| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Alberta Agriculture, Food and Rural Development, Irrigation Branch, 100, 5401 First Avenue South, Lethbridge, AB, Canada T1J 4V6
b University of Alberta, 3rd Floor University Hall, Edmonton, AB, Canada T6G 2J9
c Department of Renewable Resources, 751 General Services Building, University of Alberta, Edmonton, AB, Canada T6G 2H1
* Corresponding author (Shelley.A.Woods{at}gov.ab.ca)
Received 10 February 2005.
 |
ABSTRACT |
|---|
 TOPEXECUTIVE SUMMARY ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
Abbreviations: BTC, breakthrough curve • FT, field-scale transect • IT, intensive transect • MR, mass recovery • PT, pedon-scale transect • SD, standard deviation
|
INTRODUCTION |
|---|
|
TOPEXECUTIVE SUMMARYABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
In semiarid climates, the complexity of field-scale unsaturated transport is magnified because annual potential evapotranspiration is greater than annual precipitation. Periodic occurrences of water input in excess of evapotranspiration rates can occur, occasionally resulting in deep drainage and transport of solute below the root zone. These events are typically seasonal, such as during rapid snowmelt, fallow periods, or large precipitation events. As a result, deep drainage in semiarid environments is usually small (<10 mm yr–1), transient, and difficult to quantify (Gee et al., 1994; Tyler et al., 1996; Scanlon et al., 1999; Joshi and Maulé, 2000). At the field scale, factors controlling the spatial redistribution of precipitation (e.g., topography), and the subsequent capacity of plants to use this water, will strongly influence the spatial variability and spatial scale dependence of water and solute transport below the root zone.
It is well known that landform shape (e.g., surface profile [downslope] curvature and plan [across-slope] curvature) strongly influences the spatial redistribution of water within fields. A wetness index (ratio of upslope contributing area to downslope dispersal area/slope gradient) developed by Beven and Kirkby (1979) has been widely used to estimate soil water storage and wetness patterns from digital elevation data of catchment basins. Studies have shown that field-scale patterns of soil water storage in semiarid regions tend to persist, i.e., are time stable, because of the influence of topography on the redistribution of snow, snowmelt, runoff, and throughflow, although the persistence of the spatial pattern is scale dependent (Kachanoski and de Jong, 1988; Zebarth and de Jong,1989). On the Canadian prairies, topography often results in poorly developed surface drainage and numerous small depressions that intermittently hold water (de Jong and Kachanoski, 1987). These small depressions (potholes and sloughs) are a focal point of the prairie hydrological cycle, and can be areas of groundwater recharge, discharge, or both (Lissey, 1968).
In semiarid regions, topography also greatly influences the field-scale spatial distribution of soil properties due to its controlling effects on water-related soil-forming processes (Joel, 1933; King et al., 1983; Miller et al., 1985; Pennock et al., 1987). King et al. (1983) concluded that cultivated soils of the Canadian prairies could only be grouped on a macro scale by convex upper slopes (shallow soils), concave lower slopes (deep soils), and depressional areas (gleyed soils). Pennock et al. (1987) classified prairie landscapes into landform elements based on digital elevation estimates of surface curvature, slope, and gradient. The landscape units were used to describe the spatial distribution of longer term hydrologic regimes and soil morphology. Significant unexplained local-scale soil variability exists within these landscape units. This local-scale variability has been attributed to the influence of the microtopography of the original native, uncultivated soil surface (Kachanoski et al., 1985a, 1985b). Cultivation of a native soil drastically disturbs its surface and eliminates microtopographic features that were responsible for local-scale soil profile development (Kachanoski et al., 1985c).
Topography can significantly influence the field-scale spatial variability of crop water use and yield through its influence on water redistribution and soil properties (de Jong and Rennie, 1969; Moulin et al., 1994; Jowkin and Schoenau, 1998; Si and Kachanoski, 2002). In the semiarid northern Great Plains, water is recognized as the most limiting factor for plant growth and crops generally use most of the water that is available (de Jong and Rennie, 1969; Campbell et al., 1988; Hanna et al., 1982). Several studies have found a high correlation between crop yield and water availability (growing season precipitation and available soil water). This suggests that water use by plants can buffer the impact of spatial and temporal variations of surface water input (to the soil) on subsequent deep drainage and movement of solute below the root zone. Studies showing a quadratic relationship between yield and available water (de Jong and Rennie, 1969; Campbell et al., 1988) suggest that the buffering effect is nonlinear and is reduced at higher amounts of available water. The amount of water redistribution by topography vs. the capacity for plants to use the redistributed water could greatly affect the magnitude, spatial pattern, and spatial scale dependence of deep drainage and solute transport at the field scale, but this has not been studied.
A number of studies have investigated solute transport at the field scale using applied tracers (Schulin et al., 1987; Butters et al., 1989; Ellsworth et al., 1991; Hamlen and Kachanoski, 1992; van Wesenbeeck and Kachanoski, 1995; Ward et al., 1995). Highly variable boundary conditions and soil properties in most field soils has also resulted in the development of stochastic and probabilistic models to describe and predict unsaturated solute transport at the field scale (e.g., Bresler and Dagan, 1979; Jury, 1982; Simmons, 1982; Yeh et al., 1985). Generally, two limiting horizontal spatial scales of transport have been measured: the average local-scale dispersion associated with one measurement point (e.g., one core or one solution sampler), and the asymptotic or field-scale dispersion incorporating field variations of local-scale advection. A method for measuring the transition from the local scale to the asymptotic or field scale for the solute travel-depth variance (dispersion) as a function of horizontal spatial scale was developed by Van Wesenbeeck and Kachanoski (1991). Most field solute transport experiments using applied tracers have been performed under relatively high-flow steady state or quasi-steady state conditions that are generally not typical of semiarid environments. The horizontal spatial scale dependence of solute transport, from local to asymptotic or field scale, has not been examined under long-term, low transient flow conditions typical of semiarid areas.
Recently, the long-term (>30 yr) movement of a Cl– tracer applied to a field under semiarid conditions near Saskatoon, Saskatchewan, Canada, was reported for a 10-m-long transect (Dyck et al., 2003, 2005). The studies examined in detail the spatial variability of the Cl– movement and its relationship to spatial variability of soil horizons and layers. The 10-m transect was located on an upper slope position within a 6.1-m-wide by 90-m-long strip plot where Cl– (as KCl) was applied on the surface 34 yr before sampling. The objectives of this study were to expand the studies of Dyck et al. (2003, 2005) and quantify the transport characteristics of the Cl– tracer at the field scale after 34 yr of low transient flow under semiarid conditions. This includes the transition of the local-scale travel-time variance (dispersion) to the field-scale travel-time variance (dispersion). The influence of topography on the magnitude and spatial redistribution of surface water at the site, during a typical 19-mo fallow period, was measured and its relationship to the field-scale changes in Cl– tracer distribution after 34 yr (i.e., after approximately 17 fallow recharge periods) determined.
|
MATERIALS AND METHODS |
|---|
|
TOPEXECUTIVE SUMMARYABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
|
Precipitation was recorded at an Environment Canada meteorological station at Harris, Saskatchewan, Canada, 25 km southwest of the study site. Although this was the nearest meteorological station, the historic precipitation data may differ slightly from the actual precipitation that occurred at the study site. Annual precipitation averaged 390 mm between 1966 and 2000, with a standard deviation of 75 mm. During that period, the wettest months were normally May to August. Between 1966 and 2000, there were 6 yr where total annual precipitation was <300 mm. There were 3 yr where total annual precipitation exceeded 500 mm and five additional years where precipitation was between 450 and 500 mm. Dyck et al. (2003) gave a summary of meteorological data for Rosetown, Saskatchewan, Canada, located 55 km from the site.
Annual lake evaporation in the area has been estimated at 690 to 710 mm (Morton, 1983) and 718 m (Joshi et al., 1997), which is nearly double the mean annual precipitation. The site has been farmed for >60 yr under a crop (usually wheat [Tritcum aestivum L.])–fallow rotation, with crops grown on odd-numbered years and fallow in even-numbered years (Dyck et al., 2003). In a 2-yr crop–fallow rotation, crops were planted and harvested in a 4- to 5-mo period (May–September) and then left fallow for 19 to 20 mo.
Plot Layout and Position of Core Samples
The strip that had received a total of 1.54 kg Cl– m–2 (3.36 kg KCl m–2) was core sampled extensively using a drill rig in the summer of 2000. Locations and elevations of the plot and all of the undisturbed sample cores were determined with a differential GPS (geographic positioning system) and latitude and longitude values were converted to UTM (universal transverse mercator) values with units in meters (Northing and Easting). A 90-m FT (field-scale transect), along the 1.54 kg Cl– m–2 strip was core sampled every 1 m in 2000 (Fig. 1). The southernmost point of the FT was fixed to a Northing of 0 m and an Easting of 0 m. The relative surface elevation at point 0,0 was set to 0 m. The KCl was applied in two passes of a fertilizer spreader. To avoid a position of gap or overlap, the sample transect (Easting = 0 m) was positioned 1 m off-center of the original KCl strip (Fig. 1). The 10-m-long PT (pedon-scale transect), sampled within the FT between Northings of 7 and 17 m, was reported in detail by Dyck et al. (2003, 2005). A second 10-m-long IT (intensive transect) was sampled at the bottom of the slight convex along the FT, between Northings of 37 and 47 m. Despite the concave shape at the IT, water did not collect there during spring melt or large rainfall events. A total of 51 undisturbed cores was taken at 0.2-m intervals along each of the PT and IT 10-m-long transects (Fig. 1). Outside of the original KCl experimental plot, 25 m east of the PT, a core was taken to characterize the background (unfertilized) Cl– concentrations in the soil (Fig. 1).
Coring
All of the cores, 53 mm in diameter, were taken to approximately 6 m in depth, using a truck-mounted drill rig–press. A few of the cores were not as deep due to problems with the drilling equipment or because stones or impermeable soil were encountered at depth. The cores taken in 2000 were taken approximately 10 mo after the previous harvest, during a fallow year. The 2001 core samples were collected approximately 15 mo after that, following a crop harvest.
Cores were examined in the field and the depths and thicknesses recorded for horizons, depositional layers, and the occurrence of pedogenic carbonates. Core morphology was classified according to the Canadian System of Soil Classification (Agriculture Canada Expert Committee on Soil Survey, 1987). Cores taken in 2000 were then sliced into 0.10-m increments, bagged, and labeled. The additional cores taken in 2001 were sliced into 0.20-m intervals. Wet weights were recorded in the field. In the lab, samples were dried at 60°C for 48 h and dry weights were recorded for determining gravimetric and volumetric soil water content and bulk density. For this portion of the study, a total of 171 cores and 10 382 samples were taken.
Laboratory Analysis
Dried samples were ground and sieved to generate aggregates that passed through a 2-mm sieve. For each sample, a 15-g subsample was combined with 30 g of water and shaken for 1 h. Suction filtration was used to obtain the water extracts containing the soluble salts. Electrical conductivity of the 2:1 water/soil extract was measured using a conductivity cell. Extract Cl– concentrations were determined colorimetrically on an autoanalyzer (Tel and Heseltine, 1990), converted to soil Cl– (in kilograms Cl– per kilogram of soil) and then multiplied by the sample bulk density (in kilograms per cubic meter) to determine the resident soil Cl– (in kilograms of Cl– per cubic meter of soil). The resident soil Cl– was corrected by subtracting the layer background resident soil Cl– for each sample, given by Dyck et al. (2003). Error may result from 2:1 dilutions at very low Cl– concentrations as the detection limit is approached.
Solute (Chloride) Transport
Soil water content vs. depth and resident soil Cl– concentration vs. depth BTCs (breakthrough curves) were created for each individual core sampled. In addition, average soil water content profiles and Cl– BTCs were calculated for each of the three transects (FT, PT, and IT) by averaging for each depth increment along the given transect. The 95% confidence intervals were also calculated from this data. The studies by Dyck et al. (2003, 2005) indicate that, in the PT transect, the initial movement of the surface-applied Cl– through the root zone was relatively quick, with an average travel depth of approximately 1.3 m 4 yr after application. This was followed by very slow downward vertical movement (average travel depth of 1.68 m 34 yr after application) and significant horizontal movement (0.76 m). Thus, depth BTCs of a single core gave effective downward (vertical) solute (Cl–) transport properties but did not quantify all of the transport that has occurred. Nevertheless, the width of the application strips (6.1 m) and the dominant role of root zone soil water recharge in downward transport indicated that individual and average (along the transects) depth BTCs should give good estimates of effective downward (vertical) solute (Cl–) transport properties and their spatial scale relationships.
Moment analysis (Jury and Roth, 1990) was used to calculate mass recovery, MR (kg m–2), mean vertical travel depth, E[z] (m), and travel depth variance about the mean, V[z] (m2) for each BTC using
 | [1] |
 | [2] |
 | [3] |
For the maximum horizontal scales, i.e., the transect lengths of the PT, IT, and FT, values of E[z] and V[z] were calculated. For mass-weighted values, average values of Cl– for each depth increment (i.e., E[C(z)]) were calculated first and then moment analysis was performed on the average Cl– BTC.
The value of V[z] as a function of horizontal spatial scale was calculated using the individual or local-scale BTCs at each of the three transects following the methodology proposed by Van Wesenbeeck and Kachanoski (1991). With this technique, a number of average BTCs were calculated for each increment (sampling increment) of horizontal scale along a transect. For example, the FT has n = 91 individual BTCs located 1 m (sampling interval) apart from each other. Averaging six adjacent individual FT BTCs was assumed to give one realization of a BTC for a horizontal scale of 5 m (the maximum distance separating the individual BTCs). A graph of V[z] as a function of horizontal spatial scale should indicate how dispersion was changing as a function of scale (Van Wesenbeeck and Kachanoski, 1991). A sill on the graph should indicate an estimate of where the transport depth variance was no longer a function of horizontal spatial scale and, therefore, the minimum plot scale necessary to account for horizontal variability in E[z] (Van Wesenbeeck and Kachanoski, 1991).
Topography and Catchment Area
Measurements of position and elevation were taken at the site for each of the core samples and for other locations in the surrounding area. These data were used to generate a map of surface topography (Golden Software Inc., 1999) using the kriging method. To estimate the catchment area along the FT, estimates for aspect were generated at 1-m-spaced nodes for an area with limits of Easting = –60 to 0 m and Northing = –10 to 100 m, indicated by diamond-shaped symbols on Fig. 2
. This area was chosen, as it was upslope and within a reasonable distance of the FT, to potentially allow surface redistribution of precipitation. Aspect was generated using a digital terrain modeling procedure (Golden Software Inc., 1999) and was given in azimuths, where an azimuth of 0 sloped to the north, an azimuth of 90 sloped to the east and an azimuth of 180 sloped to the south.
|
Spatial Redistribution of Surface Water
Additional soil cores (50-mm diameter, 1.6-m depth) were taken at 50 sampling locations after harvest in November 2001 (Fig. 2). The same sites were similarly sampled just before planting the next crop in late May 2003. All of the cores from the two sampling dates were sliced into 0.10-m increments and volumetric soil water content was determined from wet and oven-dry sample weights. The increase in soil water storage to a depth of 1.6 m was determined at each site and a contour map was generated to give an interpolated estimate of net surface water redistribution (Golden Software Inc., 1999; Fig. 2).
|
RESULTS AND DISCUSSION |
|---|
|
TOPEXECUTIVE SUMMARYABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
|
Layer Descriptions
The major soil layers recorded were similar to Dyck et al. (2003) and include Ap, Bm, Cca/Ck, indistinct varving (Varving 1), distinct varving (Varving 2), silt, and sand (Table 1; Agriculture Canada Expert Committee on Soil Survey, 1987). The varving layers consist of seasonally sorted glaciolacustrine deposits of alternating silt and clay. The Varving 1 layer consists of indistinct silt and clay banding, while the underlying Varving 2 is distinct. The Varving 2 layer is not apparent south of Northing = 11.4 m. At the bottom of the sample range is a layer of brownish-yellow fine to medium sand that extends to a depth of approximately 22 m (Menely, 1975). The water table sits within the sand layer, at a depth of approximately 15 m (Menely, 1975). For a more detailed description of the surface and underlying layers, see Dyck et al. (2003, 2005). Anisotropy, caused by microscale pore structure and layering (Zhang et al., 2003; Raats et al., 2004), such as that found in the varving layer, can influence water flow and promote lateral water flow.
|
|
Soil Water Content
The average soil water content profiles taken in 2000 at the time of Cl– sampling are given in Fig. 4a, 4b, and 4c for the FT, PT, and IT transects, respectively. Dyck et al. (2003) concluded that cumulative net infiltration by fall, winter, and spring precipitation was responsible for the increase in soil water storage in the upper 0.5 m of the PT profile. The sampling took place during a fallow year, approximately 10 mo after the previous harvest, so the depletion of water in the Cca/Ck horizon was attributed to the prior crop-year evapotranspiration and root-zone water use. The depth of root-zone water extraction in the PT was estimated to be 1.2 m, the start of the varving layers (Dyck et al., 2003). At the IT, cumulative net infiltration for the 10 mo since the previous harvest appeared to have penetrated to 1.2 m, compared with 0.5 m in the PT. If the lowest soil water content (0.15 m m–1) in the Cca/Ck horizon is used as a reference for the postharvest water content, then the subsequent net increase in soil water storage was approximately 90 mm for the IT and 60 mm for the PT. Although this difference is small (30 mm), it indicates that the surface topography, although slight, might have affected redistribution of precipitation. The soil water content increased significantly in the Varving 1 and Varving 2 layers of all profiles. Assuming somewhat similar matric pressure head vs. soil water relationships for similar layers, the larger depth gradient of soil water content within the varving (1 and 2) and silt layers of the IT transect suggested increased downward water flux compared with the PT transect.
Increased soil water storage (to a depth of 1.6 m) during the 19-mo fallow measurement period (November 2001 to May 2003) was, on average, 125 mm. Environment Canada climate data from the Harris meteorological station indicated that 676 mm of precipitation fell during the same 19-mo measurement period. The net storage as soil water of only 18.5% of the fallow-period precipitation, is similar to a number of studies on the Canadian prairies (Staple and Lehane, 1952; Campbell et al., 1987). Measured increases in soil water storage varied considerably across the catchment area and ranged from 36 to 245 mm (Fig. 2). The interpolated increases in soil water storage at each point (1-m spacing) along the FT ranged from 46 to 196 mm (Fig. 3b). The increase in soil water storage in the PT transect was approximately 118 mm, compared with 163 mm in the IT transect. As expected, the increased soil water storage (45 mm after 19 mo of fallow) in the IT compared with the PT was higher than the 30-mm difference after 10 mo of fallow (Fig. 4). The increased soil water storage in the IT was consistent with the significantly higher catchment area index and the deeper soil profile development for the IT transect (Table 1).
Solute Transport
The average Cl– BTCs are shown in Fig. 4d (FT), Fig. 4e (PT), and Fig. 4f (IT). A detailed discussion of the Cl– BTC for the PT has been given by Dyck et al. (2003). Briefly, at the PT, the peak Cl– concentration sits near the top of the Varving 1 layer, within the 1.3- to 1.4-m sample depth increment (Fig. 4e). The 95% confidence intervals indicate a very low degree of spatial variability. The Cl– MR (mass recovery) for the average PT BTC was 104% (Table 2). The average MR for the 51 individual PT cores was 103% (Table 2). The mean E[z] of Cl– for all PT cores was 1.70 m, with a V[z] of 0.86 m2, and the E[z] of the average PT BTC was 1.73 m, with a V[z] of 0.96 m2 (Table 2). The spatial variability of E[z] for individual locations within the PT was very low (CV [coefficient of variation] <5%), which was consistent with low estimated spatial variability of soil water storage (Fig. 3b). The interpolated increase in soil water storage varied from 111 to 125 mm across the PT. The PT BTC was skewed, with E[z] greater than the depth to peak Cl– concentration. The tailing of the Cl– BTC extended to 6.0 m. Dyck et al. (2003) concluded that the PT BTC reflects transport processes dominated by root-zone extraction and upward movement of water during cropping years, recharge of water during the fallow period, and transient deep net drainage averaging approximately 3 mm yr–1.
|
The coefficient of spatial variability of E[z] for individual locations in the IT was 9.8%, approximately double that of the PT (CV = 4.8%). This is consistent with the substantially higher spatial variability of the 19-mo increase in soil water storage in the IT (range 125–183 mm) compared with PT (range 111–125 mm). In the IT, there was a clear trend of increasing soil water storage from 125 mm of recharge at Northing = 37 m, to 181 mm of recharge at Northing = 47 m (Fig. 3b). Thus, although the IT is only 10 m long, it appears to be located in a transition area from lower to higher soil water storage. The southern section of the IT had soil water storage similar to the PT [with the PT E[z] = 1.7 m], which probably explains the bimodal nature of the IT BTC. The first IT BTC peak was at z = 1.6 to 1.7 m, similar to the PT E[z] = 1.7 m; the second IT BTC peak was much deeper at 2.3 to 2.4 m. This sharp transition from slow to faster water and solute transport has been observed under ephemeral ponds in topographic depressions and has been termed depression-focused recharge (Derby and Knighton, 2001; Hayashi et al., 2003). In these studies, conditions were often saturated due to accumulation of water in ponds and potholes occurring after spring thaw or large rainfall events. At our study site, the topographic differences were only slight and conditions remained unsaturated, yet the sharp transition in water and solute flow was nonetheless observed.
At the IT, E[z] ranged from 2.24 to 3.19 m for individual cores, with an average E[z] = 2.76 m and an average V[z] = 1.21 m2 (Table 2). The E[z] of the IT BTC was 2.74 m and the V[z] = 1.52 m2 (Table 2). The average travel depth of the IT BTC was 1.01 m deeper than the PT BTC, which is remarkably similar to the 1.0-m difference in depth to peak concentration between the PT and the IT (second peak). This suggests that during 34 yr (or 17 x 2-yr fallow recharge periods), the Cl– has been transported 1.0 m deeper, on average, in the IT than the PT. Assuming piston displacement, and a soil water content–transport volume of 0.25, deep drainage in the IT is estimated to be 7.5 mm yr–1 (or 15.0 mm per 2-yr fallow period) greater than in the PT. The PT deep drainage was estimated at 3 mm yr–1 (Dyck et al., 2003). The increase of 7.5 mm yr–1 in IT deep drainage was attributed to the influence of topography on the surface redistribution of precipitation and subsequently increased soil water storage in areas of high catchment area. If the measured increase in IT net soil water storage of 22.5 mm yr–1 (or 45 mm per 2-yr fallow period) is typical for other fallow periods, then crop evapotranspiration in the IT is estimated to be 15.0 mm yr–1 greater than evapotranspiration in the PT. Thus, the crops grown had the capacity to use some, but not all, of the redistributed water. This is consistent with the findings of de Jong and Rennie (1969) and Campbell et al. (1988), who demonstrated that water was the most limiting factor for plant growth in semiarid regions and that crops generally use most of the water that is available, which can buffer the impact of variable surface water redistribution. These studies suggest that the buffering effect was nonlinear and was reduced at greater amounts of available water.
The higher average and spatial variability of soil water storage in the IT than the PT is consistent with the MR of the two transects. The IT MR averaged 78%, compared with 103% for the PT (Table 2). The SD of the MR for the IT was 48%, compared with 11% for the PT. Within the IT, the MR was highly negatively correlated (r = –0.85, p < 0.001) to increased soil water storage from redistributed surface water. The decreased MR was probably from the increased deep drainage and lateral flow from layers. Analysis of two-dimensional transport in relation to subsurface layering was beyond the scope of this study.
The field-scale BTC reflected the combined variability of all transport processes and MR. It showed that the greatest concentration of Cl– had accumulated in the center of the Cca/Ck horizon (Fig. 4d). The FT (Fig. 4d) and PT (Fig. 4e) BTCs had a similar shape but there were several key differences. The thicker tailing on the FT was indicative of a greater amount of deep transport at the field scale than the PT. The 95% confidence intervals showed a greater variability in the Cca/Ck horizon for the FT. The confidence intervals indicated that there was a very small degree of variability in the Cl– concentration from the surface to the bottom of the Bm horizon and from the center of the Varving 2 layer downward. The FT E[z] ranged from 1.43 to 3.17 m, with an average local-scale E[z] of 2.03 m and a SD of 0.41 m (Table 2). The E[z] of the field-scale BTC was 1.92 m (Table 2). Local-scale V[z] ranged from 0.27 to 1.85 m2, with an average of 1.17 m2 (Table 2), and the field-scale V[z] was 1.26 m2 (Table 2).
The mass recovery MR of the Cl– was highly variable across the FT. This reflects a number of factors including variability of background levels of Cl–, three-dimensional transport of Cl–, and likely variability in the initial application rate. The initial application of the tracer was performed by two separate side-by-side passes of the surface spreader and overlap of applied Cl– probably occurred in the center of the strip. The transect location was offset from the center of the strip by 1.0 m to minimize the influence of overlap, but it might still possibly be a factor. The MR of the field-scale BTC was 144% and the average MR of the 91 FT cores, or local-scale mass recovery, was also 144%. Cores from individual locations had MR values ranging from 25 to 276% (Table 2). The field-scale MR was much greater than at both the PT and IT, indicating that, outside of the two 10-m transects, there were many locations or cores with very high MR.
Catchment area (five-core floating-point averages), elevation, increased soil water storage, E[z], and MR were compared with position along the FT (Fig. 3). At the center of the FT, where elevation was lowest, catchment area and increased soil water storage were high, E[z] was deep, and MR low. A correlation matrix was calculated for the 91 FT points, comparing elevation, catchment area, increased soil water storage (19 mo), 34-yr Cl– MR, E[z], V[z], and depth to peak Cl– concentration (Table 3). Many of the relationships were significant (p < 0.001), demonstrating the strong spatial covariance between these variables. The relationships between E(z), MR, and surface elevation and increased soil water storage indicated that surface topography controlled surface water redistribution, which, in turn, drove convective transport. The significant spatial correlations between increased soil water storage measured during a single 2-yr fallow period and Cl– transport properties (r = 0.65 for E[z], r = –0.74 for MR) after 34 yr also suggested that the measured water redistribution pattern was typical and recurring, i.e., time stable. The slope of the linear regression relationship between E[z] (after 34 yr) as a function of interpolated increased soil water storage for a 2-yr fallow period was 7.5 m depth m–1 water storage. If the measured spatial distribution of increased soil water storage was typical for fallow periods during the 34 yr, and the water transport volume of the soil was estimated at 0.25, then the regression slope suggests that, during 34 yr, 
10% of increased soil water storage from surface redistribution ended up as deep drainage and 90% was used by the crop.
|
|
|
SUMMARY AND CONCLUSIONS |
|---|
|
TOPEXECUTIVE SUMMARYABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
The Ap horizon thickness was similar throughout the study site; however, the Bm and Cca/Ck horizons were more developed in a slight depression than on a nearby level area. After 34 yr, the mean E[z] of surface-applied Cl– on the slight knoll was 1.70 m below the surface, while it was 2.76 m in the very slight depression, only 20 m away, with a grade between transect centers of just 1.8%. After a 19-mo fallow period, there was considerably more redistributed surface water stored in the soil profile in the very slight depression than on the nearby level area, despite the fact that there was only a small difference in elevation at the two 10-m transects. The spatial pattern of increased soil water storage (from one fallow period) was significantly correlated to solute transport properties (from 34 yr or 17 fallow periods). Deep drainage in the slightly concave lower slope IT was estimated at 10.5 mm yr–1, compared with 3 mm yr–1 in the slightly convex upper slope PT. The regression of the increase in soil water storage vs. E[z] across the FT suggests that 90% of redistributed surface water ending up as stored soil water was used by plants and 10% contributed to increased deep drainage.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPEXECUTIVE SUMMARYABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
. Water Resour. Res. 21:457–464.This article has been cited by other articles:
|
|
 |
|
  B. C. Si and E. de Jong Determining Long-Term (Decadal) Deep Drainage Rate Using Multiple Tracers J. Environ. Qual., October 16, 2007; 36(6): 1686 - 1694. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. L. Corwin, J. Hopmans, and G. H. de Rooij From Field- to Landscape-Scale Vadose Zone Processes: Scale Issues, Modeling, and Monitoring Vadose Zone J., March 8, 2006; 5(1): 129 - 139. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||
