SW—Soil and Water: Two-component Analysis of Flow through Macroporous Soil

Download SW—Soil and Water: Two-component Analysis of Flow through Macroporous Soil

Post on 15-Jun-2016

212 views

Category:

Documents

0 download

TRANSCRIPT

  • mrt

    1. Introduction a diameter of 600 lm for water at surface tension ofIn "eld soil heterogeneity caused in part by the pres-ence of fractures, "ssures, channels, root and worm holes,peds and aggregates would result in multi-modal distri-bution of pore-size (Uttermann et al., 1990; Durner,1994). Since pore size directly a!ects soil hydraulic con-ductivity, a multi-modal pore-size distribution wouldresult in a multi-modal distribution of subsurface #owand solute transport (White, 1985; Jarvis et al., 1991).When two pore-size classes are considered in the analysiswe have a bimodal pore-size distribution, and the corre-sponding subsurface #ow distribution is bimodal. Apartfrom having a bimodal pore-size distribution, the occur-rence of a bimodal subsurface #ow distribution alsodepends on such conditions as antecedent soil watercontent and rainfall intensity (Jarvis et al., 1991).

    Various categorizations have been used in the litera-ture for de"ning macropores. For instance, macropores

    0)07 Nm~1 with a contact angle of 03. The matric poten-tial of !5 cm was selected as one which is greater thanthe maximum matric potential of !10 cm which wasapplied in laboratory experiments for determination ofsoil water characteristics of the study "eld (Diiwu, 1997;Diiwu et al., 1998a).

    One approach to modelling #ow in macroporousmedia considers the structural units (peds, aggregates,clods) as sources or sinks for the more mobile water inthe macropores. Another approach considers the bi-modal medium as two superimposed media, each with itsown hydraulic properties. There is the macroporedomain in which subsurface #ow and solute transmissionis very fast compared to slow subsurface #ow and solutetransmission in the micropore domain (Beven &Germann, 1981; Jarvis et al., 1991; Chen & Wagenet,1992). Exchange of mass between the macropore andmicropore domains depends on the degree of saturationJ. agric. Engng Res., (2001) 78 (1), 109}116doi:10.1006/jaer.2000.0617, available online at http://wwwSW*Soil and Water

    Two-component Analysis of Fl

    J. Y. Diiwu1; R. P. Rudra1; W

    1School of Engineering, University of Guelph, Guelph, ON, Canad2Land Resource Research Centre, Agriculture and Agri-Food Can

    (Received 28 April 1999; accepted in revised form

    Subsurface hydrographs, obtained during rainfall simmacropore and micropore components by applicationThe corresponding solute concentrations in the two doTime-domain re#ectometry was then used to estimate sian upper depth in the A horizon of the soil pro"le. The54% of total subsurface #ow and from 1 to 61% of tohave been de"ned as those pores having diameters from30 to 300 lm. They have also been de"ned based on thematric potential at which they drain such as !6 cm ofwater; by volume fraction such as 0)001}0)05; or byin"ltration rate such as 1}10 mmh~1; or as those poreswhich empty in 24 h (Beven & Germann, 1982; White,1985; Chen & Wagenet, 1992). In this paper, macroporesare de"ned as those pores which empty at matric poten-tials greater than !5 cm of water. This corresponds to

    0021-8634/01/010109#08 $35.00/0 10.idealibrary.com on

    ow through Macroporous Soil

    . T. Dickinson1; G. J. Wall2

    a N1G 2W1; e-mail of corresponding author: jdiiwu@uoguelph.caada, Guelph, ON, Canada N1H 6N1; e-mail: gwall@uoguelph.ca

    22 July 2000; published online 25 October 2000)

    ulation on 1m by 1m plots, were separated intoof a dual-porosity concept and mass balance analysis.mains were also determined by mass balance analysis.

    ilar out#ow hydrographs and breakthrough curves atesults show that the macropores contributed from 6 toal solute mass transported through the soil pro"le.

    ( 2001 Silsoe Research Instituteof the soil (Jarvis et al., 1991). The two-domain concepthas been presented in various forms such as the mo-bile}immobile form by Beven and Germann (1981), aswell as the dual-porosity model by Gerke and vanGenuchten (1993) and Durner (1994). The di$culty withthese models is the problem of accounting for the ex-change of mass between the two domains. Besides, forsoils with unstable aggregates it would be extremelydi$cult to accurately determine the hydraulic properties

    9 ( 2001 Silsoe Research Institute

  • Uof the two domains separately due to variability of soilproperties in the "eld.

    In modelling subsurface #ow and contaminant trans-port through macroporous soil, the uncertainties in thetwo-domain approach may be quanti"ed by means ofprobability measures of the hydraulic variables of thetwo domains. Combining this with dual porosity, thehydraulic processes in each of the macropore andmicropore domains may be characterized by a distinctprobability distribution (Diiwu, 1997). Derivation of thedistinct probability distributions for the macropore andmicropore domains requires that #ows and solute con-centrations be separated into macropore and microporecomponents if such components were not separatelymeasured during data collection. Also, for decision-making on water management it may be useful to be ableto assess the separate contributions of the macropore andmicropore domains to subsurface #ows and solute mass-es transported through the soil system. This may benecessary since macropores tend to transmit water andcontaminants rapidly beyond the root zone and intostreams and the groundwater system. This paper there-fore discusses the application of the dual-porosityconcept to separate subsurface hydrographs and break-through curves into macropore and microporecomponents. The analysis of spatial variability of thephysical and hydraulic properties of soil in the study "eldhad pointed to the possibility of the occurrence of macro-pore #ow in the "eld (Diiwu, 1997; Diiwu et al., 1998a,1998b). Hence, the need for subsurface #ow separation inthe study. Some existing soil water characteristic modelshave been found to perform poorly when applied on soilwith macropores (Diiwu et al., 1998c). Separation ofsubsurface #ow would enable such models to be success-fully applied to the micropore domain, along with analternative model for the macropore domain.

    2. Materials and methods

    2.1. Rainfall simulation experiments

    Subsurface hydrographs and breakthrough curveswere measured during rainfall simulation on 1 m by 1 mplots under no-tillage treatment. The plots wereconstructed by installing large steel plates approximatelyof the same dimensions as the plots, at a depth of about55 cm in the soil pro"le to serve as catchment pansto collect subsurface #ow during rainfall event. Thesteel plates were installed by gently driving them hori-zontally into place in the B horizon using hydraulic jacks.

    J. Y. DIIW110The investigation therefore involved the A and B hor-izons of the soil pro"le. The thickness of the A horizonat the site varies between 25 and 30 cm and the portionof the B horizon above the pans varies between 25 and30 cm.

    Plastic bottles were installed in a pit for collecting thesubsurface #ow for subsequent sampling. At the soilsurface, aprons were placed around the plots to directsurface #ow into a V-shaped trough from which sampleswere collected at 1 min intervals to determine volume ofruno! generated and the concentration of tracer presentin the runo!. Pairs of time-domain re#ectometry (TDR)probes, each probe measuring 20 cm in length and 0)2 cmin diameter, were installed horizontally at depths of 2)5,25, and 50 cm in each plot to measure in situ soil watercontent during rainfall simulation. The dielectric con-stant manually measured by TDR (Tectronix instrumentmodel 1502C) was used to infer the water content of thesoil via the emperical relationship by Topp et al. (1980).

    Rainfall intensity of 15)6 cmh~1 was simulated byusing the Guelph Rainfall Simulator II with a 12)7 mmfull jet nozzle that was maintained at a height of 1)5 mabove the soil surface and was operated at a pressurebetween 48 and 55 kPa (Tossel et al., 1987). Water forrainfall simulation was supplied by a pump at a rate of2)4 m3h~1. To avoid moving the simulator from one plotto another, three rainfall simulators were used, one oneach of plots 1}3. Characteristics of the three simulatorswere the same, and they were used to produce similarrainfall events. For each experiment bromide tracer solu-tion was applied on the soil surface of each plot usinga hand-held spray. The rainfall simulator was thenturned on and left to run for about 15 min, during whichtime ponding was established and maintained. Over the15 min period the simulator on plot 1 produced a totalvolume of 31 290 cm3 of water, on plot 2 the simulatorproduced 31 800 cm3 of water, and on plot 3 this was32 900 cm3. The slight di!erences in total volume ofwater produced are attributable to unavoidable sourcesof error such as the precise timing of start and end ofsimulation, the variation in nozzle pressure during simu-lation, and the e!ect of wind on the simulated raindrops(Tossel et al., 1987). Surface and subsurface #ows weresampled for volume and bromide concentration at aninterval of 1 min, over a 45 min period from the begin-ning of rainfall simulation.

    2.2. Separation of hydrographs into macroporeand micropore components

    The application of the dual-porosity concept inhydrologic modelling requires that the macropore andmicropore #ow components be separated and the

    E A .corresponding mass of contaminant in each componentdetermined. Everts and Kanwar (1990) representedthe transport of solute to subsurface drains by a mass

  • balance equation using a hydrograph separation tech-nique proposed by Pinder and Jones (1969). The tech-nique was adapted in this study for separating subsurface#ow into macropore and micropore components. Theequations for subsurface #ow and concentration of solutecan be written as

    Qt"Q

    mi#Q

    ma(1)

    Ct"QmiCmi#QmaCma

    Qt

    (2)

    where Qtdenotes the #ow rate measured at the pan in

    cm3 min~1, Qmi

    denotes the portion of Qtassociated with

    the micropore domain in cm3 min~1, Qma

    denotes theportion of Q

    tassociated with the macropore domain in

    cm3 min~1, Ctdenotes the total concentration of solute

    measured in Qtin lg cm~3, C

    midenotes concentration of

    solute associated with the micropore domain in lg cm~3,and C

    madenotes concentration of solute associated with

    the macropore domain in lg cm~3. The objective of theseparation performed in the study was to obtain valuesfor Q , Q , C and C from the hydrographs and

    TWO-COMPONEFig. 1. Subsurface hydrograph measured at pan on plot 1 for 26

    mi ma mi mabreakthrough curves observed at the pan, such that theconditions of mass balance expressed in Eqns (1) and (2)above were satis"ed.

    For the #ow separation a typical hydrograph obtainedfrom the rainfall simulation experiments and shownin Fig. 1, was partitioned into three stages. The partition-ing was based on the assumption that the hydraulicMay, 1993 event and corresponding estimated macropore andmicropore components: , total yow; , macropore yow;

    , micropore yowconductivity of the macropore domain was several ordersof magnitude greater than the hydraulic conductivity ofthe micropore domain. This condition is re#ected bydouble peaks in the measured hydrograph (Gerke & vanGenuchten, 1993). It was also assumed that there wasnegligible mixing between the two domains, and that themacropores were empty of antecedent soil water prior toeach rainfall simulation event (Everts & Kanwar, 1990).The entire duration of simulation was considered tocomprise three stages. In the "rst stage, the processes ofwater redistribution and drainage were considered to beinitially dominated by macropores. This was followed bya mixed stage when both macropores and microporeswere conducting water, before macropores emptied. Afterthe macropores were empty the micropores were respon-sible for drainage and redistribution of water. The ob-served hydrograph was then partitioned into the threestages as follows.

    Stage 1 from the beginning of rainfall simulation up tothe ,rst peak of the hydrograph

    In this case, it was assumed that the observed #ow wasonly from the macropore domain. The "rst peak wasattained a few minutes before rainfall simulation ceased.Also it was assumed that micropore #ow was juststarting.

    Stage 2 between the two hydrograph peaksIn this case, both macropore and micropore #ows were

    likely to be taking place, and hence any observed #owsinclude contributions from the two domains. While thecontribution from the macropores decreased as theyemptied, that from the micropores increased. By the timethe second peak was attained, all macropores have emp-tied, and so the micropore domain would be the only onecontributing to the observed subsurface #ow.

    Stage 3 beyond the second peakDuring this stage, all the observed #ow was assumed to

    be through the micropore domain and all macroporeshave emptied.

    With this partitioning of the hydrograph the separ-ation of subsurface #ow was required for stage 2 only.Moreover, there was a point in time during this stage atwhich the contributions to #ow from the two domainswere equal. The #ow separation was thus achieved byconstrained spline interpolations through the two peaksand the point of in#exion between them. The recedinglimb of the macropore component of the hydrograph wasspline "tted through three points*the "rst peak at whichQ

    ma"Q

    tand Q

    mi"0, the point on the time axis corre-

    sponding to the second peak, where Qma"0 and

    Qmi"Q

    tand the third point midway between the two

    111NT ANALYSISpeaks at which Qma"Q

    mi"0)5 Q

    t. The intermediate

    points of this spline curve were obtained by adjusting theproportions of Q

    mabetween 1 and 0)5 as well as between

  • used for the computations

    QtA

    *t"

Recommended

View more >