Furrow Irrigation Water-Quality Effects on Soil Loss and InfiltrationR. D. Lentz,* R. E. Sojka, and D. L. Carter
ABSTRACTIrrigation-induced erosion is a serious problem in the western USA
where irrigation water quality can vary seasonally and geographically.We hypothesized that source-water electrical conductivity (EC) andsodium adsorption ratio (SAR = Na/[(Ca + Mg)/2]°-s, where concen-trations are in millimoles of charge per liter) affect infiltration andsediment losses from irrigated furrows, and warrant specific consider-ation in irrigation-induced erosion models. On a fallow Portneuf siltloam (coarse-silty, mixed, mesic Durixerollic Calciorthid), tail-watersediment loss was measured from trafficked and nontrafflcked furrowsirrigated with waters of differing quality. Treatments were the fourcombinations of low or high EC (0.6 and 2 dS m'1) and low or highSAR (0.7 and 12 [mmolc L~ lfs). Slope is 1%. Twelve irrigations weremonitored. Each furrow received two irrigations. Main effects forwater quality, traffic, and first vs. second irrigations were significantfor total soil loss, mean sediment concentration, total outflow, netinfiltration, and advance time. Average tail-water soil losses were 2.5Mg ha-' from low EC/low SAR furrows, 4.5 Mg ha-' from low EC/high SAR furrows, 3.0 Mg ha'1 from high EC/high SAR furrows;and 1.8 Mg ha"1 from high EC/low SAR furrows. Elevating waterEC decreased sediment concentration from 6.2 to 4.6 g L~', butincreasing SAR increased sediment concentration from 6.2 to 8.7 gL"'. Net-infiltration decreased 14% in high SAR compared with lowSAR'treatments. Soil loss increased 68% for second irrigations, andnet infiltration fell 23% in trafficked furrows, but water-quality effectswere the same. Water quality significantly influenced infiltration anderosion processes in irrigated furrows on Portneuf soils.
OF THE ESTIMATED 250 MILLION HA irrigated world-wide5 at least 60% is surface irrigated. Soil erosion
from irrigation, especially furrow irrigation, contributesto nonpoint-source pollution (Hajek et al., 1990) and isa serious threat to crop productivity in many regions(Carter, 1993).
Agricultural research has focused primarily on rainfall-induced soil erosion, with comparatively little attentionto furrow-irrigation-induced erosion. A common as-sumption has been that erosion in rills is mechanisticallyequivalent to that in irrigated furrows. While shear pro-duced by concentrated flow causes soil detachment andentrainment in both, there are several important differ-ences: (i) rilh phenomenon often includes an additionalforce, raindrop impact, which detaches and transportsadjacent soil particles to the rill stream; (ii) a furrowstream initially advances over dry soil, resulting in rapidwetting and destabilization of dry, low-cohesion soilaggregates and increased furrow erosion losses (Kemperet al., 1985), whereas rill soils are prewetted by precipita-tion; (iii) downstream flow rates decrease in furrows aswater infiltrates, but increase in rain-fed rills owingmainly to tributary inflow, hence, furrow flow rates andpotential erosion losses are greater in the upper reachesof a furrow, not in the lower reaches as for rills; and
USDA-ARS, Northwest Irrigation and Soils Research Lab., 3793 N 3600E, Kimberly, ID 83341. Received 26 Sep. 1994. *Corresponding author([email protected]).
Published in Soil Sci. Soc. Am. I. 60:238-245 (1996).
(iv) salinity and sodicity of rainwater are uniformly low,while irrigation water quality can vary widely, geographi-cally and temporally, even within short distances andshort intervals.
Few studies have attempted to determine how irrigationwater quality influences furrow erosion, although thisinformation may be necessary to understand and modelerosion processes in surface irrigated systems.
Three main factors influencing furrow erosion are theshear stress of flowing water on the furrow perimeter,cohesivity of soil particles (which affects the stabilityand size-distribution characteristics of furrow soil), andstream transport capacity (Kemper et al., 1985; Troutand Neibling, 1993). Water quality may influence flowshear by controlling furrow intake and, hence, down-furrow flow rate. In soil column studies, SAR and ECof infiltrating water reduced soil permeability and infil-tration rate (Fireman and Bodman, 1939; Quirk andSchofield, 1955; Frenkel et al., 1978). The most signifi-cant impact has been shown to be on depositional sealformation (Shainberg and Singer, 1985; Brown et al.,1988). Soils were more sensitive to water quality impactswhen mechanical disruption (i.e., flow shear) accompa-nied water application (Quirk and Schofield, 1955; Osterand Schroer, 1979). The extent of water-quality impacton soil permeability was shown to depend on soil texture(Frenkel et al., 1978); clay mineralogy (McNeal andColeman, 1966); presence of soluble soil minerals(Shainberg et al., 1981), soil binding agents, Al and Feoxides, and organic matter (Goldberg et al., 1988);Na/K ratio of soil saturation extracts (Robbins, 1984);and constancy of irrigation water quality (Oster andSchroer, 1979). Sinclair et al. (1992) measured no effecton intake when they applied CaCl2-treated water (SARnot specified) to furrows in hard-setting sandy loamsoils. Their gravimetric sampling scheme, however, waslimited in extent and may have inadequately representedsoil water conditions. Evans et al. (1990) measuredseason-long intake rates using a recirculating furrowinfiltrometer. Furrow intake was higher for more salinewater treatments, even when irrigating with high SARwaters (EC = 9.2 dS m"1 and SAR MOO vs. EC =0.1 and SAR = 0.97).
Furrow irrigation water quality affected soil cohesivityby altering clay dispersion (Velasco-Molina et al., 1971;Frenkel et al., 1978; Malik et al., 1992; Shainberg etal., 1992) and aggregate stability characteristics (Smithet al., 1992). Irrigating with high-SAR water increaseddouble-layer thickness and zeta potential of soil colloids,leading to aggregate destabilization and enhanced chemi-cal dispersion (Malik et al., 1992), especially when soilaggregates were exposed to the mechanical disturbanceprovided by flow shear (Peele, 1936; Oster and Schroer,1979). The resulting soil structure was less cohesive and
Abbreviations: EC, electrical conductivity; SAR, sodium adsorption ratio;ESP, exchangeable sodium percentage.
238
LENTZ ET AL.: IRRIGATION WATER-QUALITY EFFECTS ON SOIL LOSS AND INFILTRATION 239
Table 1. Properties of Portneuf soil (plow layer).Particle-size distribution
Texture
SQt loam
Sand
100-170
Silt
650-700
Clay
180-200
Dominant clayminerals!
I»K = M> V
Cation-exchangecapacity
18-20
ECJ
dSm-'0.5-0.7
ExchangeableNa
percentage
1.6-1.8
pH
7.9-8.2
OMt
10-17
Aggregatestability §
%, w/w89
t Coarse clay fraction: I = illite, K = kaolinite, M = montmorillonite, V = vermiculite.I EC = electrical conductivity (saturated paste extract); OM = organic matter.§ From Lehrsch and Brown (1995).
more susceptible to detachment and transport forces ofthe furrow stream. Arulanandan et al. (1975) measuredthe fluid shear stress required to initiate erosion from apacked sample of a cohesive soil. At a given SAR, thesoil critical shear stress increased as the EC of the erodingfluid increased. This laboratory study was unable to fullysimulate field furrow conditions (e.g., it did not accountfor initial low soil water content of furrows or infiltra-tion). Hence, the observed response of soil shear tochanges in fluid EC may differ from that occurring inthe field. Soil dispersion also decreased with increasingelectrolyte concentration of the percolating solution(Quirk and Schofield, 1955), even when soil SAR washigh (Velasco-Molina et al., 1971; Arora and Coleman,1979; Shainbergetal., 1981).
Water chemistry may influence the sediment transportcapacity of the furrow stream indirectly via impacts onflow shear (i.e., infiltration-induced flow rate effects),and by modifying the character of entrained soil particlesand aggregates. Water quality affected flocculation,which determined the size and density of detached soilmaterial (Arora and Coleman, 1979; Goldberg and Glau-big, 1987). Compared with dispersed suspensions, theaggregate-size distribution of flocculated suspensions wasskewed toward larger sizes. However, Gregory (1989)reported that in flowing water, hydrodynamic shear in-creased floe breakage and limited maximum floe diame-ters to between 50 u.m and several millimeters, dependingon the strength of flow shear. The relatively greaternumber of large coalescent masses in flocculated suspen-sions requires greater energy for transport, settles faster,and is less likely to be entrained in the furrow stream.For example, increasing EC of irrigation water enhancedsoil flocculation (Arora and Coleman, 1979) and in-creased settling rates of sediment suspended in water(Robbins and Brockway, 1978).
We hypothesized that irrigation water-quality impactssediment loss from furrows via effects on soil erodibility,flocculation, and infiltration. Water with low EC and/or high SAR should promote dispersion and developmentof a slowly permeable surface seal. This would decreaseinfiltration and increase stream velocity. Greater streamvelocities may stimulate detachment and sediment trans-port processes and increase soil loss from furrows. Inaddition, low EC and/or high SAR should weaken soil
aggregates and decrease the soil's resistance to shear.Alternatively, high EC/low SAR water would improveaggregate strength, promote flocculation and develop amore permeable surface seal, stabilize infiltration, andinhibit soil removal and transport processes. Our objec-tives were to: (i) determine whether EC and SAR ofinflowing water affects sediment loss from irrigated fur-rows, and (ii) relate furrow sediment loss to net infiltra-tion and outflow.
METHODSThe study was conducted from July through mid-August
1991 near Kimberly, ID, on a 1.6-ha field of fallow Portneufsilt loam. Properties of the Portneuf soil are presented in Table1. Slope is 1%. The previous potato (Solatium tuberosum L.)crop was harvested, the field was fall disked, and disked androller-harrowed in the spring. Pre-(spring) and post-emergence(mid-July) herbicides were applied to control weeds. Irrigationwater, diverted from the Snake River, was conveyed to furrowheads via gated pipe fitted with adjustable spigots.
Furrows approximately 20 cm deep were formed with 75°V-shaped implements attached to the tractor's rear tool bar.Furrow spacing was 0.76 m, and furrow length was 114 m.Furrows were established in mid-June. Unused furrows werecultivated and reformed in late July to prevent soil consolidationand to remove any weed residue.
Four water-quality treatments consisted of combinations oftwo EC and two SAR levels. The chemistry of water-qualitytreatments selected satisfied the following criteria: (i) EC andSAR levels were moderate relative to the entire range ofirrigation water qualities available in the western USA, yethad demonstrated potential for altering soil properties (e.g.,infiltration rate) in surface irrigation applications (Oster andSchroer, 1979); (ii) treatment levels were obtainable by simpleamendment of local irrigation water, and (iii) treatment choiceswould permit preparation of low and high SAR waters thathad similar EC. The targeted EC levels were low EC = 0.5dS m-' and high EC = 2 dS nr1. The targeted SAR levelswere low SAR = 0.7 and high SAR = 12. The controltreatment (low EC/low SAR) was untreated Snake River water,which is characterized in Table 2. Typically, the compositionof untreated Snake River water is relatively stable during theperiod encompassed by the study, e.g., EC and SAR vary lessthan 8% (Carter et al., 1973). Water chemistry of furrowstreams was adjusted by metering solutions of NaOH, CaCl2,and NaCl into inflows. The NaOH solution was employedbecause it increased irrigation-water SAR levels with minimal
Table 2. Composition of untreated Snake River water (i.e., low electrical conductivity [EC] low [SAR] sodium adsorption ratio treatment).Ca2 Mg2 Na + HC03- so2,- ci- ECt SARt pH
2.30 1.44 1.13• mrnolc L"1
0.12 2.95 0.81 0.78dSm-
0.42(m moLL-')05
0.82 8.3
240 SOIL SCI. SOC. AM. J., VOL. 60, JANUARY-FEBRUARY 1996
Table 3. Treatment tail-water characteristics.
Irrigation watertreatment
low EC/low SARhigh EC/low SARlow EC/high SARhigh EC/high SAR
Samplenumber
49715473
ECt
mean- dS m0.52.10.71.7
SD-i _0.10.90.40.7
SARtmean
(m mole0.90.59.19.3
SD
L-')os
0.20.17.63.9
pHmean
8.67.59.27.4
SD
0.10.30.70.1
Table 4. Significance of F values from analysis of variance onmeasured responses.
t EC = electrical conductivity; SAR = sodium adsorption ratio.
influence on EC (via the reaction: Na+ + OH~ + Ca2+ +HCO3- ?* CaCCM + C02t + Na+ -I- H2O).
Either peristaltic pumps or constant-head discharge deviceswere utilized to introduce salt solutions into furrow inflowsbeneath gated-pipe spigots. Metering rates were checked peri-odically throughout each irrigation. Eighteen to 24 tailwatersamples were randomly collected from furrow treatments dur-ing each irrigation. Soluble Ca, Mg, and Na in the sampleswere determined by atomic-absorption spectrophotometry, andEC was measured with a conductivity meter. These measure-ments indicated that tail-water EC was very close to targetvalues, but SAR in runoff water of high SAR treatments wasoften closer to nine than our target value of 12 (Table 3). TheSAR of the low EC/high SAR treatment (NaOH added) wasmore variable than other treatments. The greater variation wasprobably related to the simultaneous reaction that occurredduring solution addition (see above), and may have resultedfrom differing reaction rates or endpoints present in eachfurrow system.
The study employed a split-plot experimental design. Eachreplicated block consisted of four plots representing the fourwater-quality types. Each plot was split into a wheel-trafficked(one tractor pass) and non-wheel-trafficked furrow subplot.The response from each subplot was taken as the mean of thethree furrows. Subplots received two irrigations. The secondirrigation occurred 7 to 9 d after the first. Furrows were notdisturbed between irrigations. For statistical purposes, thetwo irrigation events were referenced as irrigation episode.Interaction effects between irrigation episode (first vs. secondirrigations) and treatments were examined by including irriga-tion episode as a split of the subplot. The experiment wasrepeated in a new block six times (six replicates) from Julythrough mid-August. The entire data set included completemeasurements from 288 irrigated furrows. Values from threefurrows were averaged for each subplot response, so the statisti-cal data set includes 96 responses: (6 replicates) X (4 waterqualities) x (2 traffic types) x (2 irrigations) = 96.
Surface soils were sampled for soil water content beforeeach irrigation. Inflow rate was 19 L min"1 for all irrigationsexcept for the initial irrigation in the last three repetitions;inflow was increased to 23 L min~' during these irrigationsto promote more rapid furrow stream advance across the driersurface soils in these plots. Inflow and outflow rates weremeasured, and runoff samples collected for each furrow duringeach irrigation. Irrigation and runoff initiation times were notedin all irrigations. Furrow runoff rate was measured usingcalibrated, long-throated flumes (Bos et al., 1984) installed nearfurrow outfalls. Outflows were measured and runoff samplescollected at 30-min intervals during the first 1 to 3 h of anirrigation; once outflows stabilized, samples were taken hourly.
One-liter runoff samples were collected from the flumeoutlet. Sediment concentration in tail-water samples was mea-sured using the Imhoff cone technique (Sojka et al., 1992),in which 30-min settled-sediment volume is correlated withsediment weight. Three Imhoff-cone sediment samples werecollected from each treatment in each irrigation. These were
Source of Sediment Total Totalvariation loss inflow outflow
Model (overallF test)
Water quality(TRT)
Traffic (TR)Irrigation episode
(IE)TRT X TRTRT X IETR x IETRT x TR x IE
**
****
**NS*
NSNS
NS **
_ **_ **
_ **NSNS
_ **NS
MeanNet concentration Advance
intake of sediment time
**
****
**NSNSNSNS
**
****
**NSNSNSNS
*#
***#
**NSNSNSNS
, ** Significant F-value at 0.05 and 0.01 probability levels, respectively;NS = not significant at the 0.05 level.
filtered, and the papers dried and weighed, to provide sedimentmass values for calibration of sediment volume to mass correla-tion functions.
The computer program FUROFIGR (Lentz and Sojka, 1994)expedited analysis of furrow irrigation data. Analysis indicatedthat Imhoff calibration functions for treatments and irrigationepisodes were unique. Therefore, separate calibration functionswere computed for furrows from each different treatment/irrigation episode combination. This version of FUROFIGRassumed constant inflow, outflow, and sediment concentrationduring each measurement period. Total inflow, outflow, andsediment losses were summed across all periods. Net infiltrationwas the difference between total inflow and total outflow, andmean sediment concentration was the ratio of total sedimentloss to total outflow. Furrow advance time was the numberof minutes required for inflowing water to initially traversethe entire dry furrow length. Analysis of variance comparedtreatment effects. Mean separations in tables and error barsin treatment-comparison figures were obtained from confidenceintervals derived from pairwise Student's Mests (a = 0.10).A 0.10 significance level was used to offset the lack of powerassociated with the conservative f-test.
RESULTS AND DISCUSSIONAnalysis of variance (Table 4) showed that water
quality, irrigation episode, and traffic significantlyaffected total tail-water soil loss, total outflow, net infil-tration, mean tail-water sediment concentration, and ad-vance time. Significant interactions were a water qualityX irrigation episode effect on soil loss and a traffic Xirrigation episode effect on total outflow.
Water-Quality EffectsThe water-quality effect on total furrow soil loss was
examined separately for each irrigation episode becauseof the significant interaction of water quality X irrigationepisode on sediment loss. The soil loss pattern in responseto water-quality treatments was similar in both irrigations(Fig. 1A and 2A), but treatment effects were slightlylarger for the second irrigation (Table 5). In the secondirrigation, sediment loss for control (low EC/low SAR)furrows was 2.9 Mg ha"1, and all soil loss mean separa-tions were significant (Table 5). For both irrigation epi-sodes, soil losses from high EC/low SAR treatmentswere 67 to 72% of control losses. Low EC/high SAR
LENTZ ET AL.: IRRIGATION WATER-QUALITY EFFECTS ON SOIL LOSS AND INFILTRATION 241
0)
re
E3o
(AWO
OCO
0>
'•5a>co coo
cE
O
40
30
20
10
0
10
8
6
4
2
0
10864
Low EC/Low SAR
High EC/Low SAR
Low EC/High SAR
High EC/High SAR
C
25 50 75Runoff Period /0/-
100
Fig. 1. Influence of irrigation water quality on: (A) cumulative soilloss; (B) mean sediment concentration; and (C) outflow during thefirst irrigation. Plotted values are means of all furrows includedin each water-quality treatment. (Note.v axis scale differences withFig. 2.)
treatment soil losses were 1.43 to 2.03 times those ofcontrols, and high EC/high SAR soil losses were 1.04to 1.31 times control amounts.
The response of mean sediment concentration to water-quality treatments was similar to that of total soil loss(Fig. IB and 2B). Sediment concentration was greatestin low EC/high SAR, smallest in high EC/low SARirrigated furrows, and intermediate for the control (lowEC/low SAR) and high EC/high SAR furrows (Table5). Total outflow, net infiltration, and furrow advancewere without water-quality interactions, thus treatmentvalues from all irrigations were analyzed together (Table6). In general, total outflow was 25 % greater, net infiltra-tion was 14% less, and furrow advance was 19% fasterfor high SAR than for low SAR treatments.
For these irrigation waters, the impacts of increased
si:li sE -J3 .—O Q
3020
CO 100
Low EC/Low SAR
High EC/Low SAR
Low EC/High SAR
High EC/High SAR
• 60) C
CO Oo
-—, 15_l 12
^ 9
6
3
B
E
o
108642
0
C
25 50 75Runoff Period (%)
100
Fig. 2. Influence of irrigation water quality on: (A) cumulative soilloss; (B) mean sediment concentration; and (C) outflow during thesecond irrigation Plotted values are means of all furrows includedin each water-quality treatment. (Note y axis scale differences withFig. 1.)
EC on sediment concentration and soil loss were theopposite of those observed when SAR increased (Fig.1 and 2). Elevating water EC reduced the total sedimentloss and mean sediment concentration in runoff, but hadlittle effect on net infiltration or rate of furrow advance(Tables 5 and 6). Conversely, increasing the SAR ofthe irrigation water increased total sediment loss and themean sediment concentration in outflow, reduced netinfiltration, and increased the rate of furrow advance.Increasing both EC and SAR resulted in lower soil lossthan occurred when SAR alone increased (Tables 5 and6). These results strongly support our hypothesis.
Clay DispersionLow EC/high SAR irrigation water enhanced chemical
dispersion and destabilization of soil aggregates in furrow
Table 5. Effect of water quality on furrow soil loss in each irrigation.First irrigation Second irrigation
Low SARt High SAR Low SAR High SARMeasured response Low ECt High EC Low EC High EC Low EC High EC Low EC High ECTotal soil loss,Mean sediment
Mg ha-'cone., g L"1
2.15.4
b*b
1.4 c3.7 c
3.0 a6.5 a
2.24.8
bb
2.9 c7.0 b
2.1 d5.4 c
5.9 a10.9 a
3.8 b7.6 b
t EC = electrical conductivity; SAR = sodium adsorption ratio.f Means within a row and irrigation episode with dissimilar letters are significantly different (P < 0.10); each value is the mean of 36 furrows.
242 SOIL SCI. SOC. AM. J . , VOL. 60, JANUARY-FEBRUARY 1996 .
Table 6. Overall influence of water quality on furrow flow rateand infiltration.
Low SARt High SAR
Measured response Low ECJ High EC Low EC High EC
Total inflow, mmTotal outflow, mmNet infiltration, mmFurrow advance, min
107$40 b§67 ab59 ab
10736 be71 b63 a
10647 a59 a51 be
10748 a59 a48 c
t EC = electrical conductivity; SAR = sodium adsorption ratio.t Mean separations were not appropriate because of a nonsignificant
ANOVA; each value is the mean of 72 furrows.§ Means within a row with dissimilar letters are significantly different (P
< 0.10).
streams. This was evident from both the Imhoff conesamples and their calibration functions, which relatedsettled sediment volume to mass of sediment per liter.Turbidity of the Imhoff cone supernatant (after settlingfor 30 min) was greater in low EC/high SAR samplesthan in high EC/low SAR samples, signifying greaterclay dispersion and clay content in the low EC/highSAR sediment. Furthermore, the slope of the calibrationfunction for low EC/high SAR samples (0.73) was sig-nificantly less (P < 0.05) than that for samples from thelow EC/low SAR treatment (0.84), indicating that thebulk density of the low EC/high SAR settled sedimentwas less than that of the low EC/low SAR treatment.We attributed its low bulk density to the dispersed, lessconsolidated condition of the low EC/high SAR sediment.
That irrigation water having an SAR of « 9 and ECof —0.1 would promote clay dispersion may be surpris-ing for two reasons. First, clay dispersion is a functionof a soil's ESP; it's possible that the soil in the furrowdid not equilibrate with the source water during the irriga-tion, thus soil ESP would have been less than the water'sSAR value. Patience and numerous leaching volumesare required to equilibrate soil in a laboratory columnwith infiltrating water. However, erosion and infiltration(sealing) are soil surface and furrow-stream phenomena,therefore, it is the impact of water chemistry on surfacesoil that is most critical. The kinetic energy conditionsand soil/water ratios occurring in the surface 2 millime-ters and flow-suspended sediment of irrigated furrowsdiffer markedly from those present at the soil surface incolumn experiments. Mixing and redistribution of sur-face soil caused by flow shear and low soil/water ratiosof furrow-stream suspensions may have accelerated theequilibration process in furrows relative to laboratorycolumns. If furrow soils did not equilibrate with appliedwater, then water-quality impacts on furrow processes insubsequent irrigations could be greater than that observedhere.
A second concern is that the salt concentration of thelow EC/high SAR water (~1 mmolc L"1) was enoughto mask any dispersing effects of the water's higher SAR(Table 3). Shainbergetal. (1980) reported that infiltratingwater with a salt concentration of 2 mmolc L~' preventedclay dispersion in soil columns when water SAR < 10.However, the influence of EC and SAR on clay disper-sion, infiltration, or hydraulic conductivity varies withsoil type, and other researchers reported that clay disper-sion can occur with SAR = 10 source water at electrolyte
concentrations of < 10 mmolc L ' (Curtin et al., 1994;Frenkel et al., 1978; Goldberg and Forster, 1990).
While the pH of the low EC/high SAR tail water was1 to 1.8 units higher than for other treatments (Table3), the pH effect on clay dispersion-flocculation wasconsidered to be slight. Goldberg and Forster (1990)reported that the critical coagulation concentration valueof soil clay suspensions with SAR < 25 was unchangedwhen the suspension pH was increased from 7.5 to 9.5.
When furrows were irrigated with low EC/high SARwaters, rapidly wetted aggregates were destabilized andsediment was readily detached and transported, resultingin high in-stream dispersed sediment concentrations,compared with furrows irrigated with low SAR water.Greater stream sediment concentrations lead to enhancedsoil loss, particularly since outflows were also greaterin high SAR treatments than low SAR waters (Table 6).
An increase in EC promotes clay flocculation andsettling of suspended sediments in irrigation water (Rob-bins and Brockway, 1978). Settling of flocculated parti-cles produces an open, more porous depositional sealthan that produced by dispersed systems (Southard etal., 1988). The percentage of water-stable aggregatescan be raised by elevating water EC (Smith et al., 1992).Increased Ca levels stabilize the microaggregate fraction(Peele, 1936). Thus, a greater proportion of larger andmore stable aggregates are present mat are more resistantto transport forces than smaller aggregates and primaryparticles. Furthermore, increasing the EC of the erodingfluid increases soil resistance to shear forces (Arulanan-dan et al., 1975). Water-quality impacts on soil aggrega-tion could fully explain the sediment concentration andsoil loss effects we observed. However, water-qualitytreatments also affected infiltration and, hence, runoffrates (given equivalent inflow rates used across treat-ments). This suggests that soil erosion was also influencedby induced changes in furrow stream hydraulics, al-though not measured.
Surface Seal and Infiltration RelationshipsFor Portneuf soil, the surface seal that forms during
irrigation and erosion of furrows is instrumental in con-trolling furrow infiltration (Segeren and Trout, 1991).Evidence of seal formation and its effect on infiltrationand runoff can be seen in outflow-duration curves (Fig.1C and 2C). Those from the first irrigation (Fig. 1C)show a more gradual rate of increase in outflow, withsteady state nearly achieved by irrigation's end (95 % ofmaximum outflow occurred at 80% runoff). In contrast,those from the second irrigations (Fig. 2C) show a morerapid rise to steady state (95 % of maximum outflow at50% runoff). The first response resembles that producedwhen infiltration and seal development are initiated to-gether, while the second is indicative of an infiltrationevent initiated after a soil seal has already formed(Moore, 1981). Though surface antecedent soil watercontent for second-irrigation furrows was greater thanthat for the first (15 vs. 10% water content by weight),this probably had a smaller impact on the shape ofoutflow-duration curves than seal formation. Moore
LENTZ ET AL.: IRRIGATION WATER-QUALITY EFFECTS ON SOIL LOSS AND INFILTRATION 243
(1981) showed that whether surface seals were devel-oping or had already formed, a change in initial soilwater content tends to shift the soil' infiltration rate-duration curve rather than alter its shape. Thus, thechange in outflow-duration curve shape observed in thefirst and second irrigation most likely resulted fromdifferences in seal development.
An infiltration rate comparison between irrigated fur-rows having normal and subnormal seal development(unpublished furrow infiltrometer data) indicated thatpermeability of the developing seals tends to exert controlover furrow infiltration rate very early in Portneuf irriga-tions (in the first 5-10 min. of an infiltration event).These data, which were corroborated with in-furrowhydraulic resistance measurements (Segeren and Trout,1991), suggest that an imbibition-induced decline in soilwater potential gradient exerts a smaller influence onfurrow infiltration rate than does soil surface permeabilityin Portneuf soils.
Early in the first irrigation (0-20% runoff period),all treatments showed a parallel rapid increase in outflowinduced by the formation of a depositional layer andattendant abruptly decreasing infiltration rates (Fig. 1C).Later, the slope of the outflow curve decreased notice-ably, but this occurred sooner in the irrigation for lowSAR treatments (20-35% of runoff) than for high SARtreatments (45-55% of runoff). We hypothesize that theparallel rapid increase in outflow for all water-qualitytreatments was caused by the initial development ofa depositional layer with restricted permeability. Thesubsequent slope change noted above may reflect theaccumulation of differences in processes or factors con-trolling infiltration. Seal-forming processes under thehigh EC/low SAR treatment no longer produced theprevious steeply declining infiltration rates (i.e., increas-ing outflow rates) because more complete blockage ofthe depositional layer's open and porous structure wasnot as easily accomplished by the small quantities ofundispersed colloidal materials available, or by soil ag-gregates themselves, which were generally too large toenter soil pores. In contrast, under the low EC/highSAR treatments, the depositional seal developed, and itspermeablility declined, at a fairly uniform rate throughoutthe irrigation, steadily reducing infiltration. Well-dispersed particles common under these conditions pro-duced a depositional layer with smaller average poresize. Even then, small dispersed colloids could enter soilpores and penetrate several millimeters into the soilbefore lodging in and blocking soil pores. In this phase,continued colloid penetration maintained the steeply in-creasing outflow-rate curve of the low EC/high SARtreatment (Fig. 1C).
Furrow seal and infiltration processes apparently linkirrigation water quality to stream erosivity. High SARwaters produced a significant reduction in furrow netinfiltration and an increase in furrow outflow comparedwith low SAR counterparts (Table 6). We attributed thislower net infiltration to less permeable depositional sealsformed in response to high SAR waters (Shainberg andSinger, 1985). The high SAR treatments also had thegreatest soil loss, especially in the second irrigation
100
10 20 30 40 50 60 70 80Total Outflow (mm)
Fig. 3. Cumulative soil loss from nontrafflcked furrows irrigated withwaters of contrasting quality as a function of total outflow (bothirrigations).
(Table 5). The inverse relationship between infiltrationand furrow erosion was expected since decreased waterintake produces an increase in flow velocity, which en-hances flow shear, detachment, and sediment transportcapacity of the furrow stream (Kemper et al., 1985). Thisinfiltration-furrow erosion relationship was presentedgraphically in a preliminary analysis of our researchresults (Lentz et al., 1993). Increased soil loss was alsoattributed to the weakening of furrow aggregates andincreased soil dispersion under low EC and/or high SARconditions (Velasco-Molina et al., 1971; Arora and Cole-man, 1979), which promoted sediment detachment pro-cesses in the furrow stream.
Increased EC alone had an inconsistent influence onfurrow net infiltration, corroborating results of Sinclairet al. (1992). However, outflow-duration plots (Fig. 1Cand 2C) suggest a trend toward increased infiltrationwith increased EC, especially for low SAR treatments.
Figures 3 and 4 compare high EC vs. high SAReffects on outflow and soil loss relationships. Note thecurvilinear association between soil loss and outflow. Inaddition, soil loss in the high SAR system increasedmore sharply with outflow rate than did that in the highEC system. It appears that other factors besides outflowmay be controlling soil loss in the high EC/low SARtreatment, i.e., soil cohesivity and aggregate stability.These data suggest that the EC effect on furrow soil
CO
Q)75DC(A(0
OCO
200
Outflow Rate (g s )Fig. 4. Effect of irrigation water quality on soil loss rate vs. outflow
rate relationships during the initial furrow runoff phase. Furrowsare nontrafflcked (both irrigations).
244 SOIL SCI. SOC. AM. J., VOL. 60, JANUARY-FEBRUARY 1996
Table 7. Effect of irrigation episode on measured response whereinteractions were not present.
Table 8. Effect of traffic on measured response where interactionswere not present.
Measured response First Second
Soil loss, Mg ha"1
Mean sediment cone.,Total inflow, mmTotal outflow, mmNet infiltration, mmFurrow advance, min
g L~2.2t5.1 b
112t39t73 a80 a
3.77.7 a
1034756 b40 b
t Mean separations were not appropriate because of a significant interactionor nonsignificant ANOVA; each value is the mean of 144 furrows.| Means within a row with dissimilar letters are significantly different
credibility characteristics was more important than itsinfluence on infiltration and furrow stream erosivity.
Irrigation Episode EffectsA significant interaction between water quality and
irrigation episode was found. Compared with the initialirrigation, soil loss tended to be greater in second irriga-tions, but the difference was significant only for the lowEC/high SAR treatment. The increase in soil loss forsecond irrigations compared with initial irrigations was97% for low EC/high SAR and 73% for high EC/highSAR water, but only 40% for the low SAR waters. Theinfluence of high SAR water on total soil loss was muchmore potent during the second irrigation. Overall, sec-ond-irrigation mean sediment concentration was 51%greater, net infiltration was 23% lower, and furrowadvance 50% faster than in the first irrigation (Table 7).
Several factors contributed to these results. First, netinfiltration in previously irrigated furrows was reduced,which increased total outflow (Table 7) and enhanceddetachment and transport forces. Permeability in thesesoils was reduced owing to their consolidation and subsi-dence, and to formation of a surface crust. In addition,surface antecedent soil water content for second-irrigationfurrows was 50% greater than when first irrigated. Thisresulted in a downward shift in the soil infiltration rate-duration curve (Moore, 1981), hence soils in the secondirrigation absorbed water more slowly (shorter furrowadvance time). Second, the first irrigation may havealtered the chemistry of in-furrow surface soil, ampli-fying water-quality impacts on furrows in the secondirrigation.
Traffic EffectsCompared with non-wheel-trafficked furrows, wheel-
trafficked furrows in general had 64% greater total soilloss, 22% higher mean sediment concentration, 17%less net infiltration, and 34% faster furrow advance(Table 8). The compacted soils in trafficked furrowsabsorbed water less rapidly, produced higher runoff andstream velocities, and enhanced flow-induced erosion.Furthermore, the interaction of irrigation episode andtraffic increased outflow significantly for non-traffickedfurrows in the second irrigation. In the first irrigations,total outflow from trafficked furrows was already highbecause compaction had reduced net infiltration. Thus,
Measuredresponse
Soil loss, Mg ha"1
Mean sediment cone., g L~'Total inflow, mmTotal outflow, mmNet infiltration, mmFurrow advance, min
Non-wheel-trafficked
2.2 bt5.6 b
107t37*70 a70 a
Wheel-trafficked
3.6 a7.2 a
1074958 b44 b
t Means within a row with dissimilar letters are significantly different(P < 0.10); each value is the mean of 144 furrows.
t Mean separations were not appropriate because of a significant interactionor nonsignificant ANOVA.
the potential for increased outflow from second-irrigationtrafficked furrows was reduced.
CONCLUSIONSThe EC and SAR of inflowing water significantly
impact infiltration and erosion processes in irrigated fur-rows, supporting the hypothesis that water quality influ-ences furrow-irrigation-induced soil loss via effects onboth soil credibility and furrow-stream erosivity (viainfiltration effects). While the Portneuf soil is similar tomany irrigated soils in the Pacific Northwest, it clearlyis not representative of all soil types found in otherirrigated regions. Chemistry of soils themselves willinfluence their susceptibility to water-quality impacts.For example, soils with relatively high ESP would besensitive to low SAR water, but would be less sensitiveto high SAR water than soils with low ESP. Soils highin organic C content would be less sensitive to high ECirrigation waters than soils with low organic C (Hieland Sposito, 1993). However, the soil dispersion andaggregate stabilizing processes affected by irrigation wa-ter quality have been shown to be important for soilshaving a range of soil textures (Shainberg and Singer,1985; Evans et al., 1990), clay contents (Malik et al.,1992), and clay mineralogy (Quirk and Schofield, 1955;Velasco-Molina et al., 1971; Arulanadan et al., 1975).Likely, many irrigated soils will be susceptible in somedegree to water-quality impacts. Furthermore, waterquality is not a constant in all irrigated agriculture.Farmers sometimes blend surface and well sources tomeet volume demands, and surface water quality oftenchanges seasonally as snowmelt is displaced by returnflows and drainage in surface water sources. Therefore,the chemistry of irrigation water must be carefully consid-ered when assessing irrigation-induced erosion. Mathe-matical models that describe irrigation processes andpredict infiltration and erosion rates will need to accountfor water-quality effects on both soil erodibility andstream hydraulics parameters. Moreover, soil erosionparameters used to predict rainfall-induced erosion maybe subject to error if (i) rainfall simulators were usedto derive parameters, and (ii) chemistry of source waterused in rainfall simulators differed from that of rainwater.Under certain circumstances, these considerations mayprovide an unexpected source of error in output derivedfrom the Revised Universal Soil Loss Equation, WaterErosion Prediction Project, and other predictive technol-
LENTZ ET AL.: IRRIGATION WATER-QUALITY EFFECTS ON SOIL LOSS AND INFILTRATION 245
ogies. More study is needed to define water-quality pa-rameters to improve erosion prediction for a wide rangeof soils. Further study of water-quality erosion impactsunder rainfall regimes may also be warranted, since thechemistry of raindrops involved in initial interrill andrill erosion high on the landscape may differ significantlyfrom that of downstream waters, such as those flowingin ephemeral gullies, channels, and streams.
ACKNOWLEDGMENTSThe authors thank Dr. Bruce E. Mackey for assistance in
statistical analysis, and Dr. J.M. Bradford, Dr. C.E. Brock-way, Dr. G.R. Foster, Dr. J.E. Gilley, Dr. C.W. Robbins,Dr. M.J. Singer, Dr. P. Shouse, and Dr. K.K. Tanji for theirhelpful comments on a previous draft of the manuscript.
