Infiltration rate assessment of some major soils

John Diamond and Thomas Shanley

Teagasc, Johnstown Castle Research Centre, Wexford ABSTRACT
An evaluation of the risk of overland flow is needed in order to minimise the risk of transferring pollutants from soil to rivers and lakes. The infiltration capacity of soil influences the occurrence of overland flow. The objectives of this study were to assess the infiltration capacity of some extensive soils in Ireland and also its spatial and temporal variability. Infiltration capacity was measured using double-ring infiltrometers at one poorly drained, one imperfectly drained and eight freely drained sites. The first series of measurements was performed for one day in summer. Eight years later a second series was performed for two days in winter and two days in summer at the same sites. On average, eight replicates were required in summer and fifteen in winter to estimate the mean with 50 percent precision. Infiltration capacities were reasonably stable between the two series but there was a significant difference between seasons. There was a significant relationship between infiltration capacity and the antecedent soil water content, which contributed to the seasonal effect. Capacities in summer were 3.5 times the winter values. In summer, the infiltration capacity exceeded or equalled the five-year return rainfall rate, except on the poorly drained soil, indicating that the risk of overland flow, due to infiltration excess, is negligible on well drained soils in summer. In winter, the infiltration capacity at nine of the ten sites, predominantly freely drained, was below the five-year return hourly rainfall indicating a risk of overland flow in winter. Key index words: Infiltration, soil, year, season.

Introduction Infiltration is the process by which water enters the soil. It separates water into two major hydrologic components - surface runoff and subsurface recharge. The assessment of runoff risk has assumed an increased importance because of concerns about the associated pollution hazards. Accurate determination of infiltration rates is essential for reliable prediction of surface runoff. As environmental impact assessments are concerned with long-term effects, it is essential that the infiltration data on which they are based should be reasonably stable over decades. For planning purposes it is essential to know the stability of infiltration data under Irish conditions and whether the infiltration capacity of individual soils is adequate to cope with the anticipated hydrologic loads. The infiltration process Infiltration rate usually shows a sharp decline with time from the start of the application of water. The constant rate approached after a sufficiently large time is referred to as the steady-infiltration rate. The process is described by the equations of Kostiakov (1932) and Horton (1940), which show a decreasing infiltration rate as a function of time. As these equations and the related experiments were empirical the coefficients of the equations have no physical meaning.
Irish Geography, Volume 36(1), 2003, 32-46.

Infiltration rate assessment

33

Green-Ampt (1911) and Phillip (1957a) developed mathematical solutions to physically based theories of infiltration. The Green-Ampt theory considers water to move downwards as piston flow. The system is assumed to consist of a uniformly wetted near-saturated transmission zone above a sharply defined wetting front of constant pressure head. Application of the Darcy equation for saturated flow through porous media to this simple system yields: i = K(H + L - h)/L (1)

where i is infiltration rate, K is hydraulic conductivity, H is ponding depth, h is pressure head at wetting front and L is depth to wetting front. As the wetting front deepens, the value of the pressure head (h) becomes relatively insignificant compared to the depth (L) and the infiltration rate (i) approaches the hydraulic conductivity (K) of the transmission zone. Ponding depth affects infiltration most at the beginning of infiltration while L is small. As infiltration progresses and the depth to wetting front (L) increases, the effect of ponding depth decreases and eventually becomes negligible when L is sufficiently deep (Bouwer, 1986). The mathematical and physical analysis of the infiltration process developed by Phillip (1957b) separates the process into two components - that caused by a sorptivity factor and that influenced by gravity. Sorptivity is the rate at which water will be drawn into a soil in the absence of gravity; it comprises the combined effects of adsorption at surfaces of soil particles and capillarity in soil pores. The gravity factor is due to the impact of pores on the flow of water through soil under the influence of gravity. The Phillip model takes the form of a power series but in practice an adequate description is given by the two-parameter equation: i = S t- + A (2)

where i is infiltration rate, S is sorptivity, t is time and A is a gravity factor related to hydraulic conductivity. The sorptivity (S) is influenced by the initial and final moisture contents. As the moisture content approaches saturation, sorptivity tends to zero and the infiltration rate becomes equal to the field saturated hydraulic conductivity. This implies that the steady infiltration rate reached after a long time should be largely independent of the antecedent moisture content (Phillip, 1957b). The theoretical analyses outlined above imply that the steady infiltration rate is a function of the pore configuration of the soil. Soils derived from shale, sandstone or limestone, containing little or none of the swelling clay mineral montmorillonite (Kiely, 1971), are predominant in Ireland (Gardiner and Radford, 1980). Consequently the steady infiltration rate should remain stable over time unless the soil structure is altered by animal or machine traffic. Objectives The objectives of the investigation were to evaluate the infiltration capacity of the dominant component of major soil associations and to evaluate the reliability of infiltration tests. Materials and methods Sites Measurements were made at ten sites that were under old pasture. The sites were originally used in a study of the transfer of radiocaesium from soils to crops after the Chernobyl incident (MacNeill et al., 1992). A selection criterion for the original study was that

34

Diamond and Shanley

Figure 1: Location of sites.

tillage crops should be available in the area and consequently the sites were confined to the east, south-east and midland regions of Ireland (Figure 1). The sites were chosen to represent the dominant component of major soil associations. Infiltration measurements had been performed at the sites in summer 1988 (series 1). Repeat measurements were made in 1995 or 1996 (series 2) and, except at Clonroche, were located in the same field as near as possible to the original site. At Clonroche, an alternative site on the same soil series was selected following the sale of the original site. The classification, drainage and texture characteristics of each site are given in Table 1. Two sites, Clonroche and Dundalk, represent Soil Association 14; the latter is a stony phase. Most of the sites are well drained or moderately well drained reflecting the preponderance of free drainage in the more intensively utilised agricultural areas in the country. The Castlecomer site, an impermeable Gleysol, was anticipated to represent soils of high runoffrisk while the Kilcock site represents the transition between the free draining soils and Gleysols. Textures of the free draining sites range from sandy loam to loam in the A horizon and from sandy loam to clay loam in the B horizon. Structure of the A horizon was moderately well developed at eight sites; the structure was strongly developed at the Dundalk site and was weakly developed at Castlecomer. Sites were not visibly trampled to any significant extent, except at Birr, in the second series of measurements, where sheep trampled the site in the interval between the summer and winter tests.

Infiltration rate assessment Table 1: Classification, drainage and texture characteristics of sites. Site No. 1 2 3 4 5 6 7 8 9 10 Name Association1 Group2 Drainage class Well Moderate Well Well Poor Well Imperfect Well Well Well WRAP3 class High High Very high Very high Very low High Moderate Very high Very high Very high Texture (horizon) A B Loam Loam Sandy loam Loam Clay loam Loam Loam Loam Loam Loam Clay loam Clay loam Sandy loam Loam Clay loam Clay loam Clay loam Loam Loam Stony loam

35

Clonroche Kinsealy Birr Gurteen Castlecomer Clonmel Kilcock Cappoquin Midleton Dundalk

14 38 30 31 22 34 40 6 13 14

Dystric Cambisol Orthic Luvisol Eutric Cambisol Orthic Luvisol Dystric Gleysol Orthic Luvisol Gleyic Luvisol Leptic Podzol Eutric Cambisol Dystric

3 Winter

Gardiner and Radford (1980); 2 Definitions according to FAO (1974); Rainfall Acceptance Potential (Farquaharson et al. 1978).

WRAP (Table 1) is an assessment of the capacity of different soils to accept rainfall without run-off; it depends on soil water regime, depth to an impermeable layer and permeability of horizons above the impermeable layer (Farquaharson et al., 1978). A high WRAP implies a low run-off potential and conversely for a low WRAP. The WRAP classes shown in Table 1 pertain to the specific sites and are higher for the Cappoquin, Dundalk and Gurteen sites than that given by Gardiner (1986) for the principal soil of the soil association in which these sites occur . The WRAP classes of the other seven sites are the same as that given by Gardiner (1986). Measurement of infiltration Infiltration rate was measured by the same method in both the first series of measurements, performed in 1988 and in the second series, performed in 1995-1996. Doublering infiltrometers, consisting of two concentric rings, were used (Figure 2); the rate of fall of water was measured in the inner ring while a pool of water was maintained at approximately the same level in the outer ring to reduce the amount of lateral flow from the inner ring.

36

Diamond and Shanley

Figure 2: Double-ring infiltrometer.

Six double-ring infiltrometers were set up at each site. The diameters of the inner rings were 280 mm, 300 mm or 320 mm; the outer diameters were 540 mm, 560 mm, or 580 mm. The slight variation in diameter allowed nesting of the rings during transport. Rings were 250 mm deep and were made from 12-guage steel with sharpened bottom edges; they were driven into the ground to 50 mm depth. Grass was cut to near soil level and a pad was placed inside the inner ring to prevent puddling. The inner and outer edges were tamped to seal possible cracking. Generally the water level was kept at or above 50 mm depth; the difference in height between the inner and outer rings was kept to a minimum. The amount of water required per site, on the first day in summer, varied from 0.3 m3 at Castlecomer to 3.5 m3 at Cappoquin; a smaller amount was generally required on the second day. The rate of fall of the water level in the inner cylinder was measured at 1, 5, 10, 15, 30, 45 and 60 minutes and at 30-minute intervals thereafter. The test was conducted for one day in the first series of measurements; in the second series, the process was stopped on the first day (day 1) once a steady infiltration rate had been found and the process was repeated on the following day (day 2). The duration of each test was from 4.5 to 7.5 hours in winter and from 6.5 to 8.5 hours in summer. Additional measurements were only made at Clonroche, where slow infiltration rates were found, to test for surface compaction; the surface layer (0-100 mm) was removed and the tests were performed on the exposed 100 mm surface. Flow rate through a finite ring overestimates the true infiltration rate due to the effects of lateral flow and ponding. The configuration used in this research, 300 mm inner ring and 600 mm outer ring, could overestimate the steady infiltration rate by a factor of 0.3 (Baver et al., 1972; Wu et al., 1997). Adding a correction for 50 mm ponding depth (Wu et al., 1997) brings the factor to 0.4. In the second series, at each site, ten soil samples were taken at random from the surface layer (0-60 mm) in the area outside the outer rings. These were bulked for the determination of gravimetric moisture content. Attempts to obtain undisturbed samples in winter, for the determination of bulk density, were unsuccessful due to the weak consistence of the wet soil.

Infiltration rate assessment

37

In the following summer, at least twelve undisturbed samples were obtained from each site, without evident disturbance, using 50 mm x 50 mm stainless steel cylinders. Hence bulk density data were only available for summer. Soil samples had not been taken in the first series which limited the assessment of the influence of antecedent moisture contents to seasonal effects. Statistical analysis Statistical analysis was performed with the SAS statistical package. The General Linear Model procedure (GLM) was used for the analysis of variance. The GLM procedure is appropriate for unbalanced designs, which have cells that contain a different number of values from other cells; in series 2 summer, there were two values, day 1 and day 2, whereas, in series 1, there was only one value (day 1). Duncans multiple-range test (Duncan, 1975) was performed on the means of sites. The sample size necessary to estimate the mean infiltration rate for a given confidence interval was determined by the Stein procedure (Steel & Torrie, 1980). Levels of significance are indicated as follows: * = p < 0.05, ** = p < 0.01, *** = p < 0.001. Results and discussion Moisture content and density At each site the initial moisture content at 0 - 60 mm depth was lower in summer than in winter (Table 2). Bulk density was low at all sites and was mostly less than 1 Mg m-3 This is attributed to the abundance of grass roots at the surface. It indicates that compaction by machines or animals was negligible prior to the time of sampling in summer, series 2. Table 2: Antecedent soil water content and bulk density. Site Gravimetric water content kg kg-1 Summer Winter 0.53 0.32 0.18 0.40 0.84 0.55 0.23 0.48 0.06 0.21 0.38 0.61 0.53 0.44 0.69 1.38 0.64 0.81 0.82 0.52 0.57 0.70 Volumetric water content m3 m-3 Summer Winter 0.51 0.31 0.20 0.42 0.51 0.47 0.20 0.44 0.07 0.20 0.33 0.59 0.50 0.51 0.72 0.84 0.55 0.71 0.75 0.58 0.56 0.63 Bulk density Mg m-3 Summer 0.96 0.95 1.16 1.05 0.61 0.85 0.87 0.92 1.13 0.99 0.95

Clonroche Kinsealy Birr Gurteen Castlecomer Clonmel Kilcock Cappoquin Midleton Dundalk Mean Influence of sites

Highly significant (***) differences were found among sites in each series of measurements (Table 3). The grouping of sites by the Duncan Multiple Range Test showed differences in the classification of individual sites between the two series of measurements; it is not possible to determine whether these differences were due to antecedent moisture

38

Diamond and Shanley

contents as the moisture contents were only determined for the second series. In winter, two groups were distinguished; one group consisted of a single site, Dundalk, which had very high infiltration rates. When this site was removed from the analysis the highest group still consisted of a single site, Midleton. Infiltration tests are a poor discriminant of soils in winter, because, as discussed below in the section on seasonal variation , the infiltration rates are more influenced by hydrologic regime than by soil properties. In summer, the Duncan Range Test (p < 0.05) distinguished three to five classes. The Castlecomer site was always in the lowest class; the low rates on this site are consistent with the fine silty texture, massive structure and gley features. In summer the sandstone derived sites at Midleton and Cappoquin were always in the top two classes (Class A or B, Table 3). This is consistent with the free drainage and coarse texture of these soils. Table 3: Mean steady infiltration rate (mm hr-1) of each site and grouping by Duncans Multiple Range Test. Site Series 2, Day 1 Summer Winter Infiltration Duncan Infiltration Duncan rate group rate group 28.0 37.7 24.7 95.0 18.0 87.3 102.3 196.7 117.3 88.3 c c c b c b b a b b 1.3 22.7 17.3 13.0 2.0 25.7 14.7 14.0 40.7 222.3 cx bc bc bc c bc bc bc b a Series 1, Day 1 Summer Infiltration Duncan rate group 116.8 31.7 73.0 49.8 14.7 21.5 54.8 133.7 132.6 87.8 abx ef cd de f ef de a a bc

Clonroche Kinsealy Birr Gurteen Castlecomer Clonmel Kilcock Cappoquin Midleton Dundalk

Series 2, Day 1 Summer Winter Infiltration Duncan Infiltration Duncan rate group rate group Clonroche Kinsealy Birr Gurteen Castlecomer Clonmel Kilcock Cappoquin Midleton Dundalk
xsites

8.0 29.0 53.7 114.0 5.3 30.0 96.3 160.0 103.0 96.0

dx cd c b d cd b a b b

0.8 11.7 11.3 8.7 1.7 10.3 6.0 3.7 22.0 122.0

bx b b b b b b b b a

with the same letter, within a column, are not significantly different (=0.05).

Infiltration rate assessment

39

Table 4 compares the infiltration rates of two soil groups, classified according to the texture of the soil profiles. In each set of measurements, the infiltration rate of the coarse group was much higher than the medium to fine group. The ratio of infiltration rates between the two texture groups ranges from about 5:1 in winter to about 2:1 in summer. The high ratio in winter is due to the unusually high rates at the Dundalk site. When this site is excluded the ratio is similar to that found in summer. Some overlap exists between the textural groups; for example the medium textured Kilcock site has a higher rate than the coarse textured Birr site. As it is not possible to vary soil texture independently of other characteristics it is not inferred that the infiltration rates are caused by texture. It is more likely due to associated characteristics including the configuration of pores and fissures. Nevertheless it seems that infiltration rates above 100 mm hr-1 in summer are invariably associated with coarse texture whereas values less than 50 mm hr-1 are most likely on medium to fine soils. All sites in the coarse group belong to the very high WRAP class, whereas sites in the fine group range from high to very low WRAP. Sites with a very high WRAP tend to have greater infiltration rates than sites with a lower WRAP. The Castlecomer site has the lowest WRAP and also had consistently slow infiltration rates, but the Kilcock site, with a moderate WRAP, had a high infiltration rate in summer. Table 4: Infiltration rates (mm hr-1) of coarse and medium/fine textured soils. Year 1 2 Season Summer Summer Winter Day 1 1 2 1 2 Coarse 95 104 105 63 34 Medium/fine 45 55 40 13 6

Variation within sites and sampling requirement Intra-site variation in steady infiltration rates was substantial and varied widely between sites (Table 5). In summer, series 2, the coefficient of variation (CV) ranged from 0.14 at Cappoquin to 0.87 at Kinsealy. The range in CV was broadly similar in series 1 (CV = 0.09 to 0.97) but there were substantial differences at individual sites between the two series; it is not possible to determine the influence of antecedent moisture contents on the CV differences between series because moisture contents were not available for the first series. Intra-site variation was much higher in winter; the CV ranged from 0.40 at Dundalk to 1.05 at Midleton (Table 5). The spatial variation in summer was similar to the variation in field saturated hydraulic conductivity (CV = 0.5) found in glacial till soils (Diamond, 1984). The similarity is to be expected, as the steady infiltration rate is theoretically equal to the mean vertical hydraulic conductivity of the profile (Bouwer, 1986). Estimates of the number of observations required to estimate the mean infiltration rate of a site at three levels of precision are shown in Table 5. As the variation tends to be greater at lower infiltration rates the levels of precision are defined as a proportion of the mean, namely, 1, 0.4 and 0.2 times the mean. On average eight

40

Diamond and Shanley

Table 5: Mean steady infiltration rate, standard deviation (sd), coefficient of varation (cv), and number of measurements required at three levels of precision (p < 0.05). Site mean mm hr-1 sd cv Confidence interval (x mean) 0.2 0.4 1

Series 1, Summer, Day 1 Clonroche Kinsealy Birr Gurteen Castlecomer Clonmel Kilcock Cappoquin Midleton Dundalk Mean 116.8 31.7 73.0 49.8 14.7 21.5 54.8 133.7 132.6 87.8 22.3 21.9 6.5 19.1 11.5 20.9 18.2 47.3 41.0 9.9 0.19 0.69 0.09 0.38 0.78 0.97 0.33 0.35 0.31 0.11 24 317 5 97 406 624 73 83 63 8 170 Series 2, Summer, Day 2 Clonroche Kinsealy Birr Gurteen Castlecomer Clonmel Kilcock Cappoquin Midleton Dundalk Mean 8.0 29.0 53.7 114.0 5.3 30.0 96.3 160.0 103.0 96.0 6.4 25.3 16.0 40.3 3.0 20.2 37.3 22.2 44.7 36.2 0.81 0.87 0.30 0.35 0.56 0.67 0.39 0.14 0.43 0.38 430 503 59 83 211 301 99 13 124 94 191 Series 2, Winter, Day 2 Clonroche Kinsealy Birr Gurteen Castlecomer Clonmel Kilcock Cappoquin Midleton Dundalk Mean 0.8 11.7 11.3 8.7 1.7 10.3 6.0 3.7 22.0 122.0 0.7 7.1 8.8 7.7 1.5 4.4 3.6 2.0 23.2 49.3 0.92 0.61 0.78 0.88 0.90 0.43 0.60 0.54 1.05 0.40 556 244 401 516 539 123 235 190 732 108 364 139 61 100 129 135 31 59 48 183 27 91 22 10 16 21 22 5 9 8 29 4 15 107 126 15 21 53 75 24 3 31 23 48 17 20 2 3 8 12 4 1 5 4 8 6 79 1 24 102 156 18 21 16 2 43 1 13 1 4 16 25 3 3 3 1 7

Infiltration rate assessment

41

replicates were required on day 2 in summer and 15 in winter to estimate the mean rate of a site with 50 percent precision (CI = 0.1 x mean). In summer this rises to 48 at twenty percent precision and to 191 at ten percent precision. Some individual sites had extraordinarily high sampling requirements e.g. Kinsealy which required over 500 observations for ten percent precision. The number of observations required for a given level of precision was higher in winter than in summer. In general about twice as many were required in winter (Table 5). With eight replicates, the 50 percent precision level was achieved at seven sites in summer and at only three sites in winter. The large numbers of replicates required, even on homogeneous soil areas, make the attainment of greater than 50 percent precision impractical. The sampling requirements of a particular site are not predictable. Eight should be adequate in summer for most sites comprising a single soil series and phase. Additional measurements should be made as indicated by the variance. Land areas comprising more than one soil unit will require a corresponding increase in the number of observations. Duration of tests Regression analysis shows a significant correlation between the steady infiltration rates on day 1 and day 2 in summer (R2 = 0.754***) and also in winter (R2 = 0.770***). The regression equations (Table 6) are based on the steady infiltration rates of individual rings, and not on the site means, as the individual rings were in exactly the same location on both days. Table 6: Regression showing relation between steady infiltration rates at various times. Dependent Independent variable y variable x Y2SD2 Y2SD2 Y2SD1 Y2WD2 Y2SD1 Y1SD1 Y1SD1 Y2WD1 Regression Equation y= 9.06 + 0.78x y= 29.1 + 0.71x y=41.55 + 0.53x y=0.58 + 0.50x R2 Significance

0.754 0.370 0.177 0.770

p<0.001 p<0.001 p<0.01 p<0.001

S1 = Series 1; S2 = Series 2; D1 = Day 1; D2 = Day 2; S = Summer; W = Winter

The mean steady infiltration rate on day 2 was slightly lower than on day 1 in summer; analysis of variance indicated that the difference was not statistically significant (Table 7). In winter, the mean steady infiltration rate on day 2 was about half that on day 1 and the difference was not statistically significant (Table 7). Table 7: Mean steady infiltration rates (mm hr-1) by series, season and day. Winter Series 1 Series 2 Significance day 1 day 1 day 2 Summer 69.4 78.1 69.5 NS Significance

38.1 19.8 NS

p<0.001 p<0.001

42

Diamond and Shanley

The smaller difference between successive days in summer than in winter is contrary to that implied by the theory of the infiltration process (Phillip, 1957a). The initial moisture contents were invariably lower in summer than in winter; it should therefore take longer to reach the steady-state in summer and hence the difference between successive days should be greater. At three sites in summer, the infiltration rates were higher on the second day. At Birr there was a twofold increase and the infiltration rate increased in three rings over the last seven hours. This is consistent with the wetting front reaching the coarse gravelly C horizon at 700 mm depth on the second day; the absorption of water into a two-layer composite is ultimately controlled by the second material in the absorbing sequence (Wilson and Hoff, 1997). Also the release of air compression ahead of the wetting front could contribute to the increase (Morel-Seytoux, 1976). The Gurteen and Dundalk sites had a slight increase (c. 10 percent) on day 2: these sites have stony/gravelly C horizons at 600 mm depth. Although the mean difference in winter between day 1 and day 2 was not statistically significant, the relatively large reduction in infiltration rate on day 2, at some sites implies that the true steady infiltration may not have been achieved on day 1 at all sites: The two-day test is therefore desirable in winter to reduce the risk of overestimates at individual sites. An alternative way would be to apply the correction indicated by the regression equation to the day 1 data; however, this is unreliable as the same relationship may not obtain under a different moisture regime. Although on average the difference is small in summer, the two day test is still desirable because of large differences at individual sites such as, for example, Birr. Also, the increased rate found on the second day, at sites such as Birr, would indicate a risk of groundwater contamination. Long-term variation The difference in mean infiltration rate between series 1 and series 2 was very small (Table 7) and was not statistically significant (p < 0.05). Regression analysis (Table 6) showed reasonable conformity of individual measurements between series (R2 = 0.37***). At the nine sites in which measurements were repeated in the same field the mean infiltration rates of sites were significantly related (R2 = 0.53*) on day 1 and very significantly related (R2 = 0.70**) on day 2. Given the large intra-site variability, and that the tests were not performed in exactly the same position in both years, this implies that the soil structure and associated pore configuration were reasonably stable over the eight year period. The low bulk density (Table 2) indicates that compaction by animal or machine traffic was insignificant at the time of sampling in summer, series 2. The long term stability of steady infiltration rates in summer indicates that infiltration rate tests can have a role in the long-term prediction of surface runoff. Seasonal variation Infiltration rates were substantially greater in summer than in winter. On day 1 the steady infiltration rate in summer was twice the winter rate. The infiltration rate in summer on day 2, which is a better measure of the steady rate, was 3.5 times the winter rate. The differences between seasons were statistically very significant (p < 0.001). The stability of infiltration rates between years indicates an overall long-term stability of soil structure. Seasonal changes in soil structure, especially in the subsoil, are unlikely to be significant. Abrupt changes in infiltration rate can occur in soils composed of swelling clays (Lim et al., 1998). However, in Ireland, the smectitic minerals that form swelling clays are not associated with shale,

Infiltration rate assessment

43

Figure 3: Relationship between steady infiltration rate (mm hr-1) and antecedent soil water content (kg kg-1 dry soil) of surface layer (0-60 mm) in series 2, day2. The solid line represents the fitted linear model based on log transformed infiltration rates. The dashed lines are 95% confidence limits.

sandstone or limestone which comprise the parent materials of the sites. Except at Birr, there was no visible evidence of compaction by animals or machines having occurred in the interval between the summer and winter tests. There was a clear difference in the antecedent water content of the surface layer (0-60 mm) between summer and winter (Table 2). A scatter diagram of winter and summer infiltration rates in relation to antecedent soil water content in series 2, day 2 is shown in Figure 3; the regression line was fitted by least squares analysis to the log transformed infiltration rates which formed a normal distribution. This relationship was described by the equation: Log (infiltration rate)=2.158 -1.634*(gravimetric water content) (R2 0.520, n=20, p < 0.001). The R2 decreased slightly when volumetric water content replaced gravimetric water content. Least squares analysis does not perform well when outliers are present e.g. Clonroche site which lies outside the 95% confidence limits (Figure 2.). Robust regression analysis, which downweights the influence of outliers, increased the R2 to 0.755 (p < 0.001). It was concluded that a change in moisture regime from summer to winter was the most likely cause of the large seasonal difference in infiltration rates. Comparisons of infiltration rates across seasons, under Irish rainfall conditions, are therefore inaccurate unless they can be corrected for antecedent soil water content. Ideally, the soil water regime in the subsoil, and hysteresis effects, would need to be included. The Clonroche site, which appears to be anomalous (Figure 2), illustrates the influence of the former; this well drained site had the slowest infiltration rate in winter. When the surface

44

Diamond and Shanley

layer (0-100 mm) was removed, the rates were slightly slower at 100 mm depth than had been found at the surface, indicating that the slow rates were not caused by compaction of the surface layer. Pressure head, measured with tensiometers at an adjacent site in the same field, was 20-25 hPa above 600 mm depth and decreased to -4 hPa at 1200 mm depth. This indicated the presence of a perched water table, which would impede the downward flow of water and cause slow infiltration. Infiltration is analogous to hydraulic conductivity whose relationship with water content is strongly hysteretic (Mualem, 1986). At a given time the hydraulic conductivity depends on the history of wetting and draining. Staple (1965) found that conductivities were ten times greater for wetting than drying at moisture contents of 0.3 to 0.4 and stated that the duration of wetting may be involved in the hysteresis pattern of conductivity. Implications for overland flow Since rainfall is generally intermittent in Ireland the intensity over hourly periods is more appropriate than daily rates to evaluate the risk of overland flow. The expected rainfall, with a return period of five years for times of one hour, is less than 18 mm in the lowland regions of the country (Logue, 1975). Obviously the expected rainfall or the maximum liquid application rate should not exceed the infiltration capacity if surface runoff is to be avoided. However the measured values could overestimate the true steady infiltration rate by a factor of 0.4 (Baver et al., 1972; Wu et al., 1997). In summer the adjusted steady infiltration rates were at or above 18 mm hr-1 at all sites except Castlecomer. This implies that, except on impermeable gley soils, the risk of overland flow in summer is negligible on soils that are wettable and lack visible evidence of abnormal compaction by machines or animals. The steady infiltration rate defines a minimum absorption capacity. At any given time, especially in summer, the soil will be capable of storing additional water both in and on the soil; the amount will depend on the antecedent weather conditions and on micro relief. This reduces still further the actual risk of overland flow. However the infiltration capacity of the Gley at Castlecomer, in summer, was less than the five-year return rainfall implying a risk of overland flow even in summer. Analogous impermeable Gleys are extensive in west Limerick, west Clare, Leitrim and Cavan (Soil Associations 22, 25, 27; Gardiner and Radford, 1980); they probably, generally, have a risk of overland flow similar to the Castlecomer site and some areas with higher rainfall would probably have a greater risk. In winter, the infiltration capacity at all sites, except Dundalk, was less than the five-year return hourly rainfall indicating a risk of overland flow on both freely drained and poorly drained soils in winter. Four sites, including two freely drained sites, were below 5 mm hr-1 and two sites were below 2.5 mm hr-1. The Code of Good Agricultural Practice (Anon., 1996) recommends that the appropriate rates of application of soiled water by irrigation in sensitive and other areas should not exceed 2.5 mm hr-1 and 5.0 mm hr-1 respectively. Six sites could cope with the maximum rates but irrigation at these rates in the other sites could result in overland flow. With the possible exception of the impermeable Gleys, it is not possible to discriminate between sensitive and other areas in winter on the basis of infiltration tests. The risk is determined by the interaction between weather conditions and hydrologic regime, which change with time. Experiments at Hoarstone (Ksat =5 mm hr-1) and Castlebridge (Ksat = 1.25 mm hr-1) showed a high frequency of runoff: 12 to 17 runoff events each year (Sherwood and Fanning, 1981). This is attributed to the low permeability on both sites which is within the range of

Infiltration rate assessment

45

gley soils derived from glacial till (Diamond, 1984). The summer infiltration rates on the freely drained soils are much higher than the hydraulic conductivity rates at the Hoarstone and Castlebridge experimental sites and would consequently have a much lower frequency of overland flow. There was a pronounced seasonal effect. Mean monthly runoff was less than 1 mm at Hoarstone for seven months (April - October inclusive) and for three months (July September inclusive) at Castlebridge and runoff was substantially greater for the remaining months at both sites. The low runoff period corresponds to the normal grazing season at Hoarstone. On free draining lowland soils the low runoff period should be at least as long as at Hoarstone and is probably longer due to greater permeability. Conclusion Infiltration capacity was reasonably stable over an eight year period and accordingly infiltration tests can have a role in assessing the long-term risk of overland flow and the associated pollution hazard. Substantial variation occurred within sites and the variation was greater in winter than in summer. In general, for 50 percent precision, eight measurements are required in summer, and fourteen in winter, on sites composed of a single soil series or phase. A correspondingly greater number will be required on land areas composed of a number of soils. Infiltration tests performed in summer are preferable, as tests performed in winter are unlikely to reflect stable soil characteristics. To assess the risk of overland flow over the whole year, additional information will be required on the duration and degree of wetness. Although the mean difference between day 1 and day 2 was not statistically significant, the two-day test is desirable to reduce the risk of overestimates at individual sites. There was a pronounced seasonal effect, which was attributed to the influence of antecedent soil water content on the measured infiltration capacities. In summer, infiltration capacity was on average 3.5 times the winter rate. Except on an impermeable Gley, the infiltration capacity in summer equalled or exceeded the hourly rainfall expected once in five years. This implies that the risk of overland flow is likely to be very small in summer on freely drained soils that have not been compacted by animals or machines. Irrigation rates of 5 mm hahr-1 or 2.5 mm hrhr-1, permitted by the Code of Good Agricultural Practice, exceed the infiltration capacity of some soils, including free draining soils, in winter. This implies that there may be a significant risk of overland flow in winter. Acknowledgements We thank Mr P. Sills for technical assistance. This research was part-funded by European Union Structural Funds (EAGGF). References
ANON. (1996) Code of Good Agricultural Practice to protect waters from pollution by nitrates. Dublin: Department of Agriculture, Food and Forestry/Department of Environment, 22. BAVER, L.D., GARDNER, W.H. and GARDNER, W.R. (1972) Soil Physics. New York: John Wiley & Sons, 365-376. BOUWER, H. (1986) Intake rate: cylinder infiltrometer, In: Klute, A. (ed.) Methods of Soil Analysis, Part 1, Physical and Mineralogical Methods. Agronomy Monograph No. 9 (2nd Edition). USA: American Society of Agronomy, Inc. - Soil Science Society of America, Inc., 825-844.

46

Diamond and Shanley

DIAMOND, S. (1984) Land evaluation methodology: A case study in east Waterford, Ireland, In: Haans, J.C.F.M., Steur, G.G.L. and Heide, G. (eds) Progress in Land Evaluation. Rotterdam: A.A. Balkema, 7-25. DUNCAN, D.B. (1975) t-Tests and intervals for comparisons suggested by the data, Biometrics, 31, 339-359. FAO/UNESCO (1974) Soil map of the world 1:5,000,000. Paris: Unesco. FARQUAHARSON, F.A.K., MACKNEY, D., NEWSON, M.D. and THOMASSON, A.J. (1978) Estimation of run-off potential of river catchments from soil surveys. UK Soil Survey, Technical Monograph No. 11. GARDINER, M.J. and RADFORD, T. (1980) Soil Associations of Ireland and their Land Use Potential. Dublin: An Foras Talntais. GARDINER, M.J. (1986) Soil, site and climatic factors in slurry disposal, Irish Journal of Agricultural Research, 25, 273-284 GREEN, W.H. and AMPT, G.A. (1911) Studies on Soil Physics:1. Flow of air and water through soils, Journal of Agricultural Science, 4, 1-24. HORTON, R.E. (1940) An approach toward a physical interpretation of infiltration-capacity, Soil Science Society of America Proceedings, 5, 399-417. KIELY, P.V. (1971) Clay mineralogy of selected soil profiles, In: Finch, T.F. (ed.) Soils of county Clare. Dublin: An Foras Talntais, 88-93. KOSTIAKOV, A.N. (1932) On the dynamics of the coefficient of water-percolation in soils and on the necessity of studying it from a dynamic point of view for purposes of amelioration, Transactions Congress International Society for Soil Science, 6th, Moscow Part A, 17-21. LIM, H.S., McINNES, K.J., WILDING, L.P. and HALLMARK, C.T. (1998) Macropososity and initial moisture effects on infiltration rates in Vertisols and vertic intergrades, Soil Science, 163 (1), 2-8. LOGUE, J.J. (1975) Extreme Rainfalls in Ireland. Technical Note No. 40. Dublin: Meteorological Service. MacNEILL, G., DUFFY, J.T., CUNNINGHAM, J.D., COULTER, B., DIAMOND, S., McAULEY, I.R., and MORAN, D. (1992) Transfer characteristics of radiocaesium from soils to permanent pasture, Analyst, 117, 521-524. MOREL-SEYTOUX, H.J. (1976) Derivation of equations for rainfall infiltration, Journal of Hydrology, 31, 203-219. MUALEM, Y. (1986) Hydraulic conductivity of unsaturated soils: prediction and formulas, In: Klute, A. (ed.) Methods of Soil Analysis, Part 1, Physical and Mineralogical Methods. Agronomy Monograph No. 9 (2nd Edition). USA: American Society of Agronomy, Inc. - Soil Science Society of America, Inc., 799-823. PHILLIP, J.R. (1957a) The theory of infiltration: 4. Sorptivity and algebraic infiltration equations, Soil Science, 84, 257-264. PHILLIP, J.R. (1957b) The theory of infiltration: 5. The influence of the initial moisture content, Soil Science, 84, 329-339. SHERWOOD, M. and FANNING, A. (1981) Nutrient content of surface runoff water from land treated with animal wastes, In: Brogan, J.C. (ed.) Nitrogen losses and surface runoff from landspreading of manures. The Hague: Martinus Nijhof, W. Junk, 5-16. STAPLE, W.J. (1965) Moisture tension, diffusivity, and conductivity of a loam soil during wetting and drying, Canadian Journal of Soil Science, 15(1), 78-86. WILSON, M.A. and HOFF, W.D. (1997) The absorption of water into a porous solid consisting of two dissimilar layer, European Journal of Soil Science, 48, 79-86. STEEL, R.G.D. and TORRIE, J.H. (1980) Principles and procedures of statistics. Singapore: McGrawHill, 119-120. WU, L., ROBERSON, M.J. and SHOUSE, P.J. (1997) Numerical evaluation of ring-infiltrometers under various soil conditions, Soil Science, 162 (11), 771-777.