This action might not be possible to undo. Are you sure you want to continue?

BooksAudiobooksComicsSheet Music### Categories

### Categories

### Categories

Editors' Picks Books

Hand-picked favorites from

our editors

our editors

Editors' Picks Audiobooks

Hand-picked favorites from

our editors

our editors

Editors' Picks Comics

Hand-picked favorites from

our editors

our editors

Editors' Picks Sheet Music

Hand-picked favorites from

our editors

our editors

Top Books

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Audiobooks

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Comics

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Sheet Music

What's trending, bestsellers,

award-winners & more

award-winners & more

Welcome to Scribd! Start your free trial and access books, documents and more.Find out more

Elsevier Science Publishers B.V., Amsterdam - - Printed in The Netherlands

357

[31 ON THE R E L A T I O N S H I P B E T W E E N THE TIME C O N D E N S A T I O N A P P R O X I M A T I O N A N D THE F L U X - C O N C E N T R A T I O N R E L A T I O N

MURUGESU S I V A P A L A N and P.C.D. MILLY

**Water Resources Program, Department of Civil Engineering and Operations Research, Princeton University, Princeton, NJ 08544 (U.S.A.)
**

(Received September, 1987; accepted after revision April 12, 1988)

ABSTRACT Sivapalan, M. and Milly, P.C.D., 1989. On ~he relationship betwee ,~the time condensation approxi. mation and the flux-concentration relation. J. Hydrol., 105: 357-367. The time condensation approximation (TCA) has been used widely in hydrological modeling since it was introduced in the 1940's. This paper proposes a theoretical foundation for the validity of the TCA applied to constant-rainfall infiltration by recourse to the flux-concentration relation (FCR) used in soil physics. When the FCR is independent of boundary conditions, the TCA is exact. The success of the TCA is the result of two important properties of the FCR. The first is the fact that for a variety of flow processes the FCR is nearly the same regardless of the actual boundary conditions. Secondly, the FCR-based solutmns of the Richards equation are relatively insensitive to the form of the FCR. For a delta-function soil, the TCA is exact. The largest errors associated with the TCA occur in the case of linear soil, where it leads to a 19% error in the calculation of time to ponding.

INTRODUCTION

The phenomenon of soil moisture movement has been studied to the extent that both Childs (1968) and Philip (1973) have said (White et al., 1979) that "further refinements of its conceptual bases are unlikely to be profitable." Even so, the best existing solutions of infiltration problems, for example, are those that have been obtained for only the simplest of boundary conditions, namely that of surface saturation. More recently, great progress has also been made on the problem of constant-flux infiltration (Broadbridge and White, 1988; White and Broadbridge, 1988). However, due to the feedback effects inherent in soil moisture movement, the boundary conditions are quite complex and time variable. Thus, the governing boundary conditions cannot be assigned a priori but have to be updated continually. Taking the example of infiltration, since initially the rate of rainfall is less than the infiltration capacity, '~he actual infiltration rate is equal to the rainfall rate and the moisture content at the surface will be determined as part of the solution of the problem. After ponding has taken place at the surface, the governing boundary condition

0022-1694/89/$03.50

© 1989 Elsevier Science Publishers B.V.

one needs only to solve the infiltration problem for a single boundary condition. what is needed is a simple and reasonably accurate method for predicting infiltration and evaporation amounts on a continuous basis during periods of in:ermittent storm and interstorm periods. the concept of time condensation or time compression was first introduced by Sherman (1943) in the context of partitioning rainfall into infiltration and overland flow during erratic rainfall events. it is equivalent to assuming that the soil moisture profile. once the soil surface is saturated. The underlying assumption of the TCA as originally applied was that the infiltration capacity of the catchment at any given time within a storm depends only on the volume of infiltration from the storm until that time. At present. But these are far too expensive and time consuming for practical applications in field-scale modelling of the rainfall-runoff process. He used a form of Philip's (1957) infiltration equation to derive an equation for the infiltration capacity as a function of cumulative infiltrat. In the latter case. regardless of its distribution in time.on for a ponded surface and used the latter as the base equation for the determination of rainfall infiltration.flux infiltration has recently followed the development by Parlange (1971) and Philip (1973) of the fl~lx-concentration relation (FCR).358 becomes the specification of saturation moisture content at the surface and the rate of infiltration is obtained as part of the solution. The physical basis for the validity of TCA was never analyzed though numerical tests have shown it to be reasonably accurate and useful when applied to the problem of one-dimensional infiltration into a homogeneous soil. numerical methods have advanced to the stage that permits the accurate solution of the nonlinear and hysteretic governing equations for the quite general initial and boundary conditions. For field applications of the TCA the base . The problem is even more complicated in practice since rainfall and evaporation events are quite nonuniform and random and since hysteresis effects also need to be taken into Consideration. is dependent only on cumulative infiltration and is independent of the actual rainfall history. If this is true it means that to predict infiltration capacity resulting from any time-varying boundary condition. The simplest such solution is for a continuously ponded surface. A good illustration of the application of the TCA m variable-rainfall infiltration is given by Milly (1986). Time condensation approximation Apparently. It was in this regard that the time con~lensation approximation (TCA) was introduced into hydrological practice i~ the early 1940's to be used in conjunction with empirical field-scale infiltration functions. Significant progress on assigned. who also applied the YCA to the analogous problem in evaporation. The purpose of the present exercise is to explore the theoretical underpinnings of the TCA by use of the FCR. The same assumption is implicit in the application of the TCA to a homogeneous one-dimensional soil profile. An analogous situation exists in the case of evaporation from the soil surface.

They found generally good agreement. The flux-concentration relation The work of Parlange (1971) led to Philip's (1973) introduction of the fluxconcentration relation (FCR). D(O) is the moisture diffusivity. The governing equation considered here is the one-dimensional form of the Richards equation for homogeneous soil: dO d [ D dO e-7 = & K] (la) where 0 is the volumetric moisture content. (lb). (4b) to eqn. to the initial condition [eqn. the errors observed in their comparisons were due to the redistribution taking place during breaks in the rainfall and due to the hysteresis accompanying it. the errors might be greatly eliminated. (la) can be simplified to: 0-7 = d-~ 0 [°00] d-~ z >I 0 0 = 0. and eqn. (lb) These governing equations will be subject. taken positive downward. More work on this aspect of the problem is needed to improve TCA-based models. (la). and z is the vertical coordinate. K(O) is the hydraulic conductivity. Reeves and Miller found that the infiltration capacities are slightly underestimated when the TCA is used. eqn. (3) or (4)]: t = 0 (2) (3) t > o z = o o = 00(0 t > 0 or: z = 0 00 D(O) o---z - K(O) = . in general.Vo(t) (4a) t )0 z = 0 00 D(O)-~z = -Vo(t) (4b) Equation (4a) applies to eqn. . Reeves and Miller (1975) and Ibrahim and Brutsaert (1968) have performed numerical tests of the TCA for erratic rainfall by comparing the cumulative infiltration and the moisture profile obtained using the TCA against those obtained by numerical solution of the hysteretic governing equations. For one-dimensional systems which are horizontal or in which the effect of moisture gradients dominates that of gravity. Their results imply that the error at the end of a break is related to the amount of redistribution and that if a correction is made in the continuous accounting of the cumulative infiltration. The FCR expresses the dependence of flux density on moisture content during various unsteady flow processes in unsaturated soils. Both D and K are usually strongly varying functions of 0.359 curve could be obtained by field experiments under saturated surface conditions. (2)] and one of the boundary conditions [eqn.

t) = or: F(®. and thus F{®. regardless of the choice of boundary conditions. The relationship given by curve C¢ is for linear soil (constant diffusivity) and was derived analytically by Philip (1973). The F(O.t) does not vary greatly over a wide range of flow processes and soil characteristics. 1973). (la) or (lb) can be found immediately. F(0. Curve A is for the so-called delta- . When Oo(t) and vo(t) are time independent. It is assumed that for all t > 0 the moisture profile varies monotonically from the value 00(0 at z = 0 to the value 0n as z --.t) = 1. (lb). B¢. both ® and F(O. Philip (1973) introduced: 0 -0n ® = Oo(t) . Firstly. (la) and the second expression applies to processes (absorption or desorption)described by eqn. that F(®.g. In general Oo(t) and vo(t) will denote surface moisture content and flux density. if F(®. when F(@.t) are nonincreasing functions of z. Figure 1 illustrates the characteristics of F(®. the set of conditions of eqn (3) may be designated as variable-concentration conditions and those of eqn (4) as variable-flux conditions.t) is not known exactly.t). then they will be called constant-concentration and constant-flux conditions respectively. (6a) v(e. that first solution estimates are relatively insensitive to errors in the guessed F(®.t). Similarly.t) applies to processes (infiltration or evaporation) governed by eqn.t) Vo(t~ (6b) The first expre~sion for F(®. oo. and C¢. and the flux-concentration relation.t) relations for absorptmn with the constant-concentration boundary condition for various soft types are given by the curves marked A. as Philip (1973) has shown.K.360 It is also required that Oo(t) and vo(t) are such that hysteresis effects are avoided. For absorption F(®. Using the terminology of Philip (1969.t) = Vo(t). (la). changes monotonically from Vo(t) to zero in the case of eqn. a first estimate of any solution may be obtained by using a guessed (F(®. (lb) and from Vo(t) to Kn in the case of eqn.t) is known exactly for any given process among those discussed above. the flux density or Darcy velocity v(O.t) is necessarily a nondecreasing function of EJ for any time t and satisfies: F(1. Secondly. (5) v(O.t) function. It turns out.t) is independent of time. With the assumptions already made. the relevant exact solution of eqn.t) for wetting processes and is based on the work of Philip (1973).t) = 0 (7) Philip (1973~ has stated two propositions on the utility of the FCR.O.t). and further. F(O.

White also confirmed Philip's proposition that the divergence between predictions using F¢ and the .~ I ! ////" . function soil. / / / J F(O..361 1.t). the F(®. all the curves in the interval A to Cc move towards A. Flux concentration relations..t) curves for "real" soils would be expected to be withia these bounds (e. working with Bungendore sand.0 0.t) = @. Curve Cf is for linear soils and c u r v e Bf is for the "real" soil. White et al. at t = 0. avd that the difference between Fc and F f is very small. the calculated moisture profiles were reasonably insensitive to the assumed form of .for which D(®) is a Dirac delta function.t). I 0.t) is time-dependent.t) in the case of absorption with the constant-flux boundary condition.2 I 0.~'~. the FCR at constant cowcentration. Curves Cc and A represent upper and lower bounds of F(@. and Cf.t) relations are equal to their counterparts for absorption expressed by curves C¢ and Cf for linear soil and Bc and Bf for "real" soils.4 0. it is smaller for "real" soils and is zero for the delta-function soil. F(®. W lite (1979) did similar work for absorption in Bungendore sand for constant-flux and constant concentration boundary conditions.It is clear that the difference between the constant-concentration and constant-flux infiltrationcurves is a m a x i m u m for linear soils at early times.t) for the constantconcentration boundary condition. F(®. with time the differences get even smaller. For the delta-function soil.2 - / O• 0.curve A holds for all t.t) 0. 1. Bf. for absorption with constant-concentration (subscript c) and constant-flux (subscript f) conditions. In the case of infiltration.4 I ® 0.t) relations for absorption with the constant-flux boundary condition for the three soil curves are given by curves marked A. A~ t -~ oo. He therefore proposed that for practical purposes F~ can be used as an approximation for Ff. as Phihp had predicted. A is for a delta-function soil for both conditions. His experimental moisture profiles confirmed Philip's (1973) prediction that for constant-flux absorption F(O. The F(®. C~ and Cf are for linear soils and Bc and Bf ar ~ for a "real" soil. lay below F~(O.~ Ct.6 I .8 0.0 Fig. The F(®. in this case F(®.t)..g. White also found. curve Be) and this is borne out by experimental and nmaerical observations. . Ff(O.~(®.t) for the delta-function soil is again given by curve A. that the FCR at constant flux. confirmed some of Philip's (1973) conclusions. Note that Bf and Cf are closer to A than are Be and C¢ respectively.8 1. (1979).

Outline of the analysis A limited special case of the application of the TCA is the problem of infiltration due to a constant applied flux v0fat the surface where it is assumed that v0f is greater than/(8. ANALYSIS OF INFILTRATION USING THE FCR Infiltration with the constant-concentration boundary condition Here the relevant governing equation is eqn. a constant. (2) and the boundary condition given by eqn.362 correct F~ can be expected to be maximum for linear soil and will decrease for materials whose diffusivity is more strongly dependent on ®. In the solutions outlined below the phenomenon of tension. After the time of ponding. the difference between F~and Fc is small and that for more "real" soil and with the passage of time. (1) with the initial condition given by eqn. the boundary condition becomes a constant-concentration condition and the infiltration rate falls below the applied flux.K® (8) . tp. In summary. In the next section.saturation has been ignored. this difference becomes even smaller. (3) where 00(0 = 08. the moisture profile at the time of ponding due to the constant. Then: Fo(e. An important part of any solution is the determination of tp and the moisture profile at this time. 08 is the saturation moisture content of the soil.flux boundary condition is shown to be the same (or nearly the same) as that which would have been obtained for the same cumulative infiltration had infiltration been taking place due to the saturation boundary condition. the validity of the application of the TCA to this special case is explored by recourse to the FCR-based solutions of the problem. The solution for t greater than tp c a n then be o utained using either of the two cases since the boundary condition as well as the initial condition would be the same (or nearly the same) in both cases. Let v0¢(t)denote the surface flux in this case. the important property of F(O. This will establish the validity of the ~pplication of the TCA to the infiltration problem with the constant-flux boundary condition.0 = v(0. From that time onward ponding takes place. the saturated hydraulic conductivity of the soil. decreasing with time. In this paper. this p'cofile will then be used as the initial condition with the constant-concentration boundary condition to obtain the solution after ponding.$) is that during infiltration even in the worst case of the linear soil.0v0o(t). In such a case during a certain initial period all of the applied flux infiltrates and the surface moisture content gradually increase until it reaches the saturation value.

[K(0) .t) .)D(O)dO f Fc(®. Back-substituting in eqn.K . ] (12) Let Qc be the net storage increase at any time due to infiltration with the constant-concentration boundary condition. Zo(0. (15) is a unique relationship between Q¢ and vo~(t) . which shows that z~(O.:(O. (12) into eqn (14).t) = . Substitution of eqns. (11): 0.(®.Oa) = (voc(t). 0 F~(o.O~) represents the variation of 0 with depth z for the constantconcentration boundary condition.t).K.)so On i/ (14) Substituting eqn.)zc(O.t) is given by the Darcy-Buckingham law as: v(O. in the two integrands above in terms of Q~ from eqn.Oa) (11) where zc(O.O~)dO or: 08 (13) Qc[voc(t) .] (15) Given D(O).[K(0).]/[V0c (t) . and Fc (®.K..] = | Zc(O. (9) The flux density v(O.) = 0. - 0. K(O).().K.08) = F. f Qc z¢(O.D(O) ~z + K(O) Let O0 (i0) gc(O. Then: 0.K.O- [K(0)- D(O)dO K. (10) and (11) in eqn. (15).t) - D(O)dO [K(0) .K.)DO)dO Yc(®.K .g.) is a unique function of Q¢.O..O.K . The corresponding relationship for the case of .t) .O. Ab Qc = [ z. eqn. ] On Equation (16) can be further rearranged by expressing voc(t) .K. integrating by parts and rearranging: O~ 1 f Q = [v0c(t).363 ® = 0--0n 0.]/[Voc(t).O. (8) and integration yield: 0.Kn] ( 0 . ] (16) ( 0 .]l[voc(t).]l[voc(t) .K.

:.t) o Os (17) ~ (0 .K.o.K.] Given f on ( 0 .364 absorption (or while gravity is negligible) i~: z~(O.K .g.t).l/[vof- 0. Oo(t) .)D(O)dO o Fd®.]/[v0f K. the relationship between Qr~ and (vof .] ( 0 . and Ff(O.t)- [K(0).O~) = o~ Qc f D(O)dO Fc(®. where Qfp denotes the ponding time value of Qf. K(0).0.[ g ( 0 ) .[ K ( 0 ) .t) ® = Let: = v(O.)D(O)dO Ff(®.] At the time of ponding 0s I zf(O.] (22) K. l / [ v o f . Zf [O.t) On Infiltration with the constant-flux boundmy condition The boundary condition here is eqn.)D(O)dO F~(O.)D(O)dO 0o(0 = 08 and eqn.) is given by: 0a 1 Q~P = [Vor.I.. (4) where the applied flux is assumed constant in time and equal to v0f.Oo(t)] = Qf .O.K.O. (21) becomes Ff(O. t ) . By analogy with the derivation of equation (16): 0ctt) I D(O)dO F d O . ] (23) D(O).K.O.]/[v0r.K .t).] (21) zf[O.Oo(t)] (20) where zf[O..O. Let: FdO.K.K.~ f (0 .08) : QfP .Knl/lvor.[ K ( 0 ) . Vof.K. (18) !9) O . 00(t)] expresses the variation of 0 with depth z for the constant-flux boundary condition. By analogy with eqn.t) . F f ( O . (15).Oo(t)] = (v0f - Kn)Zf[O. t ) - D(O)d(O [ K ( 0 ) .O. Qp and Vofare uniquely related to each other by .t).

(1979).e. i.t) = .785 Qf~ It can be shown analytically that the numerical constants in eqns. and therefore from eqns.t) is equal to Fc(®. corresponding to Qc = Qrp. Evaluation of the TCA Earlier in this sectio: ~it was shown that during ponding infiltration Qc is a unique ~anction of vuc(0 .365 eqn. zf(0. it was shown that at the time of ponding during constant-flux infiltration Qf. is a unique function of v0r . Similarly. (22?.t) O. (15) ~nd (23) Voc(t) is equal to v0f if Ff(®. If Qc is equal to Qf.t) (25) For a delta-function soil.t) o 0s zf(O.(®.t) and F c(O.)2 Q~ Vof D(O~ 0. (26a) and (26b) are in fact 2/~ and ~/4 respectively. If Ff(O. In addition. then from eqn.t) and Ff(®...K'. this relation is exact. Once again the corresponding expression for zf(O. (22). and therefore the TCA is exact. it follows that Voc(t) is equal v0f.O~) = o~ (24) f (O . Thus a sufficient condition for the validity of TCA in this problem is that: Ff(O.) 2 (26a) (27b) = 0...636 D(O~ . (23).~(O.t) expressions given by White et al. eqn (16) versus eqn.t).t) are identical. The solutions to the absorption problem in a linear soil having diffusivity equal to D were obtained numerically ltsing approximate F(O. (16) and (22) the moisture profiles are identical. Compare the moisture profile at the time of ponding of the constant-flux infiltration case to that of the ponded infiltration case.Kn.0~) for absorption is given by: Qfp ~ D(O)dO Ff(O.On)D(O)dO F. The true moisture profile at the time of ponding due to constant-fluz absorption in linear soil and tl~ avpr~ximate moisture profile based on the TCA are given by the common dimensionless equation: . Using these the TCA underestimates tp by about 19%.O~) is uniquely related to Qfp by eqn. (22). and at this time the moisture profile is given by eqn. Equation (23) is satisfied at the time of ponding tp and ca~ be solved to obtain tp.0.t) will be maximum for absorption (or infiltration at early time) in linear soils. The difference between Fc(®.. The relationship between Vow(t)and Q~ and that between v0f and Qfp are given by: Vow(t) = 0.

based simulation mvde| of moisture and energy fluxes at a bare soil surface. the errors in tp and the moisture prc.. 9th Int. P.E Z(@}vof~'5 O 2. and (b) approximate profile cbtained using th~ TCA.)/(08 . 2. Ibrahim.file obtained here for linear soils represent upper bounds to the corresponding errors in natural soils.0. (27) whereas the use of Fc(@. W.' t I ' Fig.Trans. Childs.0 ~TRUE PROFILE PROFI.-Y. Cir.0 2. Eng. J. Re~.. es-en oo 0. Comparison of dimensionless moisture profiles as a function of depth z during constant-flux absorption: (a) true profile. Am... 2.. As a result. 1968. 1971. White (1979) and White et al. and White.D.5 3. Congr. The computed dimensionless moisture profiles are presvnted in Fig.0 / (bosedontheTCA) r 3. K£PERENCES Broadbridge. the differences between the solutions will also be expected to be less for "real" soils than for linear soils.t) is maximum for linear soils. Theory of water movement in soils:1. Res. Soc. One dimensional absorption.t) and F¢ (®.t) yields the approximate profile. H..1988. Soil Sci. (1979). z(O)vof = i dO D F(O. 22 (12): 1680-1692. J. Div HYI: 113-137. Sydney. 1968. Since the difference between Ff(@. Water Resour. The true profile will be obtained if Ff(®. both Ff(@. For more "real" soils and for intermediate times. Milly. This is borne out by the numerical and experimental results of Philip (1973).C.366 @.. Analytic solution. P. Hydraul.C. Water Resour.0~).t) is used in eqn. Constant rate rainfallinfiltration:A versatilenonlinear mode].5 . e-S. P'xlange. I11: 134-137.A and Brutsaert. . Soil Sci.t) approach ® and the difference between them will be smaller. E. Intermittent infiltrationin~o soils with hysteresis. (1):Angus and Roberts®n.. The achievement~ of drainage theory in relationship to practical needs: taking stock. I.t) l (27) where ® = (0 . l.2 r 0 0. An event.... 24 (1):145--154. 1986.t) and Fc(@.5 1.

J.. White. Smiles. 116: 328-335.. On solving the unsaturated flnw equation: 1. 43: 659-664. Water Resour.. Measured and approximate flux concentration reiations for absorption of water by soil. Soil Sci. Sherman. Geophys. Soil Sci. Am. 1973. P.E.367 Philip. L. t943. 1975.. 24: 465-4~7. 1979. Adv. Philip.K.. 2.. Philip.M. J. Theory of infiltration..R. 1979. Soc. White.E. and Eroadbridge.. J. Sorptivity and algebraic infiltration equations. 5: 215-3~J5. I. White. Comparison of F-curves derived by the metkods of Sharp and Holtan and of Sherman and Mayer.. K. 1988...R. 84: 257-264. J. Trans. 1969... 4. . 43: 1074-1080. Res. Soil Sci. Theory of infiltration. J.R. I.. Res. 11 (1): 102-110. Applications and solutions. Soil Sci. Hydrosci. Union.. Constvnt rate rainfall infiltration: A versatile non-linear model. M. Absorption of water by soil: The constant-flux boundary condition... ?he flux-concentration relation. Am. Estimating infiltration for e ~ i c rainfall. I. D. 1957. Am. Water Resour. 24 (1): 155-162. E. and Mil~er. and Perroux. Reeves.

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue listening from where you left off, or restart the preview.

scribd