Soil and Environmental Analysis: Physical Methods - Chapter 2 doc

Matric potentialcmis a subcomponent of pressure potentialand is defined as the value of cpwhere there is no difference between the gaspressure on the water in the reference state and that

it is the major factor that determines the availability of water to plants After lowing for differences in elevation, differences in matric potential between differ-ent parts of the soil drive the unsaturated flow of soil water (Chap 5).

al-A Definition

The soil physics terminology committee of the ISSS provided agreed-upon nitions for total potential and its various components (Aslyng, 1963), which wereslightly modified in 1976 (Bolt, 1976) A brief summary is given here More de-tailed discussions of the meaning and significance of these definitions are given insoil physics books such as those of Marshall et al (1996) and Hillel (1998).Total potential of soil water can be divided into three components:

The pressure potentialcpis defined as ‘‘the amount of useful work that must bedone per unit quantity of pure water to transfer reversibly and isothermally to thesoil water an infinitesimal quantity of water from a pool at standard atmosphericpressure that contains a solution identical in composition to the soil water and is

at the elevation of the point under consideration’’ (Marshall et al., 1996) Similardefinitions have been given for gravitational potential, cg, and osmotic poten-tial, co, which refer to the effects of elevation (i.e., position in earth’s gravita-tional field) and of solutes on the energy status of soil water The sum of gravi-tational and pressure potential is called the hydraulic potentialch Differencesbetween the hydraulic potential at different places in the soil drive the move-ment of soil water Matric potentialcmis a subcomponent of pressure potentialand is defined as the value of cpwhere there is no difference between the gaspressure on the water in the reference state and that of gas in the soil.

The above definition of pressure potential includes (1) the positive static pressure that exists below a water table, (2) the potential difference experi-enced by soil that is under a gas pressure different from that of the water in thereference state, and (3) the negative pressure (i.e., suction) experienced by soilwater as a result of its affinity for the soil matrix In the past, some authors (Taylorand Ashcroft, 1972; Hanks and Ashcroft, 1980) have used the term ‘‘pressurepotential’’ to refer only to subcomponents 1 and 2 However, all authors useequivalent definitions for matric potential, which is subcomponent 3 Matric po-tential can have only a zero or negative value As water becomes more tightly held

hydro-by the soil its matric potential decreases (becomes more negative) Matric or soilwater suction or tension refers to the same property but takes the opposite sign tomatric potential In a swelling soil, overburden pressure can cause a slight error inapplications where it is intended to relate matric potential to soil water content(Towner, 1981)

The sum of matric and osmotic potential is called the water potentialcwand is directly related to the relative humidity of water vapor in equilibrium withthe liquid phase in soils and plants.cwis an important indicator of plant waterstatus and is also important in saline soils, where the osmotic potential of the soilsolution is sufficient to influence plant water uptake

B Units

Since potentials are defined as energy per unit mass, they have units of joules perkilogram However, it is also possible to define potentials as energy per unit vol-ume or per unit weight Thus, since the dimensions of energy per unit volume areidentical to those of pressure, the appropriate unit is the pascal (1 bar⫽ 100 kPa).Similarly, the dimensions of energy per unit weight are identical to those of length,

so the appropriate unit is the meter Because it is common to refer to the pressure

due to a height h of a column of water as a pressure head (or simply head) h, this

term is often used to describe the potential energy per unit weight The relation

c (m)

⫺1

g

where g is the density of water and g is the acceleration due to gravity

(⬃ 1000 kg m⫺3and 9.81 m s⫺2, respectively), is used to convert potentials fromone set of dimensions to another A logarithmic (pF) scale (Schofield, 1935),where

pF ⫽ log (negative pressure head in cm of water)10 (3)has also been used

MATRIC POTENTIAL

The main features of methods for measuring matric potential and the addresses ofsome manufacturers and suppliers are given inTable 1 The web sites for many ofthe manufacturers list their suppliers in many countries In considering the cost

of instruments, it is important to decide whether a data logger is required, and toconsider the cost of the logger or meter as well as the cost of the sensor, sincesome sensors are more easily logged than others and some are available withcheap loggers Consequently Table 1 should be treated only as an initial guide topurchase, because of the pace of development in the choice of loggers and meters.There are many earlier reviews of the design and use of such methods (Marshall,1959; Rawlins, 1976; Cassell and Klute, 1986; Rawlins and Campbell, 1986).Methods have been classified according to the measurement principle involvedand are discussed in detail in the following sections Tensiometers (Sec III) con-sist of a porous vessel attached via a liquid-filled column to a manometer Porousmaterial sensors (Sec IV) consist of a porous material whose water content varieswith matric potential in a reproducible manner; a physical property of the materialthat varies with its water content is measured and related to matric potential using

a calibration curve Psychrometers (Sec V) measure the relative humidity of watervapor in equilibrium with the soil solution Because they measure the sum ofmatric and osmotic potentials, they are also readily applicable for measurements

in various parts of plants

There have been large improvements in the performance and availability ofdata loggers over the past ten years, some improvements in methods for measur-ing potential, and a growing use and awareness of the importance of measure-ments of potential Despite this, there is still a need for a single sensor that can logmatric potential to a field accuracy that is sufficient for understanding water move-ment and soil aeration under wet conditions (e.g 0 to⫺100 ⫾ 0.2 kPa) whilebeing able to measure to a reasonable accuracy (say⫾ 5%) down to ⬍ ⫺1.5 MPa.This is a tall order, but it explains the continuing interest in the osmotic tensiom-eter and improved porous material sensors

III TENSIOMETERS

A tensiometer consists of a porous vessel connected to a manometer, with all parts

of the system water filled (Fig 1) When the cup is in contact with the soil, films

of water make a hydraulic connection between soil water and the water within thecup via the pores in its walls Water then moves into or out of the cup until the(negative) pressure inside the cup equals the matric potential of the soil water.The following equations are used to obtain matric and hydraulic potentialfrom the mercury manometer readings shown in Fig 1

The factor of 12.6 is the difference between the relative densities of mercury

and water c is a factor to correct for the capillary depression that occurs at the mercury–water interface If g is omitted from these two equations, they will give

the potentials in head units

Fig 1 Mercury manometer tensiometer

Tensiometers are also available with Bourdon vacuum gauges, with pressuretransducers (for data logging), and for portable use Cassell and Klute (1986) pro-vide a good discussion of methods for installing and maintaining tensiometers.

I have discussed limitations common to most designs before considering each type

be-at microscopic irregularities within the instrument At such a low pressure relbe-ative

to atmospheric pressure these bubbles expand, augmented by dissolved air comingout of solution, and can eventually block the tubing, making further readings un-reliable Filling with deaired water, which has had some of its dissolved air re-moved by boiling or by leaving it for some hours under a vacuum, is done tocounteract this effect Despite this, because dissolved air tends to move into theporous cup and come out of solution, tensiometers often incorporate an air trapthat allows air to collect without blocking the instrument (Fig 1) However, sincethis air causes the reponse time to increase (become slower), it is usual to ‘‘purge’’tensiometers at regular intervals (ca weekly or less often under cool wet condi-tions) by replacing the trapped air with deaired water (Cassell and Klute, 1986).The temporary release of suction during purging allows some water to pass intothe surrounding soil so that readings are not reliable for some time after purging

2 Response Time

Because any change in matric potential will cause a change in the volume of uid in the tensiometer, time is required for this water to move into or out of theinstrument and hence for it to respond The conductance of the porous cup andthe unsaturated hydraulic conductivity of the soil control the response time aswell as the amount of water movement required for a given change in potential(the ‘‘gauge’’ sensitivity) Mercury manometers and Bourdon vacuum gauges aremuch less sensitive than pressure transducers However, since most tensiometersoperate with some trapped air within them, and since their tubing is not com-pletely rigid, differences in response time between pressure transducers and othertensiometer types are much less than would be expected from the sensitivity ofthe gauges

liq-A tensiometer is said to be tensiometer limited if its response time is notinfluenced by soil properties, but only by the cup conductance and gauge sensi-tivity; otherwise it is soil limited Tensiometer-limited response time is inverselyproportional to cup conductance and gauge sensitivity (Richards, 1949), and cups

with 100 times greater conductivity than normal cups are available for specializedapplications It is not difficult to obtain tensiometer-limited conditions, although

in some soils tensiometers may be soil limited in drier soils (Towner, 1980).Tensiometer-limited conditions are advantageous because instrument be-havior is reproducible and not dependent on variable soil conditions (Klute andGardner, 1962) This is particularly important when the potential is changing fast.However, obtaining a tensiometer-limited response is not the main considerationwhen tensiometers are used to monitor field conditions over periods of weeks ormonths and are read at infrequent intervals Furthermore, too high a sensitivitycan cause problems if the tensiometer is then too sensitive to other factors that cancause a change in the liquid-filled volume such as temperature changes (Watsonand Jackson, 1967) and bending of the tubing In field use, all tensiometer tubingshould be shaded from direct sunlight where possible Otherwise, sudden expo-sure to the sun can cause the tubing (and any air it contains) to expand and tem-porarily perturb the readings High sensitivity/fast response tensiometers requirecareful handling and operate better under laboratory conditions

Porous cups are usually made of a ceramic and must have pores that aresmall enough to prevent air from entering the cup when it is saturated The cupmust also have a reasonably high conductance Ceramic tensiometer cups for fielduse have a conductance of about 3 · 10⫺9m2s⫺1, and even a mercury-manometertensiometer with such a cup will have a (tensiometer-limited) response time ofabout one minute in the absence of trapped air (Cassell and Klute, 1986), morethan adequate for most field use

B Mercury Manometer and Bourdon Gauge Tensiometers

A manometer scale can easily be read to the nearest millimeter, so that mercurytensiometers have a scale resolution of ⫾ 0.1 kPa However, with the smallest(1.7 mm diameter) nylon tubing commonly used for the manometer, there is asignificant capillary correction (⬃ 0.8 kPa) and hysteresis, caused by the mercurymeniscus sticking to the walls of the tube If the tube is agitated, to cause a smallfluctuation in the mercury level, an accuracy of ⫾ 0.25 kPa can be achieved;otherwise much larger errors can occur (Mullins et al., 1986) Bourdon vacuumgauges are less accurate, typically with a scale division of 2 kPa, but friction inthe gauge mechanism and the difficulty of setting an accurate zero further limittheir accuracy Mercury tensiometers suffer from the environmental hazard ofmercury and require a 1 m manometer post but are preferable if high accuracy isrequired (e.g., when measuring vertical gradients in hydraulic potential)

Mercury tensiometers can be constructed very cheaply, without the need forworkshop facilities (Webster, 1966; Cassell and Klute, 1986) Where several ten-siometers are used in the same vicinity, it is common to share a single mercury

reservoir among 6 –30 tensiometers Because the mercury withdrawn from thereservoir will cause a slight drop in its level, for high accuracy, the level should bemeasured each time a reading is taken, or the reservoir should have a cross-sectionmany times greater than the sum of the cross-sections of the tubes that dip into it.

It is also advisable to check each tensiometer for air leaks before installation This

is done by soaking the cup in water, then applying an air pressure of 100 kPa tothe inside of the tensiometer while it is immersed in water (Cassell and Klute,1986) To minimize thermal effects, the manometer tubing should be shieldedfrom direct sunlight (e.g., by facing the manometer post away from the middaysun) With prolonged outside use, some plasticizer may come out of the nylontubing and collect as a white deposit, which can eventually block the tube Wehave not found this to be a problem over a single season, but 1.7 mm tubing mayneed to be occasionally replaced over longer periods

C Pressure Transducer and Automatic Logging Systems

Because pressure transducers have a high gauge sensitivity, they are particularlyuseful when a short response time is important They can also be used with dataloggers Transducers (e.g., piezoresistive silicon types) that are not temperaturesensitive and have a precision of⫾ 0.2 kPa can be bought for ⬃ $140 Types thatare vented to the atmosphere should be used so that changes in atmospheric pres-sure have no effect

In the unusual case that matric potentials are required at a considerabledepth (say 10 m), a pressure transducer located close to the measuring depth isessential because a hanging water column will break once the tension in it ap-proaches 100 kPa

1 Automatic Logging Systems

Automatic logging systems are required at remote sites, when measurements arerequired more often than the site can be visited, and to study laboratory or fieldsituations in which many measurements are required over a period of hours ordays (e.g., drainage studies) In the former case a provision for automatic purgingmay also be necessary if weekly visits (or less frequently in wet conditions) arenot possible Systems that use a motor-driven fluid-scanning switch allow a num-ber of tensiometers to be connected each in turn to a single pressure transducer(Anderson and Burt, 1977; Lee-Williams, 1978; Blackwell and Elsworth, 1980)

It is necessary to have a transducer attached to each tensiometer if veryshort measurement intervals are required because re-equilibration, when a trans-ducer is switched between tensiometers at different potentials, can take 2 minutes(Blackwell and Elsworth, 1980) or more (Rice, 1969) The effect of temperature

Table 1 Methods, Range, Accuracy, Typical Cost, and Suppliers for Measuring Matric (cm) or(Where Indicated) Water (cm) Potential

Method, range, and accuracya

Unit cost(U.S.$)

Manufacturers/suppliersand References

Tensiometers (0 to ⴑ85 kPa)

Mercury manometer,ⱕ ⫾ 0.25 kPa 30⫹ post

& Hg

Homemade with commercialcups (Webster, 1966; Cas-sell and Klute, 1986)Ceramic cups for tensiometers 15 E, F

Pressure transducer: normal, miniature,c⫾

0.2 kPa

250, 450 B, G, HPortable Bourdon gauge,⫾ 2 kPa, but see text 1,000 C, D, F (Mullins et al., 1986)Puncture tensiometer,ⱖ ⫹ 0.7 kPa (system-

atic)⫹ portable readout

1 All suppliers of Whatman filter

paper (Deka et al., 1995)

(⫺100 to ⫺1000 kPa) ⫾ 5% ⫹ portable d

meter

Psychrometers ( cw), all for disturbed

samples except the Spanner psychrometer

Isopiestic (0 to⬍ ⫺40 MPa) ⫾ 10 kPa 15,000 (see text) (Boyer, 1995)Dew point (0 to⫺40 MPa) ⫾ 100 kPa 4,500 A

Richards (0 to⫺300 MPa) ⫾ 5–10% ⫹ meter 2,500⫹ 2,500 A (but may no longer be

available)Spanner (0 to⫺7 MPa) ⫾ 5–10% ⫹ meter 40⫹ 2,600 I (field/in situ measurement)

aAccuracy represents the best reliable reported values or manufacturers’ figures, but see text for details, since accuracy can be limited by a number of factors.

bKey (many web sites list local suppliers): A, Decagon Devices Inc., U.S.A ( http://www.decagon.com ) B, Delta

T, U.K ( http://www.delta-t.co.uk ) C, Eijkelkamp, The Netherlands ( http://www.eijkelkamp.com ) D, ELE ternational Ltd., U.K ( http://www.eleint.co.uk ) E, Fairey Industrial Ceramics Ltd., Filleybrook, Stone, Staffs., ST15 0PU, U.K F, Soilmoisture Equipment Corp., U.S.A ( http://www.soilmoisture.com ) G, Skye Instruments Ltd ( http://www.skyeinstruments.com ) H, UMS GmbH, Germany ( http://www.ums-muc.de ) I, Wescor Inc., U.S.A ( http://www.wescor.com ).

In-cCan be used with data loggers ($1000 –3000).

fluctuations on readings, which is most notable where nylon tubing is exposedabove ground (Watson and Jackson, 1967; Rice, 1969), is also minimized with thetransducer attached directly to the tensiometer Such tensiometers and loggers arecommercially available (Table 1).

2 Systems with Portable Transducers (Puncture Tensiometers)

A puncture tensiometer consists of a portable pressure transducer attached to ahypodermic needle that can be used to puncture a septum at the top of a perma-nently installed tensiometer and hence measure the pressure inside it (Fig 2)(Marthaler et al., 1983; Frede et al., 1984) In this way, one transducer and readoutunit can be used to measure the pressure in a large number of tensiometers Eachtensiometer simply consists of a porous cup attached to the base of a water-filledtube topped by a rubber or plastic septum that reseals each time the needle isremoved A small air pocket is deliberately left at the top of each tensiometer toreduce any thermal effects on the reading and the small pressure change caused

Fig 2 Various tensiometers From left to right: data logger attached to a pressure ducer tensiometer (only the top part with cover removed to reveal transducer); Webster(1966) type mercury manometer tensiometer; ‘‘quick draw’’ portable tensiometer (case,auger, and tensiometer); portable tensiometer with a pressure transducer and readout; punc-ture tensiometer without, and with, portable meter attached

trans-by the introduction of the needle The needle and sensor are designed to have avery small dead volume to minimize this However, Marthaler et al reported sys-tematic errors of⬃ 0.7 kPa in potentials close to zero (⫺2 to ⫺3.6 kPa) but agood overall relation between mercury manometer and puncture tensiometer read-ings Eventually the septum needs to be replaced, and careful insertion is required

to ensure that there is no leak into the system Consequently, these devices are not

as accurate as systems with an in situ manometer or pressure sensor

D Portable Tensiometers

Portable tensiometers with Bourdon vacuum gauges (Table 1) and ones with apressure transducer (available from UMS, Table 1) that can be read to⫾ 0.1 kPaare commercially available These can be stored with their tips in water when not

in use so that there is little accumulation of air within them, and they rarely need

to be refilled They can be used when single or occasional measurements are quired However, they cannot usually give a reliable reading quickly after insertionbecause of the effect of soil deformation during insertion Mullins et al (1986)found that re-equilibration of the disturbed soil with that surrounding it took only

re-a few minutes in soil re-at⬎ ⫺5 kPa but ⬎ 2 h in soil at ⬍ ⫺30 kPa (irrespective ofthe use of the null-point device supplied on one model)

E Osmotic Tensiometers

Peck and Rabbidge (1969) described the design and performance of an osmotictensiometer It consists of a cell containing a high molecular weight (20,000)polyethylene glycol solution confined between a pressure transducer and a semi-permeable membrane supported behind a porous ceramic The cell is pressurized

so that it registers 1.5 MPa when immersed in pure water, allowing the tensiometer

to measure matric potentials between 0 and⫺1.5 MPa However, there were lems due to polymer leakage and sensitivity to temperature changes (Bocking andFredlund, 1979) Biesheuvel et al (1999) have used an improved membrane toprevent leakage and have shown how readings can be corrected for temperatureeffects Their tensiometer had an accuracy of⬍ 10% at potentials ⬍ ⫺100 kPa.The technique is promising but requires further development and testing in soil

prob-to demonstrate that it has long-term stability and acceptable accuracy and sponse time

These sensors are made of a porous material whose water content varies withmatric potential in a reproducible manner A physical property of the material

that varies with water content is measured and related to matric potential, using

a calibration curve Sensors based on the measurement of the water content offilter paper, electrical conductivity, heat dissipation, and dielectric constant arediscussed

Irrespective of the method used to measure the water content of the porousmaterial, its physical properties determine the range of matric potentials overwhich the sensor will be sensitive and accurate Sensitivity depends on the rate ofchange of water content with matric potential, and hence on the pore size distri-bution of the porous material A major limitation to accuracy is the amount ofhysteresis that the material displays, and special materials have been developed tohave low hysteresis and good sensitivity for recently developed sensors The po-rous material is calibrated by equilibrating it at a set of known matric potentials.The reliability of published calibration curves or those supplied by manufacturersdepends on how closely the water characteristic of the sensor resembles that ofthe sensor used in the original calibration For greater accuracy, users should cali-brate all, or a representative sample, of their sensors in the range of interest Apartfrom the filter-paper technique, which is used on disturbed samples, the othersensors described here are nondestructive and can be logged Because their re-sponse time will depend on the amount of water that has to flow out of the sensorfor any given change in potential, there will be a lag in response, especially at lowpotentials Sensitivity and accuracy also vary along the sensing range Since theaccuracy figures quoted by manufacturers normally refer to optimal conditions(laboratory equilibration at constant temperature and the most accurate portion ofthe sensing range using calibrated sensors), these should be treated with consid-erable caution Finally, when left in the soil the sensors are likely to accumulatefine material, including microbial debris that can progressively clog the pores, sothat it is desirable to recheck the calibration after prolonged field use Althoughelectrical resistance sensors are becoming much less popular due to the availabil-ity of better techniques, the sections on the sensor material, response time, hys-teresis, and calibration of these sensors are of relevance to all porous materialsensors

A Filter Paper Method

The filter paper method, originally used by Gardner (1937) as a simple meansfor obtaining the soil water release characteristic, is a cheap and simple methodfor measuring matric potential that is only beginning to receive the use it de-serves The method consists of placing a filter paper in contact with a soil sample(⬎ 100 g) in a sealed container at constant temperature until equilibrium isreached The gravimetric water content of the filter paper is then determined, andthis is converted to matric potential using a calibration curve Apart from cali-brated filter papers, this technique requires only a homemade lagged sample

equilibration box, an oven set at 105⬚C, and a balance accurate to ⫾1 mg Deka

et al (1995) give a full description of how to perform the technique

The water retention characteristic of a filter paper (which is its calibrationcurve) can usefully cover a wide range of potentials from⫺1 kPa to ⫺100 MPa(Fawcett and Collis-George, 1967) At the wetter end of this range, equilibrationoccurs by liquid water flow between soil and the filter paper It is therefore impor-tant that the soil sample makes good contact with the paper and fully covers it It

is best to sandwich the paper between two halves of a core or two layers of soil.Vapor equilibrium becomes increasingly important in dryer soil, so that the paperresponds to the water potential Vapor equilibration is a slower process Althoughequilibration times from 3 to 7 days have been used (Fawcett and Collis-George,1967; McQueen and Miller, 1968; Hamblin, 1981), Deka et al (1995) have shownthat at least 6 d was required for full equilibration, even at⫺50 kPa, although thiswas still sufficient at⫺2.5 MPa Small temperature fluctuations during equilibra-tion can disturb the process and may even cause distillation (i.e., condensation ofwater on the walls of the container) (Al-Khafaf and Hanks, 1974) To avoid theseproblems, the sealed containers should be kept thermally insulated in Styrofoam(expanded polystyrene) containers, out of direct sunlight, and in a room or cup-board that does not have a large diurnal temperature variation (Campbell andGee, 1986)

Since the potential of a sample can be altered by deformation, it is important

to use an undisturbed soil core or soil that has been removed with minimal bance, to transport it with a minimum of vibration, or to equilibrate it in situ(Hamblin, 1981) Hamblin has also used the technique in situ by introducing pa-pers into slits cut with a spatula in field soils

distur-Many authors have found it necessary to impregnate their filter papers toavoid fungal degradation during equilibration Both 0.005% HgCl2 and 3%pentachlorophenol in ethanol have been successfully used by moistening the fil-ters, which are then allowed to dry before use This has not been found to affectthe calibration curve (Fawcett and Collis-George, 1967; McQueen and Miller,1968) We have not found that a fungicide was necessary for equilibration times

of up to 7 d, but this probably depends on soil type Various methods have beenproposed to cope with the soil that can stick to the equilibrated filter paper Often

it can be detached by a combination of flicking the paper with a fingernail andusing a fine brush Gardner (1937) corrected for the mass of soil adhering to thepaper by determining its oven-dry mass (when it was brushed off the dry paper)and then back-calculating what its moist mass would have been from a knowledge

of the water content of the soil sample It is also possible to use a stack of threepapers and only use the central one for measurement (Fawcett and Collis-George,1967) However, we have found that this is often less accurate than using a singlepaper and that the central paper does not always reach equilibrium

1 Calibration and Accuracy

Because filter papers have a measurable hysteresis (Fawcett and Collis-George,1967; McQueen and Miller, 1968; Deka et al., 1995) it is necessary to bring them

to equilibrium in the same way during calibration as when they are used Thus,since the filter papers are dry before use, they should be calibrated on their wettingcurve (Fawcett and Collis-George, 1967; Hamblin, 1981) Calibrations can be per-formed using a tension table, pressure plate, psychrometer, and/or vapor equili-bration to cover different parts of the calibration (Campbell and Gee, 1986; Deka

log (10 ⫺c ) ⫽ 5.144 ⫺ 6.699Mm for cm ⬍ ⫺51.6 kPa

log (⫺c ) ⫽ 2.383 ⫺ 1.309M10 m for cm ⬎ ⫺51.6 kPa (5)wherecmis in kPa and M is the water content of the filter paper in g g⫺1 The

‘‘broken stick’’ shape of the calibration curve is the result of water release fromwithin the cellulose fibers at low potentials and from between the fibers at highpotentials

With calibrated batches of filter papers, accuracies of⫾150% and ⫾180%can be expected between 0 and⫺50 kPa, and ⫺50 kPa and ⫺2.5 MPa, respec-tively (Deka et al., 1995) Where less accuracy is acceptable, the above equationcan be used with uncalibrated papers Because accuracy is mainly limited by thevariability in the properties of individual filter papers, the accuracy obtainablefrom calibrated batches can be improved by replicating measurements This isshown by the very good agreement between the mean value obtained from repli-cate filter papers and tensiometer measurements (Deka et al., 1995)

B Electrical Resistance

Electrical resistance sensors consist of two electrodes enclosed or embeddedwithin a porous material and have been used since the 1940s At equilibrium, thematric potential of the solution within the sensor is equal to that of the surroundingsoil Commercial sensors can be purchased cheaply (Table 1), and it is also notdifficult to construct large numbers of sensors at very little cost However, themethod is subject to a series of limitations that restrict the accuracy that can beobtained

The potential of the sensor is obtained by measuring the electrical resistancebetween the two electrodes, which is a function of the water content of the porous

material, and hence of its matric potential Unfortunately, the resistance is also

a function of temperature and of the concentration of solutes in the soil solution.Empirical equations to correct the resistance of gypsum sensors for temperatureeffects are available (Aitchison et al., 1951; Campbell and Gee, 1986) and havebeen reviewed by Aggelides and Paraskevi (1998) However, sensors cannot beused in saline soils unless the electrical conductivity of the soil solution is alsoknown or can be compensated for Scholl (1978) has described the constructionand use of a combined salinity –matric potential sensor designed to overcome thislimitation More commonly, the sensor is cast from, or contains, gypsum, whichslowly dissolves and maintains a saturated solution of calcium sulfate within it-self At 20⬚C, the solubility of calcium sulfate is about 1 g/dm3, which should bemore than ten times greater than the soil solution concentration in nonsaline soils,rendering gypsum sensors insensitive to the electrical conductivity of the soil so-lution in such soils

1 Sensor Materials and Measurement Range

Many authors have given construction details for gypsum sensors (Pereira, 1951;Cannell and Asbell, 1964; Fourt and Hinton, 1970) Other types of sensor materialhave been tried, including fiberglass and nylon encased in gypsum (Perrier andMarsh, 1958) and fired mixtures of ground charcoal and clay (Scholl, 1978) Thegeometry of the electrodes depends on the material used but must aim to minimizeelectrical conduction through the soil (e.g., by using concentric electrodes), whichwould bias the reading In practice, there are only two commercial sensors that arewidely available: the Watermark sensor and the gypsum block (Table 1) The Wa-termark sensor is 76 mm long and 20 mm in diameter, contains a proprietaryporous material held behind a synthetic membrane, and includes an internal gyp-sum tablet to neutralize solution conductivity effects Its range is from⫺10 to

⫺400 kPa ⫾ 10%, although the distributors claim that an accuracy of ⫾ 1% ispossible in the range⫺10 to ⫺200 kPa with individually calibrated sensors (Wes-cor web site) The gypsum block sensor is 32 mm long and 22 mm in diameterand covers the range⫺50 to ⫺1500 kPa

Gypsum sensors have a limited lifetime because they slowly dissolve in thesoil, and their calibration will consequently change with time (Bouyoucos, 1953;Wellings et al., 1985) Bouyoucos (1953) suggested that gypsum sensors may lastmore than 10 years in dry soil but that their useful life in very wet (or acid) soilmay not exceed 1 year Aitchison et al (1951) reported that gypsum sensors de-generate much faster in saline soils Both the durability and the calibration ofgypsum sensors depend on the source of the plaster of Paris used in their construc-tion and the ratio of plaster to water used in casting (Aitchison et al., 1951; Perrierand Marsh, 1958)