SSSAJ Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (9)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Gómez, J.A.
Right arrow Articles by Fereres, E.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Gómez, J.A.
Right arrow Articles by Fereres, E.
Agricola
Right arrow Articles by Gómez, J.A.
Right arrow Articles by Fereres, E.
Related Collections
Right arrow Hydraulic Conductivity
Right arrow Soil Methods/Instrumentation
Right arrow Other Cropping Systems
Right arrow Tillage
Soil Science Society of America Journal 65:291-299 (2001)
© 2001 Soil Science Society of America

DIVISION S-1-SOIL PHYSICS

Analysis of Infiltration and Runoff in an Olive Orchard under No-Till

J.A. Gómeza, J.V. Giráldezb and E. Fereresb,c

a USDA-ARS-MWA, National Soil Erosion Research Lab., 1196 Soil Building, Purdue Univ., West Lafayette, IN, 47906
b Dep. de Agronomía, Univ. de Córdoba, Apartado 3048, 14080, Córdoba, Spain
c Instituto de Agricultura Sostenible, Consejo Superior de Investigaciones Científicas, Apartado 4084, 14080, Córdoba, Spain

Corresponding author (ag1fecae{at}uco.es)


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Four infiltration techniques (falling head, ring, rainfall, and tension infiltrometer) were used to determine the saturated hydraulic conductivity, Ks, and the wetting potential front, hf, of the Green–Ampt model. Water release curves from soil cores were also used for estimating hf. The objective was to compare the performance of the different techniques for the assessment of infiltration in a no-tillage olive (Olea europaea L. subsp. europaea) orchard. Measurements were performed in two areas of the orchard, below canopy (C) and interrow among trees (IR). With the exception of the tension infiltrometer, all techniques showed significant differences in Ks and hf between C and IR areas, attributed to different compaction. Differences in Ks among techniques were within the range observed previously. The hf estimated from the falling-head technique was significantly higher than that measured with the other techniques. The discrepancies in the results obtained with the tension infiltrometer were attributed to insufficient time of measurement, leading to recommendations for a different field procedure and analysis of this technique. To assess the use of the techniques described above for the characterization of plot infiltration, rainfall and runoff were measured in a 128-m2 plot. A numerical model was then used to predict runoff using the infiltration measurements. The results showed that runoff prediction is improved when different values of Ks and hf are considered for the C and IR areas instead of a single average value. The numerical analysis of the effects of tree arrangement on runoff prediction from infiltration measurements indicated that if the trees were placed along the contour lines, runoff would decrease relative to the standard tree arrangement.

Abbreviations: C, below canopy • IR, interrow among trees


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
MORE THAN 2 MILLION HECTARES of olives, an evergreen tree crop, are grown in Spain, mostly under rainfed conditions. In many areas of low rainfall, adequate water supply to the trees has been provided by a combination of sparse plantings and mechanical weed control. Traditionally, many olive plantations are located in shallow soils on sloping ground, a common situation in Southern Spain (Andalucia). With the advent of mechanized tillage, tillage passes have been carried out more frequently and thoroughly, leading to severe erosion problems in many areas. Soil losses of >100 Mg ha-1 have been reported. Such losses have increased the need to develop measures that reduce erosion, and alternatives to conventional tillage have been sought recently. The use of cover crops and minimum and no-tillage techniques have been investigated with the goal of increasing rainfall infiltration and decreasing runoff and erosion.

Cover crops increase infiltration by eliminating surface crusting due to raindrop impact on the soil, by increasing macroporosity through root channeling, and by enhancing biological activity. For the olive, an evergreen rainfed crop, the major disadvantage of cover crops is their water consumption, which is in direct competition with tree water uptake. Although winter rainfall often exceeds evaporative demand in most areas of Spain, the variability in rainfall occurrence and the periodic droughts make cover crops a high risk option for erosion control in olive groves due to yield losses. Alternatively, no-tillage soil systems have been introduced in olive orchards during the last two decades. Presently there are {approx}40000 ha of clean-cultivated, no-tillage orchards. However some orchards have been reverted to conventional or to minimum tillage because of real or perceived problems of reduced infiltration rates caused by no-tillage. The impact of these techniques on runoff generation, soil losses, and olive yield for given topography, soil, and climatic conditions remains an open question in Southern Spain. The soil heterogeneity of the areas where olives are grown is responsible, in part, for that situation.

Simulation models have been proposed to analyze infiltration under a wide range of conditions. Infiltration models such as the empirical Kostiakov model (Kostiakov, 1932) and the more physically based models of Smith and Parlange (1978), Philip (Philip, 1957), or the Richards equation (Richards, 1931) have been used. The Green and Ampt model (Green and Ampt, 1911) is an approximate description of one-dimensional water infiltration into the soil. Its relative simplicity, physical basis (Morel-Seytoux and Khanji, 1974), and its extended applications (e.g., Mein and Larson, 1973; Vandervaere et al., 1998) suggested its convenience. The two parameters of the Green and Ampt model, effective hydraulic conductivity, Ks, and the wetting potential front, hf, can be estimated from different techniques. Ring infiltrometers (Bouwer, 1986), tension infiltrometer (White and Sully, 1987), rainfall infiltrometer (Connolly et al., 1991), and falling-head infiltrometer (Philip, 1993) are some of the techniques used or described as useful for the determination of Ks and hf.

Infiltration in orchards is expected to be spatially variable because of tree effects above and below ground. Gras and Trocmé (1977) observed that infiltration rates were higher under apple trees (Malus domestica Borkh.) than in the tree rows. Differences in infiltration between areas below trees and those in the interrow have been described in olive orchards (Vanderlinden et al., 1998; Gómez et al., 1999). The generally higher infiltration rate of the soil below the canopy has implications in runoff generation. There is no information on the influence of such differences in infiltration rates on average infiltration, and runoff generation in olive orchards. For natural vegetation in arid zones, Morin and Kosovsky (1995) proposed that the underbrush areas act as sinks for the runoff generated in the open spaces and thus can have an important role in infiltration depending on their number and distribution.

It is important to develop methods that characterize orchard runoff in order to assess the effects of changes in soil management on infiltration at the plot scale. These methods should play a major role in the calibration of the infiltration models mentioned above. Investigations on infiltration in olive orchards have been performed using different infiltrometer designs. Few of them used two or more different techniques, making it difficult to compare results from different sources. Many authors have discussed the problems of comparing different infiltration techniques that are based on different principles and that explore different soil volumes. Mohanty et al. (1994b) found a coefficient of variation (CV) of 102% among the geometric means of Ks obtained by the Guelph permeameter, the velocity permeameter, the disk permeameter, the double-tube method and the constant-head method using soil cores. Vanderlinden et al. (1998) compared the constant-head well permeameter, the falling-head lined borehole permeameter, the twin rings, and the tension infiltrometer using two disks of different radii, and found a CV of 166% among the geometric mean of Ks using the different techniques. These authors showed how different methods give variable results under various soil types and field conditions. Even when the same technique is used, the size of the infiltrometer has an influence on the results. Zobeck et al. (1985) and Shouse et al. (1994) showed a general trend towards higher saturated hydraulic conductivity as the measured area increased.

We performed infiltration measurements in an olive orchard under a long-term soil management experiment with the following objectives:

  1. Compare the performance of four different infiltration techniques using falling-head ring, tension, and rainfall infiltrometers.
  2. Compare the above four small-scale infiltration techniques with plot-scale measurements of infiltration obtained from rainfall and runoff measurements.
  3. Evaluate the influence of tree arrangement on the interpretation of small-scale measurements of infiltration and their use for runoff prediction.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Site Characteristics
The experiment was conducted in an olive orchard located in Santaella, southern Spain, (37.8° N, 4.8° W). The soil is classified as Calcixerollic Xerochrept of clay loam texture, with a surface slope of 2 to 7%. Measurements were performed on plots that are part of a soil management experiment comparing no-till with conventional tillage, which started in 1983 (Gómez et al., 1999). The no-tillage treatment is kept weed-free with applications of the herbicide simazine (6-chloro-N,N'-diethyl-1,3,5-triazine-2,4-diamine). Trees were spaced in an 8 by 8 m grid, aligned in the slope direction, and mean tree canopy radius was 2 m (Fig. 1) . Four replicate plots of 24 by 24 m in the no-till treatment were used for the point measurements of infiltration, while in one additional replication, a single plot of 16 by 8 m was isolated and instrumented for runoff collection.



View larger version (35K):
[in this window]
[in a new window]
  		Fig. 1. Description of the experimental plot with definition of interrow (IR) and below-canopy (C) zones

 
Measurements of infiltration were carried out in two distinct zones: (i) beneath the tree canopy (C), in an intermediate distance between the tree trunk and the edge of the canopy vertical projection, and (ii) in the middle of the tree interrow (IR), in visually inspected nontracked areas. Four replications were taken for any zone and infiltration technique combination, in four different zones within the orchard. All measurements were performed between the fall of 1996 and the spring of 1997.

Small-Scale Measurements of Infiltration
Falling-Head Infiltrometer
Thomas Dunne and Elisabeth Saffran used falling-head lined borehole permeameter to measure infiltration in hydrologic studies of the Amazon Basin (Philip, 1993). In this method, hereafter referred to as falling-head infiltrometer, water infiltrates into the soil from a tube inserted in the surface with falling water head. Philip (1993) presented an approximate method to calculate the saturated hydraulic conductivity, Ks, and the wetting front potential, hf, from the falling-head infiltrometer data assuming the Green and Ampt model. The application of this method to some soils of medium to fine texture generally requires a long measurement time, and often shows nonconvergence (De Haro et al., 1998) with the numerical scheme proposed by Philip (1993). To avoid this problem, Ks and hf were calculated here by numerical fitting of the observed evolution of the water head with time, as described in Eq. [1].


(1)
where t is time since the start of water falling (T); ro is one-half the internal tube radius (L); Ks is saturated hydraulic conductivity (L T-1); X is the dimensionless decrease in head water level, according to Eq. [2], modified from Philip (1993); and A is a parameter related to hf, wetting front potential (L), Eq. [3]:

(2)

(3)
where Do is the initial water depth (L), D is the water depth (L), and {Delta}{theta} is volumetric soil water increment, saturated minus initial volumetric content (L3 L-3).

A tube of a 0.09-m internal radius, inserted at the 0.05-m depth into the soil, was used. Do was 0.15 m. Soil moisture was determined gravimetrically before and after any infiltration measurement and converted to volumetric water content using soil bulk density values.

Ring Infiltrometer
Rings of 0.28 m in diameter, inserted 0.05 m into the soil, were used with a constant head of 0.05 m. The cumulative infiltration data were fitted to the Green and Ampt model (Mein and Larson, 1973), by Eq. [4]

(4)
where I is cumulative infiltration (L), Hw is water depth above soil surface (L), and the other terms are as defined above. Soil moisture was determined gravimetrically before and after any infiltration measurement and converted to volumetric water content using soil bulk density. Visual inspection after the measurements revealed subsurface lateral fluxes from the edges of the rings. Equation [5] (Bouwer, 1986) was used to correct the overestimation caused by lateral flux from around the edge of the ring

(5)
where if is the final infiltration rate, hcr is the critical pressure head estimated as hb/2, hb is the air-entry head, and f(hcr, 2rc) is the function described by Bouwer (1986), with rc the radius of the radius of the ring infiltrometer. The air-entry potentials were determined in desorption experiments using Haines cups with undisturbed samples, 0.049 m in diameter and 0.02 m high, taken in areas close to the ring test. Four samples each from the C and IR zones were placed in the cups and sealed with a base of plaster of Paris. The water retention curve was determined for the 0- to 19.61-kPa range of matric potential. Equation [6] (Brakensiek, 1977) was used for calculating hf, where hb was derived from the moisture release curve (Brooks and Corey, 1964).

(6)

Tension Infiltrometer
Infiltration measurements were taken with a suction infiltrometer (Perroux and White, 1988) 0.25 m in diameter. There are several different techniques in which this infiltrometer can be used. We used that of Ankeny et al. (1991) with one single infiltrometer and successive measurements at different potentials in the same position, applying tensions of -1.47, -0.98, -0.49, and 0 kPa of water potential, in a wetting sequence. At each potential, steady-state infiltration was attained at about 60, 30, 30, and 10 min, respectively. From these steady-state infiltration rates, we calculated the hydraulic conductivity using the expression proposed by Ankeny et al. (1991). The Ankeny method was selected because of its simplicity, and because the number of negative hydraulic conductivity values is smaller compared with other methods (Logsdon and Jaynes, 1993). It also required the sampling of fewer points than in other methods. The macroscopic capillary length, {lambda}c, was determined from the experimental data by numerical solution of the Eq. [7] (White and Sully, 1987).

(7)
where {psi}o and {psi}n are the maximum and minimum water potentials applied to the soil surface, and K({psi}) is the hydraulic conductivity as a function of soil water potential. From {lambda}c, the wetting front potential was computed using the approximate solution given by Eq. [8], with b value of 0.55 as suggested by White and Sully (1987).

(8)

We used a nylon net of 20 µm to separate the soil from treated sand placed to eliminate surface irregularities and to achieve good contact between the soil and the infiltrometer. The depth of sand layer ranged from 0.005 to 0.01 m. Soil moisture was determined gravimetrically before and after any infiltration measurement.

Rainfall Infiltrometer
We built a portable rainfall simulator constructed according to the guidelines of Peterson and Bubenzer (1986) to determine infiltration in square plots of 0.5 m. Metal boxes, 0.2 m in height, were driven 0.1 m into the soil and sealed. Rainfall was simulated with a wide-angle nozzle (8W, Spraying System Co., Wheaton, IL) located 1.5 m above the soil surface. This provided a rainfall intensity of 90 mm h-1 at 1.5-m height. The kinetic energy of the applied rainfall, estimated from the nozzle drop distribution and Kincaid's (1996) procedure was 1000 Jm-2 h-1, which is {approx}40% of the calculated energy for that intensity using the Wischmeier and Smith equation (Foster et al., 1981). Runoff hydrograph and infiltration were calculated by collecting runoff with bottles at 2-min intervals during the test that lasted 1.5 h. The Green and Ampt model, given by Eq. [4], with zero water depth above the surface, was fitted to the cumulative infiltration data. This analysis of the infiltration test by directly fitting the infiltration equation was selected among others, for example, the time to incipient ponding (White et al., 1989) because of consistency with the ring infiltrometer analysis.

Plot Measurements of Infiltration
A 16-m downslope and 8-m-wide plot, with 5% slope in the direction of the greater dimension, containing two olive trees in its center and aligned in the slope direction, was isolated with shallow levees and the runoff collected at its end. Runoff was measured with a tipping-bucket similar to that described by Barfield and Hirschi (1986) and recorded in a datalogger (CR21X, Campbell Scientific, Logan, UT). A rain gauge (Campbell ARG 100) was installed nearby to collect rainfall at 1-min intervals. Soil moisture was recorded with 0.2-m-long time domain reflectometry probes (Soil Moisture Corp., Santa Barbara, CA) installed beneath the tree and in the row zones (two replicate measurements on each location).

Cumulative rainfall and runoff were derived from rainfall and runoff records. We calculated apparent plot infiltration rate with Eq. [9] (Yu et al., 1998), which gives the water balance for a time interval j

(9)
where ij is the apparent infiltration rate of the plot area (L T-1), pj rainfall intensity (L T-1), qj the average runoff rate at the plot exit (L T-1), ej evaporation rate (L T-1), and {Delta}Sj the change in surface storage (L) during time {Delta}t. Evaporation is considered negligible, and the storage term decreases and may be ignored if data are accumulated at large time intervals. We used a time interval of 10 min to ensure this condition and to reduce the error induced by not considering runoff travel time within the plot (Yu et al., 1998).

Model Development for the Plot Analysis of Infiltration
An infiltration–runoff model was developed to evaluate the performance of the small-scale measurements of infiltration at the plot scale. Briefly, the model divided the microwatershed into square cells of variable size as defined by the user. It then calculated the infiltration, for any cell, using the Green and Ampt model with the time compression approach (Reeves and Miller, 1975). The model allows a different hydraulic conductivity value for each cell.

Surface runoff is computed by routing the excess water in the cell using the slope and aspect of each cell. Assuming uniform flow approach, Manning's equation, and the continuity equation (Eq. [10]) are solved at each cell at 1-min time step. For this, a fourth-order Runge-Kutta scheme with adaptive step-size control (Press et al., 1986) was used.

(10)
where S (L), is the water level above surface, t (T) is time, qe (L T-1) is upstream runoff into the cell, p (L T-1) is rainfall intensity at the soil surface, i (L T-1) is infiltration rate of the cell, and qs (L T-1) is runoff produced in the cell. The model allows for the runoff and run-on between cells.

The parameters needed for the model were calibrated with three different methods. First, we used saturated hydraulic conductivity and wetting front potential based on the field measurements. Surface storage was estimated with the Onstad (1984) model for a random roughness of 0.002 m measured with the chain technique of Saleh (1993). The rainfall storage capacity of the olive canopy was taken as 1.2 L m-2 derived from a previous rainfall interception experiment (Gómez et al., 2000). Second, field Ks measurements were used in all cases, while the wetting front potential was obtained from the laboratory determinations on soil cores. Third, Ks was calculated by minimizing the least square error between observed and calculated event runoff at the plot level, assuming the hf values obtained from soil cores.

The model was used to predict plot runoff with two different approaches. One used a single average value of Ks and hf for the whole plot, which were derived from averaging values obtained in the IR and C areas (Fig. 1) according to the relative extension of each area. The other used different Ks and hf values for the C and IR areas obtained from field measurement. Finally, we analyzed the effect of tree arrangement under the latter approach.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Small-Scale Measurements of Infiltration
Table 1 presents the values of Ks and hf parameters of the Green and Ampt equation as determined by all methods used. Regression analyses of the falling-head, ring, and rainfall infiltrometer experimental data against calculated values gave correlation coefficients of 0.92 or higher.


View this table:
[in this window]
[in a new window]
  		Table 1. Geometric mean values from infiltration measurements with coefficients of variation given in parentheses

 
In all methods, except for the tension infiltrometer, the C areas had smaller hf than the IR zones. Other soil properties at the experiment, which were previously reported (Gómez et al., 1999) showed greater bulk density for the IR than for the C zone. That was attributed to the effect of traffic related to agricultural operations, which is restricted mainly to the IR areas. Since the initial soil moisture contents in the C and in the IR zones were not statistically different at the test, our interpretation is that the difference in hf was caused by the different porosity of the two zones. The smaller porosity results in greater influence of the capillary forces reflected in the hf values. This agrees with results reported by Laliberte and Brooks (1966) that showed how the bubbling pressure (related to hf by Eq. [6]) increased with decreasing porosity. The lack of significant differences between tension infiltrometer measurements taken in the C and IR zones may be attributed to an overestimation of the hydraulic conductivity in the IR areas that, according to Eq. [7] and [8], leads to an overestimation of hf. Table 1 also shows that hf values depend largely on the method used for its calculation. The largest discrepancies in hf are observed when the values derived from the Philip (1993) analysis of falling-head infiltrometer data are compared with the other methods We have not found an explanation for the comparatively large hf values obtained with this method. Such values are greater than the average values reported by Rawls and Brakensiek (1989) for soils of similar textural class. This apparent overestimation of hf suggests the use of an alternative technique for checking hf values. Apart from the falling-head infiltrometer, the differences in hf values are not surprising, since we are comparing different procedures, that explore different volumes of soil, use different numerical analysis, and probably cause variable soil disturbance.

Figure 2 depicts the water retention curves for the C and IR zones. Both curves were parallel, showing that the IR zone held more water than the C zone at any tension. According to the different compaction of both zones (Gómez et al., 1999), the curve for C should cross the IR curve at potentials approaching zero (Horton et al., 1994). Reicosky et al. (1981) found such a pattern for soil cores packed at 0.99 and 1.18 Mg m-3. It is possible that the size of the soil cores in our study were too small compared with the soil volume explored by the different infiltrometers, or that the core sampler used for bulk density measurements in Gómez et al. (1999) and that size was not adequate for representing the soil structure. Another possible reason is that the number of replications was small and inadequate for characterizing both zones.



View larger version (14K):
[in this window]
[in a new window]
  		Fig. 2. Moisture release curves for soil samples taken in the interrow (IR) and below-canopy areas (C). Bars indicate standard deviation

 
All methods gave higher Ks values in C than in IR (Table 1), which may be attributed to the different compaction of both zones, as stated above. There were nonsignificant differences between Ks values determined by the falling-head and the ring methods. The tube for the falling-head infiltrometer was inserted at the 0.05-m depth, while the other methods measured surface properties. Thus, we concluded that surface infiltration was not controlled by a surface seal developed over a more permeable soil profile and that, even after nearly two decades of non-tillage, the general compaction of the soil profile (Gómez et al., 1999) seems to be the main factor reducing Ks. The Ks values obtained in IR using the rain infiltrometer were significantly higher than those by the first two methods. The higher Ks in the rain infiltrometer could have been caused by the increased likelihood of encountering cracks or large pores in a surface area larger than that sampled by the other methods (Gupta et al., 1993), also shown by Shouse et al. (1994). It is puzzling that the higher Ks occurred only in IR; it is possible that the insertion of the metal boxes, used to isolate the 0.5 by 0.5 m area, in the more compacted IR, led to higher soil disturbance and thus higher infiltration, in comparison with other techniques, than in the C zones.

Measurements performed with the tension infiltrometer yielded hydraulic conductivity at zero potential much higher than those observed with the other methods in the IR zone. In fact, values for unsaturated hydraulic conductivity were within the range of the Ks observed in the IR with the other infiltrometers. Despite that field measurements indicated a steady infiltration rate, its value was less than that predicted by theoretical models for tension infiltrometers (Warrick, 1992). Warrick (1992) suggested that insufficient equilibrium times yield higher values of Ks. Zhang (1998) showed that the determination of Ks at zero potential using the method of Ankeny et al. (1991) for short infiltration times can lead to relative errors as high as 60%. However, we have no clear explanation for the fact that the same nondisturbing technique did not show the same large differences between tension infiltrometer and the other methods in the C zone. We can only hypothesize that the greater importance of capillary forces in water infiltration in IR as shown by the hf measurements is translated into greater overestimation of hydraulic conductivity in IR as compared with C zones. Zhang (1997) proposed Eq. [11] to describe the infiltration, I, in time, t, under a disk infiltrometer.

(11)

The transition from capillarity to gravity-driven flow is determined by the values of coefficient A1, which characterizes the water flow due to the capillary forces, and A2, which characterizes the flow due to gravity. If the significance of the capillary forces in the C areas is small, as stated above, it is possible that the Ks values obtained from steady-state values at insufficient equilibrium times are similar to the Ks obtained by other methods (Table 2). On the contrary, for the same equilibrium infiltration time, there would be greater differences in the IR zone, due to importance of the term A1 in Eq. [11]. It appears that equilibrium times greater than those used in our study are a necessary condition for the correct application of this technique. Since the correct infiltration time is difficult to identify precisely-and it may be limited in the field by operational considerations—the use of procedures that correct the errors of not achieving steady state are advisable. Zhang (1998) proposed a method to correct the error due to not achieving the equilibrium infiltration time from modification of that used here, Ankeny et al. (1991).


View this table:
[in this window]
[in a new window]
  		Table 2. Analysis of variance of Ks measurements analyzing below canopy (C) and interrow among trees (IR) areas separately. Methods followed by the same letter are not statistically different at the 0.95 probability level using Tukey's test (Steel and Torrie, 1960)

 
The CV of Ks obtained in the IR area at -1.49 kPa was greater than that obtained for potentials closer to saturation. This is contrary to experience since CV is supposed to decrease for less saturated tensions due to the removal of larger pores from the flow process. No explanation for this result has been found, although a similar trend can be detected in previous studies using the same procedure (e.g., Hussen and Warrick, 1993, Table 2). Their study showed the Ankeny et al. (1991) method to be accurate and stable in the determination of the hydraulic conductivity, K({psi}), as a function of soil water potential, {psi}.

In general, the differences among the Ks values of Table 2 obtained by different infiltration techniques were smaller than those reported by others. The CV for the geometric mean of Ks [considering K({psi}) at 0 kPa as comparable with Ks] obtained for all different techniques was 77.5, 127, and 17% for the IR and C areas when both zones were independently analyzed. Despite the relatively small number of replicate measurements taken with any method, all of the methods characterized the differences between C and IR. Considering the large coefficient of variation normally observed in Ks measurements using the same technique (e.g., Mohanty et al., 1994a) due to spatial variability of soil water properties, it seems that any of the techniques used in this work could characterize the infiltration rates of non-tillage olive orchards. One should not expect better agreement among methods than the agreement found in this work since different techniques involve differences in sample size and analysis. Aside from operational considerations, one approach to evaluate the suitability of the various small-scale measurements of infiltration methods is to establish a comparison with independent measurements at the plot scale, such as the runoff measurement in closed plots. Such plots are widely used for determining the relationships between rainfall, runoff, and infiltration, and for model calibration and validation (Yu et al., 1998).

Plot Measurements of Infiltration
Table 3 presents the relation between precipitation and runoff obtained in the closed plot during winter of 1997. Runoff never exceeded 35% of rainfall even though the soil water content during that time was relatively moist and fairly constant around 0.27 cm3 cm-3, as indicated by soil water content measurements (data not shown). Apparent infiltration rate was computed using Eq. [8] from rainfall–runoff records and plotted in Fig. 3 as a function of precipitation intensity. Apparent infiltration rate increased with rainfall intensity (Fig. 3). Morin and Kosovsky (1995) proposed that relationships such as that of Fig. 3 indicate the presence of sink zones for runoff within the experimental area. In our situation, Fig. 3 suggests that, as rainfall intensity increases, the areas of higher Ks under the trees are acting as sinks where the runoff from the interrow upslope area infiltrates.


View this table:
[in this window]
[in a new window]
  		Table 3. Rainfall and runoff records for the 8 by 16 m plot obtained during winter 1997

 


View larger version (15K):
[in this window]
[in a new window]
  		Fig. 3. Apparent infiltration rate from the rainfall runoff rates for the 8 by 16 m plot as a function of rainfall intensity

 
Use of Small-Scale Measurements for Evaluation of Plot Infiltration
Use of an Average Ks Value
The geometric means of Ks and hf measured by the four methods as discussed in the above sections were used to calculate plot infiltration and runoff, using the numerical model described above, and the results were compared with measured runoff for all rainfall events presented in Fig. 3. Poor predictions of plot runoff were observed when single average Ks and hf values were used in the model. Figure 4 presents the results using Ks measured with the Dunne infiltrometer, the one method among the four tested here that had the best agreement between observed and calculated runoff. Figure 4 also shows model behavior when a numerical adjustment of Ks and the hf from soil cores was used in its calibration. It is apparent from Fig. 4 that predicting plot runoff using a single average Ks value is inadequate, as shown previously by Morin and Kosovsky (1995) for areas under natural vegetation.



View larger version (21K):
[in this window]
[in a new window]
  		Fig. 4. Relationships between observed and calculated runoff depths using an average single value of Ks for the whole plot. This value was obtained from field measurements using the falling-head infiltrometer (Curve B) or numerically by the minimizing square error between calculated and observed runoff (Curve A)

 
Use of Two Average Ks Values
The use of the geometric means of Ks and hf for the C and IR orchard zones to model runoff, gave much better agreement with the measured runoff than the approach of using a single average value. Table 4 shows the linear regression between the simulated and the observed runoff events, and Fig. 5 shows the calculated runoff using the Ks values obtained from the falling-head and ring infiltrometer data, and the hf calculated from the soil cores. The latter method and the ring method gave adjustments that did not depart from the 1:1 line. The two other methods for Ks determination (rainfall infiltrometer and tension infiltrometer at zero tension), when used in simulations, yielded very small or no runoff amounts for any of the rainfall events registered. This was mainly due to the large Ks field values reported in Table 4, and partially to the large hf estimations. The lowest root mean square error values were obtained by adjusting Ks and hf values using the least square error technique between measured and calculated runoff. The Ks values that gave the best fit were KsIR = 1.0 mm h-1 and KsC = 47 mm h-1, which are similar to the Ks values found with the falling-head and the ring infiltrometer techniques (Table 1). Several other combinations of KsIR and KsC also gave good agreement, all of them showing much greater Ks values for the C area as detected in the field (Table 1). It appears that when a two-zone infiltration model is used, the Ks measurements using the falling-head and the ring infiltrometer methods gave predictions of plot runoff close to the observed values. This prediction was unbiased if we used the hf calculated from the soil cores. Connolly et al. (1991) successfully tested a similar approach for model calibration, by providing Ks from field measurements and hf from another source. Rainfall and tension infiltrometer have been sucessfully used in the calibration of rainfall–runoff models (Connolly et al., 1991; Vandervaere et al., 1998). The correction of the sources of error for the rainfall and tension infiltrometer in our work, discussed above, probably led to a better agreement between calculated and observed runoff when such techniques where used to calibrate the two-zone model.


View this table:
[in this window]
[in a new window]
  		Table 4. Summary of regression between calculated and observed runoff volumes for different values of Ks and hf obtained from different infiltrometers. m is slope and n is intercept

 


View larger version (24K):
[in this window]
[in a new window]
  		Fig. 5. Relationship between observed and calculated runoff depths using a calibration with infiltrometer (Curves B and C) or least square error (Curve A), always considering the interrow and below-canopy zones of different Ks

 
Effects of Tree Arrangement on Plot Runoff Predictions
The above analysis has shown how the C and IR zones interact substantially. An immediate consequence is that tree arrangement should have a significant impact on runoff generation. We carried out a numerical assessment of this phenomenon by simulating the two arrangements shown in Fig. 6 . Our evaluation computed the runoff events reported in Table 3 for both arrangements using the Ks values from the falling-head infiltrometer, the hf from the soil cores, and identical IR and C surface areas in both cases. The regression of the event runoff simulations obtained for both tree arrangements is

(12)
where RA and RB refer to the event runoff from the arrangements A and B of Fig. 6. The results show that runoff is decreased by about one-third when the tree rows are placed following contour lines. Obviously, in a real world situation, the overland flow direction is going to be strongly affected by local micro relief, and not only by slope, as in our model. Nevertheless, our simulations could be relevant for analyzing runoff from high density plantings, a recent trend where trees are planted on a 7 by 4 m spacing or closer, a situation where row direction would be critical in minimizing runoff on hill plantings.



View larger version (32K):
[in this window]
[in a new window]
  		Fig. 6. Description of tree arrangement used in numerical simulation of plot runoff. (A) Trees aligned in the slope aspect in an 8 by 8 m grid. (B) Trees aligned perpendicularly to the slope aspect in a 4 by 16 m grid

 

   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
Each of the four small-scale techniques used for measuring infiltration detected significant differences in Ks and hf between the IR and C areas in an untilled olive orchard. These differences were associated with differences in compaction between the two zones. The differences in Ks among techniques were within the range of that observed in previous works. The rainfall infiltrometer and the tension infiltrometer gave higher Ks values for IR areas as compared with the falling-head and ring infiltrometers. The higher measured Ks was attributed to insufficient infiltration time for achieving a steady infiltration rate in the case of the tension infiltrometer, and to the larger area sampled by the rainfall infiltrometer that increased the probability of encountering cracks or larger pores. The differences in measured hf values among different infiltration techniques were larger than those observed for Ks, with the values from the falling-head infiltrometer remarkably larger than those obtained by the other methods. Because of these large differences in hf, we suggest the use of core samples of a size comparable with that explored by the infiltration techniques, for an additional determination of hf.

If appropriate field procedure and improved analysis (Zhang, 1998) were used for the tension infiltrometer, and if the sample area for the rainfall infiltrometer technique was reduced, then the four techniques would be expected to give comparable results when used in field surveys in olive orchards. In the case of the falling-head infiltrometer, this agreement can be expected only if the infiltration rate is not limited by soil surface sealing.

Consideration of the two different infiltration areas, C and IR, and their interaction via runoff and run-on effects proved to be relevant in the analysis of small scale measurements of infiltration for determining runoff at the plot scale. In our study, the simulation of plot runoff considering these different zones gave a much better fit to the measured runoff than the use of a single average value of Ks and hf for the whole plot. It also allows for a more precise analysis of some important factors in the spatial arrangement of olive orchards. For example, numerical analysis of the effect of tree arrangement on runoff showed that it may have a significant impact on runoff generation and should be considered in new olive plantings.


   ACKNOWLEDGMENTS
 
We thank Dr. M. Pastor for use of the experimental area and Dr. L. Andreu for helping with the tension infiltrometer technique.

Received for publication October 25, 1999.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 




This article has been cited by other articles:


Home page
 Soil Sci.Home page
P. Romero, G. Castro, J.A. Gomez, and E. Fereres
Curve Number Values for Olive Orchards under Different Soil Management
Soil Sci. Soc. Am. J., September 28, 2007; 71(6): 1758 - 1769.
[Abstract] [Full Text] [PDF]

Home page
 Vadose Zone JHome page
C. M. Regalado, A. Ritter, J. Alvarez-Benedi, and R. Munoz-Carpena
Simplified Method to Estimate the Green-Ampt Wetting Front Suction and Soil Sorptivity with the Philip-Dunne Falling-Head Permeameter
Vadose Zone J., May 12, 2005; 4(2): 291 - 299.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (9)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Gómez, J.A.
Right arrow Articles by Fereres, E.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Gómez, J.A.
Right arrow Articles by Fereres, E.
Agricola
Right arrow Articles by Gómez, J.A.
Right arrow Articles by Fereres, E.
Related Collections
Right arrow Hydraulic Conductivity
Right arrow Soil Methods/Instrumentation
Right arrow Other Cropping Systems
Right arrow Tillage

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Crop Science
Journal of Natural Resources
and Life Sciences Education		 Vadose Zone Journal
Journal of Plant Registrations Journal of
Environmental Quality		 The Plant Genome