, that Sf would be 1.0.

From the regression equation 7, the relationship between d' and f could be determined. It was then possible to find values for d and f when