Tải bản đầy đủ - 0 (trang)
II. Soil Structure: Components of the Soil–Pore System

II. Soil Structure: Components of the Soil–Pore System

Tải bản đầy đủ - 0trang



can understand this if we see how the pore system relates first to static, then

dynamic, water behavior.



Although it is most logical to describe the pore space as the void ratio e

(volume of voids to volume of solids, V&), the porosity n (the volume of

voids to bulk volume of soil, v,lV,)continues to be the most commonly used

parameter in the agricultural literature. Since porosity is equal to 1.0 - pb/

pp(pb is the bulk density and pp the particle density) and since e =

pp/pb - 1.0, the conversion can always be made by the reader. In the case of

expanding clay soils, where pore volume and bulk volume change substantially with water content, this is advisable. Alternatively, for swelling soils, the

soil weight, over a given depth and at the most swollen state, is used as the

basis for comparative values in field situations. Reduction in e occurs with

soil shrinkage as water is removed. Increased overburden pressure (loading

such as occurs when the upper part of a dry profile re-wets) and compaction

by traffic and cultivation implements are also responsible for reductions in

pore volume. Increases in surface soil e values occur in freshly tilled seedbeds,

during swelling of clay and peat soils (with vertical displacements of 0.1 to

0.2 m in the top meter), and through frost heave (with displacements of 0.1 to

0.5 m per meter of vertical movement). Freshly tilled topsoils may have e

values of N 1.25, and soils high in organic matter with peat or litter layers

may reach e values of 1.4. Subsoils generally have e values of 0.45 to 0.8, while

cemented or indurated layers drop to e values of less than 0.25.




Pores in soils range from 0.003-pm plate separations in clay particles to

pipes, cracks, or channels tens of centimeters in diameter. Pore-size distribution is generally computed from the relationship between soil water content

(0) and matric potential ($) by means of the equation:

p g h = 2 y cos a/r


where p g h is the suction, equivalent to $ ( p is the density of water, g is gravity,

and h is the hydrostatic suction); y and cos c1 are the surface tension and

contact angle of water, respectively; and r is the equivalent cylindrical

radius. Thus for large pores or pipes in which the fluid-air interface is not

controlled by capillary (hydrophilic) forces, the moisture characteristic $(0)

(Childs, 1940) cannot be used to infer pore size. The limiting diameter of a



capillary pore in a normal temperature and pressure system which will

support a meniscus is 3 x lo3 pm (3mm), equivalent to a potential of

0.1 kPa.

Pore-size distributions determined from the draining function +(O) may

contain several sources of error. In laboratory measurements sample size is

often less than the minimum structural unit, particularly in the z dimension,

where the depth is generally less than 0.03 m to reduce equilibration time.

This biases the pore-size distribution (PSD) toward smaller pore classes. In

fine-textured soils shrinkage of the sample upon desorption alters both the

PSD and the pore shape, so that PSDs thus obtained are not representative

of the proportion of the fine pores found in the fully hydrated field state.

Lawrence et al. (1979) found a 10% collapse of “domain”-sized pores

( < 0.01 pm) (see Section III,B, 1) after water desorption, compared with

values obtained by critical-point drying (Greene-Kelly, 1973) and mercury

porosimetry in small clay aggregates. Although this alteration is significant in

studies of flow in saturated subsoils, within the context of soil-water uptake

by plant roots, we consider conventional water-desorption techniques to be

more appropriate. Finally, it must be remembered that Eq. (1) assumes a

regular, cylindrical pore shape, whereas the majority of soil pores have an

Table I

Description and Dimensions of Soil Pores


Johnson et al. (1960)

Brewer (1964)

Greenland (1977)



Capillary potential


< 75

75 to 1000

1000 to 2000

2000 to 5000

> 5000



5 to 30

30 to 75




0.5 to 50

50 to 500

Luxmoore (1981)

> 500

< 10

10 to 1000

> 1000

-4 to -0.3

-0.3 10 -0.15

-0.15 to -0.06

> -0.06

< -3000

<- 60

-60 to -10

-10 to -4

> -4

< - 6 x lo7

< - 6 x lo3

-6 x lo3 to -60

-60 to -0.6

> -0.6

< -30

-30 to -0.3

> -0.3



< -4

ECD, Equivalent cylindrical diameter.

Pore name


Very fine









Bonding pore

Residual pore

Storage pore

Transmission pore


Pressure gradient pore

Gravitational pore

Channel-flow pore



Table I1

Pore Dimensions of Biological Origin or Significance

Average pore

diameter (pm)




2000-1 1,000


300- 10,000

500- 10,000


50- 100








Biological significance

Ant nests and channels


Tap roots of dicotyledons

Nodal roots of cereals

Seminal roots of cereals

Lateral roots of cereals

1st- and 2nd-order laterals

Root hairs

Root plus root hair cylinder in clover

“Field capacity” (- 10 kPa)

Fungal hyphae


Permanent wilting point ( - 1500 kPa)


Green and Askew (1965)

Barnes and Ellis (1979)

Barley (1959)

Ehlers (1975)

Bouma et a/. (1982)

Nye and Tinker (1977)

and Russell (1977)

Caradus (1979)

Griffin (1972)

irregular cross-sectional geometry (Brewer and Sleeman 1960) and fine clay

pores are mainly slit shaped (Sills et al., 1973).

There have been a large number of pore-size classifications over the past 25

years, some of which are given in Table I. The range in class limits used to

describe macro, meso, and micro pores is an indication in itself of the arbitary

nature of most classifications. In this review we use Greenland’s (1977)

terminology because it is a physically based system with categories related to

the dominant water process. While Greenland’s nomenclature is preferred,

actual class limits will vary depending on mechanical analysis and composition. There are seldom unique, nonvarying functions which describe the

material properties of each soil type, and it is probably fruitless to search for a

single universal classification of pore sizes. The biological origin and significance of some of these pore sizes are indicated in Table 11.

Since t+b(O), the moisture characteristic, and K(O), the hydraulic conductivity, are the two material properties which most completely describe the

status of soil water we need to know how a change in soil structure manifests

itself in the $(O) function. We would expect structures developed from

primary soil particles to affect potentials greater than, say, - 50 to - 100 kPa

(assuming a close packing density of spherical particles) in sands and greater

than - 1.0 to - 1.5 MPa (-1 x lo3 to - 1.5 x lo3 kPa) in clays, where

domain separations are 0.01 pm. Soil pores of > 50 pm ECD, which affect




the higher potential range (equivalent to > - 6 kPa), are normally developed

from interaggregate or interped pores. These pores are the ones most

frequently altered by disturbance, including disturbance which may occur in

sample collection and preparation.

An early laboratory study of particular clarity which shows the change in

$(O) with structural alteration was described by Elrick and Tanner (1955).














Miami Silt Loam

Cu I t i v a t e d







FIG.1. 1/40)curves for undisturbed (m), sieved (<2-mm aggregates) (O),and puddled ( 0 )

Wisconsin topsoils at pa = 1.0 ton/m3. (Adapted from Elrick and Tanner, 1955.)



Some of the results are reproduced in Fig. 1. Puddled soil pastes and sieved

soils with bulk densities of 1.0 ton/m3 were compared with undisturbed cores

from the same sites over the full field desorption range. Most of the significant

difference between treatments lies in the> - 100 kPa range, corresponding to

an alteration of the interped pore geometry, as we would anticipate. Although

Elrick and Tanner do not themselves draw attention to the difference in $(O)

between the virgin and cultivated sites of the same soil, that difference is large,

with a 37 % difference in the water held between -0.01 kPa and - 50 kPa.

Such a dramatic reduction in the transmission porosity frequently accompanies a change from native vegetation to agricultural land use. Unger (1975)

demonstrated that the difference in water retention between undisturbed and

sieved samples is equivalent to the effect of tillage. Although tillage recreates

transmission pores, these are often transitory, and tillage-created cracks may

collapse within the season as the result of raindrop impact, compression of

soil by roots, and wetting-drying cycles (Hamblin and Tennant, 1981;

Dexter, 1977).

The effects of tillage on PSD and the $(O) function are not always

consistent. Some studies comparing zero tillage with ploughing show a

volume reduction of zero-tilled transmission pores within the topsoil (0-0.1

or 0-0.2 m, depending on cultivation depth) but equivalent volumes for both

treatments in the subsoil (Douglas et al., 1980). Other studies show increased

transmission porosity in surfaces of zero-tilled soils through the development

of earthworm channels and other soil faunal pores (Ehlers, 1975; Lal, 1976;

Barnes and Ellis, 1979). These conflicting views arise, I suggest, because we

are not always comparing like with like. Soils which are measured in the first

year or two of a changed management regime may not have stabilized, and

measurements taken at different times in the year are therefore not comparable. Significant soil structural changes can occur within a few weeks in

freshly tilled topsoils (Hamblin, 1982). Longer-term changes are reported

after one or several years. An increased proportion of transmission pores in

undisturbed soils depends on the number of biological active days in the year

and the level of macrofaunal activity (Ehlers, 1973). Reduction in transmission porosity of ploughed soils depends on the frequency and intensity of

tillage operations, the rainfall intensity, and the structural stability of the

system (see Section 111).

There are so many combinations of management and pore-geometry

interaction that the literature in this area offers little consistent predictive

information on specific tillage practices. Deep cultivation (subsoiling), for

example, is designed to increase transmission porosity in massive or compacted subsoils, but the superimposition of different tillage systems on such

deep-tilled soils may alter the results (Negi et al., 1981). Similarily, the crop

species can interact with the soil to produce different pore sizes, depending on



the geometry of the root system (tap roots may enlarge transmission pores

greatly and yet compress the surrounding pore walls), the root density, and

total depth. Fahad et al. (1982) studied the influence of crop species on soil

structure in an experiment which compared continuous monoculture with

various crops in rotations. There was a reduction in aggregate stability

associated with the continuous monoculture which was correlated with

reduced 8 at all values of II/, but particularly in the > -60-kPa range

(transmission zone).

Structural differences may have a significant influence when a range of soils

within a textural group are compared. Conventionally, soil physicists and

agronomists have tended to attribute most of the differences in the shape and

position of the $(8) function to differences in clay and sand content. This view

is oversimplified.A recent statistical analysis by Williams et al. (1983). found

pedality (the presence or absence of visible ped structure) to be the most

important distinguishing attribute within groups of similar moisture characteristics. While there were usually good positive correlations between the

values of 8 at low potentials (such as -1500 kPa) and clay content, the

correlations at - 10 to -0.1 kPa were generally much weaker. Greacen and

Williams (1983) made an interesting study of 244 horizons of Australian soils

to compare poorly and well-structured members of 14 textural groups.

Between potentials of -10 and -1500 kPa [the available water capacity

(AWC)], well-structured soils in the coarse- and medium-textured soils

contained one-third to two times the water of comparable poorly structured

soils. The greatest differences occurred in the fine sandy loam to loam classes.

Similarly,studies on the effects of cultivation damage and subsoil compaction

in silty soils (Hamblin and Davies, 1977; Ehlers, 1973) showed the special

susceptibility of these soils to loss of transmission porosity.



Water flow in porous media depends upon a hydraulic gradient composed

of the total difference in pressure potential between two points. The flux is

dependent upon the product of this gradient and the water content, and the

rate of flow is controlled by the pore geometry. The effect of pore geometry is

described by the Darcy “constant” K , which, in experimental terms, is

normally the unknown. Q (the flux) and d$/dz (hydraulic gradient) are the

measured variables. In the one-dimensional, steady-state system,



- KdII//dz

where Q is the flux in cubic meters per second, K is the hydraulic conductivity

in meters per second, and d$/dz is the hydraulic gradient over depth z.



Although an expression describing the pore structure thus sits centrally in

the flow equation, this parameter is clearly not a constant. [A more realistic

description might be obtained from modeling the PSD as bundles of

different-sized capillary tubes and applying an appropriate model such as the

Poiseuille-Hagen equation (see Section VI). Yet even this is obviously a

simplification of the real complexity of the porosity of soils.] Variations in

measured values of saturated conductivity ( K , ) at any one site may have

coefficients of variation [CV = standard error divided by sample mean; as a

percentage, i.e., (s/X)lOO] of 100-200 %, while CVs of 200-400 % are reported

for K ( 0 ) (Warrick and Nielsen, 1980). In comparison, the CVs of static

properties such as 6 and pb are generally less than 10 %. While the methods

for measuring K , and K(6) are less precise than for other parameters, it is the

inherent heterogeneity in the pore geometry both vertically in the profile and

spatially in the landscape which accounts for most of this variation. The

functional form of K ( 6 ) thus varies considerably depending on .the dimensions over which it is measured. For one-dimensional, steady-state flow, a

logarithmic function based on geometric means has received a consensus

(e.g., Bouwer, 1969; Nielsen et al., 1973).

Laboratory determinations of K are generally made on undisturbed cores

and values are obtained vertical to the soil surface. Simple field methods for

K , and transient K ( 0 ) also measure one-dimensional K vertically. Hydrologic

methods for saturated groundwater flows generally use pump methods,

which measure horizontal flux. Such values of K cannot strictly be extrapolated to different spatial scales or vectors except where the material is

isotropic, which is rare in agricultural soils. The theory, measurement, and

implications of the anisotropy of pore structures on K have developed from

many independent sources, but much of the more recent work rests on that of

Childs and his colleagues (Childs, 1952; Childs et al., 1957). In a field study of

anisotropy, Childs et al. (1957) measured vertical ( K , ) and horizontal (Kh)

contributions to flow in East Anglian fluvioglacial materials. In these recent

depositional sediments, anisotropy ( K JKv) varied from

to lo", with

about half of the 13 sites being isotropic. Soils of similar texture varied

substantially both in K , values (e.g., three alluvial clays had values ranging


to 10-9m/s) and anisotropy; the same three soils had K d K ,

values of 0.03, 200-70,000, and 1.0.

We should expect K JKv values of less than 1 in soils with strongly marked

vertical cracking (Vertisols, Natric Ultisols) and K JKv values of greater than

1 in soils with laminar bedding planes and high proportions of platy particles

such as silt. However, our assumptions on the preferred flow vector can be at

fault. Bouma et al. (1981) give an interesting example of the hydraulic

conductivity surrounding pipe drains in massive clay subsoils. Pipe installation was predicted to have resulted in the smearing of the layer beneath the



pipes, and backfilling to have given a more open structure above the pipes,

but measuring K above (a) and below (b) the pipes ( K , , Kb), they found the

Kb values to be the same as the surrounding undisturbed soil, whereas the K ,

values were generally less than K,.

While considerable progress has been made in the number of measurements and methods used to obtain K(B) over the past two decades, structural

heterogeneity has hampered data collection and reduced the utility of many

of the measurements made. Alternative indirect approaches for obtaining

K(8) which attempt to reduce the measured variability will be discussed in

Section IV.

Pore continuity has particular significance in agricultural environments,

where so many management practices (tillage, land clearance, root-crop

harvesting, and landforming) tend to disrupt naturally continuous pores and

produce one or more structurally discontinuous layers. I consider this has

great significance in crop-water relations and is not, as yet, always appreciated. A typical example of the effect this has upon the hydraulic gradient is

0 -10 -20 -30 -40 0

-10 -20 -30 -40

Hydraulic Head (kPa)

Frc.2. Hydraulic gradients at 0.1 (0)

and 0.01 ( 0 )pm/s flow rates for undisturbed,

wheat-fallow, and wheat-pea rotations of a Walla-Walla silt loam. (After Allmaras et al., 1982.)

given by Allmaras et al. (1982), and is shown in Fig. 2. A reduction in the

proportion of larger pores in the top 30 cm of an old (50 year) cultivated silt

loam gave a 10-fold reduction in K(8) at potentials greater than - 10 kPa

compared with undisturbed grassland. This reduction in K(B) was accompanied by longer periods of wet, anaerobic conditions in winter and spring,

increased nitrogen loss, lower pH, and lower biological activity. Steady-state

hydraulic gradients of the long-cultivated sites showed a double inflection,

which is probably very typical of many agricultural soils.



In recent years much attention has been focused on the role of more or less

vertical, continuous large pores in both saturated and unsaturated water

flow. These have been generally termed “macropores”, as in a review by

Beven and Germann (1982) which summarizes their hydrological significance. Earthworm channels tend to be randomly distributed, with spatial

densities as high as 900m-2 (Bouma et al., 1982) in temperate Europe or

500 m - 2 in Mediterranean climates (Barley, 1959). They have been reported

as continuous to 0.7 m by Ehlers (1975) and 1.6 m by Bouma et al. (1982), but

Barnes and Ellis (1979) noted considerable annual variation in depth

depending on the depth and duration of wetted soil profiles. Despite

considerable swelling and shrinkage movement of the clay soil studied by

Barnes and Ellis, the wormholes persisted for several years. Cultivation has

been found to reduce the numbers of both holes and worms very drastically

(Ehlers, 1975). One of the major factors in long-term inprovement of many

zero-tilled soils has been the development of wormhole transmission pores.

Earthworms are not found in alkaline soils, however, nor are they abundant

in soils with low organic matter, especially where there is little surface plant

litter. We might therefore anticipate more widespread effects of large biopores from root channels than from wormholes.

Persistence of channels created by crop roots is a common observation,

although it is seldom quantified. Many incidental references to new roots

growing down old root channels occur (e.g., Ehlers et al., 1983). Where such

channels do not become hydrophobic from detached, lignified, cortical cells,

they could have a pronounced effect on infiltration and through-drainage.

Some of the pioneering work on the effect of root growth on pore structure

was carried out by Barley (1953, 1959). Barley considered the effect of

earthworm channels less significant than that of root channels in a red brown

earth, since the number of root channels exceeded wormholes of > 500 pm

diameter by a factor of 10 in the topsoil. Later data (Barley, 1970) for cereal

root numbers in the same soil suggest that at 0.5 m root channels would

outnumber wormholes by lo3. Barley argued that the large diameter of

wormholes reduced their role in water movement (though not their significance to gas exchange), as they would not fill until the soil (in this case a red

brown earth or Haploxeralf) was near saturation. The functioning of these

larger transmission pores in soils is not yet completely resolved. It is central,

however, to the concept of preferred pathway flow, and we return to it in

Section IV,B,l. In passing, it is interesting to note that when Sedgley and

Barley (1958) measured the effects of grass roots on the hydraulic properties

of a sandy loam, they found a reduction in K ( 0 ) at Ic/ = - 3.0 kPa when roots

had been grown in the soil. They interpreted this as compression by the roots

which had reduced pore space. Nevertheless, Sedgley and Barley’s work may

be open to an alternative explanation, as put forward in a study by Gish and



Jury (1983). These authors found the effect of root channels (with both living

and decayed’roots in them) was to increase the proportion of immobile water,

that is, water which has entered “dead-end” or stagnant pores, when solute

flow was measured. The proportion of immobile water increased overall from

< 10 % in precropped soil columns to 23 % where living roots were present to

39% in the case of decayed roots. Thomas and Phillips (1979) quoted

examples where more than 50 % of new water additions into wet soil bypass

the matrix water, rather than displacing it, because of preferred pathway

transport. Although solute transport falls outside the scope of this review, it is

worth noting that many predicted outflow curves of solute transport can only

be made to “fit” observed values by designating some of the volume to

stagnant pores. In field studies on solute transport, such as that of Omoti and

Wild (1979), where fluorescent or colored dyes were used, preferential

pathways are frequently a marked and characteristic feature ahead of the

general wetting front.

Intuitively we expect root channels to increase the hydraulic conductivity

at $ > - 0.3 to - 6 kPa, equivalent to ECDs of 1000 to 50 pm, which would

reduce the number of occasions when surface soils are waterlogged by

transient perched water. One could argue, however, that because cultivation

severs the continuity of root systems and probably blocks the upper openings

of root channels, these effects will not be found in many cultivated soils

(Greenland, 1977). Logically, therefore, it is in zero-tilled or pasture soils that

continuous root channels should have their most pronounced effects on K(0).

Yet many of the comparative studies in the literature between ploughed and

undisturbed soils do not bear this out. This sometimes occurs because there

are other interactions masking the effects of roots, such as the muchincreased role of earthworms in mulched, direct-drilled soils in West African

(La1 et a/., 1980) and British (Barnes and Ellis, 1979) environments. In other

cases zero-tilled crops have not grown so vigorously and have produced less

root growth. Hamblin and Tennant (1981) found that a reduced proportion

of large pores on a zero-tilled loamy sand in Western Australia (where there

were no earthworms) was reflected in the slower movement of water through

the topsoil of the undisturbed treatment and in reduced root growth

associated with higher soil strength. This situation persisted unchanged over

a 5-year continuous cropping period (Hamblin et a!., 1982) with no development of a significant number of continuous root channels or increased flow

rates over time.

Continuous vertical fissures created by shrinkage are at their most

pronounced in soils of high smectite content. They are rare in soils dominated

by “low-activity’’ clays (that is, those soils of low cation capacity and/or

charge density) such as Oxisols and Ultisols. Soils with clay contents of

greater than 30 % have a continuous clay matrix, and their physical proper-



ties are determined by clay type rather than clay content. In Vertisols these

fissures may extend to over 1 m and do not always close completely, even

when soils are fully hydrated (Warkentin, 1982). Minimum horizontal

spacings of such cracks vary from <0.07 m to >0.5 m (Chan, 1981), with

larger spacings developing at depth. The surfaces are often “self-mulching”

when subjected to wetting-drying cycles, producing large numbers of stable

peds of 1000 to 5000 pm diameter. They thus possess a distinct trimodal pore

system of large noncapillary cracks, small capillary cracks between

micropeds, and storage pores of <0.5 pm within peds. With such welldeveloped fissure systems and substantial water-dependent volume changes,

Vertisols would seem to be the extreme example of soils whose structure

influences their material properties. Curiously, their K(8) and +(S) relationships may be more predictable than many other less “stable” soils, with less

pronounced, visible structures. Hydraulic behavior and crop water use are

particularly hard to predict if these structures are unstable.




In common parlance “soil structure” is often taken to be synonymous with

structural stability. Stability should be viewed separately. A perfectly elastic

or rigid, porous body maintains dimensional integrity if a stress is imposed on

it so that the various pore scales within the system stay in the same relative

positions to each other. In soils this is not the case. When water flows small

particles are caught up by viscous drag and carried in suspension in the fluid.

Surface tension may pull grains or aggregates out of position as the air-water

interface moves from pore to pore. However, a soil is considered stable if the

pores retain physical intergrity upon wetting and drying, even if total volume

change occurs and the pore space changes its relative spatial coordinates.

This type of stability relates to the behavior of the colloids. We commonly

talk of soil “water,” but we should remember that soil water is a solution of

low ionic strength, normally with an electrolyte concentration of 0.001 to

0.01 M . The composition of this water can profoundly influence the swelling,

dispersion, and flocculation behavior of charged particles in the soil.

Aggregated soil particles may collapse (slake) and the entrained clay

particles subsequently disperse spontaneously or as a result of mechanical

forces. These forces include differential wetting of large and small clay

crystals, hydrous oxides, and organic matter; raindrop impact; shearing and

compression by implements, wheels, and roots; and rupturing of pores

Tài liệu bạn tìm kiếm đã sẵn sàng tải về

II. Soil Structure: Components of the Soil–Pore System

Tải bản đầy đủ ngay(0 tr)