Printer Friendly

Rapid identification of soil textural and management zones using electromagnetic induction sensing of soils.


Spatial variability within a paddock was largely ignored until the advent in the late 1980s of technologies that can spatially reference continuous measurement of a property of the soil or crop (Cook and Bramley 1998). Continuous proximate sensing, commonly referred to as electromagnetic induction (EMI) sensing of soil electrical conductivity, together with precise global positioning systems (GPS) have made possible accurate mapping of within-field soil variability that can lead to site-specific management (SSM), (Plant 2001). SSM, alternatively called precision agriculture, is the management of crops and/or soils at a spatial scale smaller than the whole field (Plant 2001; Johnson et al. 2003). Traditionally, within-field soil variability has been ignored when, for example, applying fertiliser to a field; however, it is now recognised that (i) significant within-field soil variability is common, (ii) this soil variability can be identified and measured, and (iii) information from these measurements can be used to modify management of that field to increase financial benefit and decrease environmental impact (Plant 2001). Johnson et al. (2003) argued whole-field management is increasingly being viewed as inefficient, with over-application of inputs in low-producing areas and sub-optimal application in areas with high-production potential.

Apparent soil electrical conductivity (E[C.sub.a]) maps provide an invaluable initial tool for the land user to assess soil variability, which can then be followed by directed soil sampling (Lund et al. 1999; Sudduth et al. 2001). Soil E[C.sub.a] is determined by those parts of the soil that have charge associated with them: salts, clays, organic matter, and water. Therefore, it can provide an indirect indicator of important soil properties such as soil salinity, pH, clay content, cation exchange capacity, exchangeable cations, clay mineralogy, moisture content, and temperature (Kachanowski et al. 1988; McBride et al. 1990; McNeill 1992; Sudduth et al. 2001; Triantafilis et al. 2001, 2002; Beecher and Dunn 2002).

Bramley and Proffitt (1999) and Bramley (2001) discussed the more efficient management of vineyards through response to small-scale variabilities within a property. Anderson-Cook et al. (2002) described the correct classification of soils into broad categories through EMI sensing, which would help variable-rate within-field fertilisation strategies in a mid-Atlantic, North American, coastal plain landscape.

Smith and Doran (1996) described the use of soil electrical conductivity to evaluate soil quality. They chose to define soil quality by indicators that would respond most quickly to soil management changes, i.e. pH, salt content, or chemical supply. The electrical conductivity of a solution is related to the total cations or anions in the solution. Although often associated with soil salinity, when salinity is not an issue, the electrical conductivity can serve as a measure of soluble nutrients, both cations and anions. Eigenberg et al. (2002) also discussed this electrical conductivity ability, when they reported use of E[C.sub.a] maps to monitor available nitrogen levels over a growing season for a cornfield. Their study showed profile-weighted soil E[C.sub.a] values were highly correlated with soil N[O.sub.3]-N in the surface 0-23 and 23-46 cm soil layers throughout the growing season. They concluded that, using real-time monitoring, E[C.sub.a] mapping was effective in identifying the dynamic changes in available soil nitrogen.

The usefulness of E[C.sub.a] values in interpreting such dynamic soil changes is explained not only by the fact that it is determined by those parts of the soil that have charge associated with them, but also that it depends on conducting pathways for electrical charge through the soil body. There are 3 pathways acting in parallel: (i) conductance through alternating layers of soil particles and their bound soil solution, (ii) conductance through continuous soil solution, and (iii) conductance through or along surfaces of soil particles in direct contact with each other (Rhoades and Corwin 1981; Lund et al. 1999).

Changes in soil physical condition, for example reduced soil porosity during compaction, and altering conducting pathways through pores and between soil particles, can therefore be expected to alter the soil E[C.sub.a]. There is little available literature on this topic; however, an EC sensor has been incorporated into a soil cone penetrometer to enable simultaneous measurement of EC and penetration force (Sudduth et al. 2000). These authors found interrelationships between EC, soil strength (measured by resistance to penetration of the penetrometer shaft into the soil), and soil moisture, and concluded that the penetrometer-EC readings should be useful in understanding profile-weighted EC measurements. EMI mapping of soils can also be used to identity compacted areas, for example, in gateways, where increased clod size relates to increasing compaction (Godwin 2001).

In this study we investigated the use of EMI soil sensing together with precise GPS technology to map a New Zealand soilscape. Our aims were to: (i) compare the resulting E[C.sub.a] map with an existing soil map to investigate how well this rapid, affordable means of soil mapping could predict known soil mapping units; and (ii) relate the E[C.sub.a] map to measured soil physical and chemical properties considered to be important soil quality indicators.

Materials and methods


A 12-ha study area within a pastoral cropping system in Manawatu province, New Zealand, was selected for the E[C.sub.a] survey (Fig. 1). The study area, part of AgResearch Aorangi Research Station, is 12 km west-north-west of Palmerston North on the floodplain of the Oroua River. It is flanked on the western side of the river by post-4000 year BP sand dunes and associated sand plains and to the east by loess-covered, uplifted marine and alluvial terraces (Cowie 1978).


The area is in a subhumid, warm-temperate climatic zone, with rainfall distributed relatively evenly throughout the year. Average annual rainfall for Palmerston North is 963mm, with a mean winter temperature of 8.5[degrees]C and mean summer temperature of 17.9[degrees]C ( statistics for 1969-1998).

The site was chosen to provide a rigorous test of the ability of the electromagnetic sensor to differentiate small but important soil differences within one soil type, influencing soil management decisions. The soils of the study area are mapped as 6 phases of the Kairanga silt loam (Shepherd 1992), classified as Gleyed Fluvial Recent soils (Hewitt 1998) or Aerie and Typic Endoaquepts (Soil Survey Staff 1998). Soil parent materials are recent fluvial deposits from the Oroua River, a very uniform silt loam in the top 20 cm, but with bands of fluvial deposits of varying coarseness at depth. A general trend of coarser parent materials closer to the Oroua River was noted' with the finest clayey phase furthest from the river.

Method of EM survey

An electromagnetic induction Geonics EM38[R] sensor was used with real-time-kinematic (RTK)-GPS and a Trimble Ag170 field computer to collect positional and E[C.sub.a] data. The computer and GPS equipment were mounted on an ATV bike, with the sensor towed behind on a rubber mat. One portion of the study area was surveyed 3 times (November 2001, February 2002, and September 2002) using the EM38 sensor in the vertical operating mode. Survey data points were collected at 1-s intervals, at an average ATV speed of 15 kph, with a measurement recorded approximately every 4 m along transects 10 m apart. During the third survey, which was of the whole study area, soil sampling and soil strength measurements were also completed.

E[C.sub.a] map production

Soil E[C.sub.a] survey points were filtered to remove: (i) RTK-GPS data classified as low accuracy, to ensure precise elevation information for topography/E[C.sub.a] comparison; and (ii) outlying E[C.sub.a] values. Determination of E[C.sub.a] outliers was based on the fit of data to a normal distribution curve. Points of interest were then examined to determine if they were genuine or a result of interference from fencelines, underground cables, or sensor anomalies. Filtered data comprising of latitude, longitude, height above mean sea level, and E[C.sub.a] were then imported into ArcMap[TM] (Environmental Systems Research Institute, ESRI[c] 1999) using a WGS84 projection, and subsequently converted to New Zealand Map Grid projection, using the (GD 1949) geodetic datum. Data were presented in a shapefile format. The points were kriged using Spatial Analyst (ESRI[c] 1999), using ordinary kriging and a spherical semivariogram model, to produce a map of soil E[C.sub.a].

Soil sampling

Forty-eight soil cores were taken using a vehicle-mounted Giddings hydraulic corer. Four sites within each of the 6 soil phases were sampled at the same time as the September 2002 EMI survey. Two cores were taken at each site, one for bulk density and volumetric soil moisture content analysis, and the other for soil chemical analysis. Mean coring depth was ~80 cm; below this depth the soil became saturated and could not be cored. The cores were extruded onto liners and then subsampled by horizon boundary and further subdivided into 20-cm lengths, where necessary. In addition, the EM sensor was hand-held at the soil surface at each soil sampling point, and used in both vertical and horizontal dipole positions. E[C.sub.a] measured in the vertical dipole mode (upright orientation) (referred to as E[C.sub.av]) provides an effective measurement depth of ~1.5 m. E[C.sub.a] measured in the horizontal dipole mode (sideways orientation) (referred to as E[C.sub.ah]) provides an effective measurement depth of ~0.75 m (Sudduth et al. 2001). The single value of E[C.sub.a] given by the EM38 at any one position is determined by the soil EC with depth, as weighted by the instrument response functions (McNeill 1992; Sudduth et al. 2001). E[C.sub.a] values are therefore depth-weighted averages. A previous 1.2-m soil core had been taken (Shepherd 1992) for soil particle-size analysis at each of these 24 sites, from which the per cent clay values (weighted for the soil profile) were derived.

Measurement of soil strength

Soil strength was assessed at each of the 24 soil sampling sites using an Eijkelkamp 06.15 penetrologger. The 2-[cm.sup.2] 30[degrees] coned penetrometer was used to take 10 replicate readings at 1-cm intervals to 80 cm depth. Readings were taken on the same day as the third EMI survey and soil samples in September 2002.

Additionally, when the E[C.sub.a] map was produced and viewed, localised zones of elevated E[C.sub.a] were clearly visible near fencelines and in gateways; these zones of elevated E[C.sub.a] did not correspond to the clay or soil map (see Fig 3), and were subsequently investigated. Four paired sites were selected: a site with an elevated E[C.sub.a] (designated 'a') was paired with a site within the same soil phase, land-use, and fertiliser regime, but with an E[C.sub.a] value within the normal range for the soft unit (designated 'b'). Soil strength readings were taken at these paired sites to investigate any relationship between farm management affect, compaction, and soil E[C.sub.a]. Soil strength measurements were taken at times when moisture was considered to be relatively uniform across the soilscape, and this was confirmed by assessing the soil moisture profiles at the same time.


Visual Soil Assessment and aggregate-size distribution

Soil physical condition was assessed in the field at the 4 selected paired sites as detailed above and annotated in Fig. 3, using the Visual Soil Assessment (VSA) method of Shepherd (2000). This was followed by a measurement of aggregate-size distribution (Shepherd et al. 2001). VSA, a field-based measurement, scores the condition of soil structure and consistence, porosity, colour, number and colour of mottles, earthworm numbers, tillage pan, degree of clod development and degree of soil erosion. VSA uses these soil properties to assess soil quality because they are easily seen and have a profound influence on soil biological and chemical properties, and hence farm productivity (Shepherd and Park 2003).

On this soil quality index, scores <10 indicate poor condition, a score of 10-25 indicates moderate condition, and >25 ranks a soil in good condition. VSA scores have been shown to be significantly related to corresponding laboratory-based measurements (Shepherd et al. 2002). The method involves a drop-shatter procedure using a clod 200 mm in diameter by 200 mm deep (Shepherd 2000; Shepherd et al. 2001). The resulting soil aggregates were then assessed and scored, before being sieved in the field through a nest of sieves (150 mm, 100 mm, 50 mm, 20 mm, 10 mm, 5 mm, 2 mm). Aggregates remaining on each sieve were weighed. The distribution of aggregate sizes was then expressed as a mean weight diameter (MWD), as described in Shepherd et al. (2001):

MWD = [summation of] (weight % sample on sieve x mean inter-sieve size/100)

This supplemented the VSA score with a quantitative value of mean aggregate size. All measurements were in triplicate. Increasing degradation of soil structure during compaction increases the overall size of soil aggregates as structural units are lost and the soil becomes more massive. Therefore, in an area where soil degradation has occurred due to, for example, stock treading or a tillage pan, it is expected MWD will increase and VSA score will decrease.

Soil laboratory analyses

One set of 24 soil cores was analysed for water content and bulk density, determined on soil cores of known diameter and length. The other set of 24 soil cores was analysed for total C, total N, pH, mineral N, Olsen P, cation exchange capacity, and exchangeable Ca, Mg, K, and Na.

One weighted mean value for each analysis was obtained for each core using the formula:

[x.sub.m] = (([x.sub.n] x [d.sub.n]) + ([x.sub.n+1] x [d.sub.n+1]) + ([x.sub.n+2] x [d.sub.n+2])....)/D

where [x.sub.m] is weighted mean value, [x.sub.n] is analysis value for subsample n of soil core, [d.sub.n] is thickness (cm) of subsample n of soil core, and D is total depth (cm) of soil core.

Soil chemical analyses were determined on air-dried and 2-mm-sieved soil samples; except for the mineral nitrogen determination, which was performed on 4-mm-sieved field-moist soil within 48 h of being collected. Total C and N were determined by a Leco CNS analyser. (Leco, St Joseph, MI).

Soil pH was measured on 8 g of soil mixed to a slurry with 20 mL of deionised water and left to stand overnight (Blakemore et al. 1987).

Mineral nitrogen (nitrate- and ammonium-nitrogen) was determined by extracting the moist soil in 2 M KCI for 60 min, in a 1:10 soil extraction ratio. The ammonium and nitrate were determined colourimetrically on a Lachat flow injection analyser (Zellweger Analytics Inc., Milwaukee, WI). The method was adapted from that described by Blakemore et al. (1987) and was used to assess plant-available nitrogen.

The method for assessing plant-available phosphorus was based on that of Olsen et al. (1954) as described by Blakemore et al. (1987). Soil was extracted with 0.5 M sodium bicarbonate at pH 8.5 at a 1:20 soil: extractant ratio with a 30-min shaking. Phosphate in the extracts was determined by a colourimetric method on a Lachat flow injection analyser (Zellweger Analytics Inc.).

The automated extractor procedure described by Blakemore et al. (1987) was used to measure cation exchange capacity (CEC). Exchangeable bases were removed and the exchange sites saturated with ammonium ions by leaching with neutral molar ammonium acetate. Excess ammonium acetate was washed from the sample with alcohol. The adsorbed ammonium ions were then displaced from the exchange sites by leaching the sample with molar sodium chloride.

Determination of ammonium in the sodium chloride solution to give the CEC was by a colourimetric method on a Lachat flow injection analyser (Zellweger Analytics Inc.).

Concentrations of exchangeable potassium, sodium, calcium, and magnesium were determined by atomic absorption spectrometry (, 2004).

Statistical analyses

Geostatistical analysis was conducted on EM38 sensor data and simultaneously acquired positional data to provide quantitative spatial and temporal analysis of E[C.sub.a]. Variograms were plotted for each of the three surveys, using GENSTAT 6.2 (Lawes Agricultural Trust[c] 2002).

Linear regression and multiple stepwise regression models were developed between E[C.sub.a] value and measured soil properties from each soil-sampling site.

Discriminant statistical analysis was used to investigate how well the E[C.sub.a] map predicted the 6 mapped soil phases within 1 soil type.

Results and discussion

E[C.sub.a] maps

Three E[C.sub.a] maps are shown in Fig. 2 for a portion of the study area, which was surveyed 3 times over the period of 1 year. These maps show the overall pattern of E[C.sub.a] remains similar, although the magnitude of E[C.sub.a] varied with time of survey. E[C.sub.a] values are highest for the third survey when soils were wettest and close to field capacity. At this time there was a regional soil moisture deficit no greater than 5-10 mm.


Figure 3 compares the E[C.sub.a] map for the whole study area with a map of per cent clay superimposed on soil units. This E[C.sub.a] map was produced from the September 2002 survey. The portion that was surveyed 3 times is shown in Fig. 2. Kriged boundary limits for this portion are different in Fig. 2. This is because kriging was standardised for the 3 survey times in Fig. 2 so that they can be easily compared. Figure 3 illustrates that increasing E[C.sub.a] values are positively related to increasing per cent clay in the profile.

These contour plots (Figs 2 and 3) show changes in E[C.sub.a] over the whole area. The pattern in the more random changes over shorter distances (i.e. roughness) is shown by the variograms (Fig. 4). For any given distance (the lag on the x-axis) they plot half the variance of the differences between E[C.sub.a] values at all pairs of sampling points that distance apart. At a short lag distance E[C.sub.a] levels should be similar, so their differences will be small with a small variance. As the lag distance increases E[C.sub.a] levels should become increasingly independent and half the variance of the differences should approach the overall variance. At this distance, the 'sill', variograms level off. A maximum lag distance of 100 m was used in this study because lengths beyond this are not replicated sufficiently to provide a random sample for estimating a variance. Duplicate point values, resulting from overlapping runs of the sensor, were eliminated from the analysis so that they did not bias results.

Variogram A plots variance of E[C.sub.a] in an east-west direction (usings northings) at each survey time. Variogram B plots variance of E[C.sub.a] in a north-south direction (using castings) at each survey time. Comparison of the 3 variograms in each plot shows that there is no levelling off (sill). They are therefore affected by factors on a scale >100 m, for example, the trends as shown by Figs 2 and 3. The intercept on the y-axis, the 'nugget', is generally around 2. This is the variance of repeated measurements at the same point, and is indicative of repeatability of EM38 sensor readings. Irregular patterns at small lags result from readings taken down fencelines where steep E[C.sub.a] gradients occur. The slightly different slopes of the variograms, greatest in September and least in February, are a consequence of differences in the overall variability, which in turn reflects soil moisture, wettest in September and driest in February. Overall, the variogram shapes are very similar, suggesting that the EM38 is measuring a factor that is spatially and temporally stable. Other reported results show this to be soil clay content and properties determined by this clay component.

Relationships of soil E[C.sub.a] with measured soil properties

During the final September 2002 survey, soil E[C.sub.a] was measured in the vertical mode (E[C.sub.av]) and horizontal mode (E[C.sub.ah]) at each of the 24 soil sampling sites. These 2 modes of measurement were highly linearly related (r = 0.96). E[C.sub.av] and E[C.sub.ah] were then correlated with weighted mean values for each soil property. Results are summarised in Table 1 and Fig. 5. Weighted mean values were determined for the whole profile (~80 cm for soil samples, 120 cm for per cent clay samples) and to 50 cm. The Olsen P values are for the top 10cm. According to the literature (e.g. McNeill 1992; Sudduth et al. 2001), the weighted mean value for the top ~75 cm of the soil profile should correlate closely with E[C.sub.ah], while the weighted mean value to ~1.5 m depth should correlate more closely with E[C.sub.av]: However, per cent clay, weighted for the whole profile (1.2 m), correlated better with E[C.sub.ah] ([R.sup.2] = 0.65) than the mean value to 75 cm ([R.sup.2] = 0.38). Table 1 shows that the best correlations were with per cent clay (to 1.2 m) ([R.sup.2] = 0.72) and exchangeable magnesium (to 50 cm) ([R.sup.2] = 0.76). Olsen P (to 10 cm) also correlated moderately well ([R.sup.2] = 0.62).


Many workers have reported strong relationships between soil moisture and E[C.sub.a] (e.g. Kachanowski et al. 1988; Waine et al. 2000). Kachanowski et al. (1988) found that soil E[C.sub.a] explained 96% of the spatial variation of soil water content, and Waine et al. successfully used EMI to produce maps of available water content. However both studies used a wide range of soil texture and moisture regimes. In comparison, the present study uses soils of one soil type, the Kairanga silt loam, with a comparatively narrow textural range (uniform silt loam topsoils over loamy subsoils) and resulting moisture-holding capacity. E[C.sub.a] was therefore not such a good predictor of soil moisture in this study, predicting about 42% of its variability.

The exchangeable cations [Mg.sup.2+], [Ca.sup.2+], [K.sup.+], and [Na.sup.+] are the forms of these elements considered plant-available by exchange with hydrogen ions from the exudates of root hairs, and are therefore important contributors to plant growth. Relationships between each exchangeable cation and both E[C.sub.av] and E[C.sub.ah] are reported in Table 1. Higher [R.sup.2] values show which cations dominated on charged clay surfaces and in the soil solution where E[C.sub.a] values are largely controlled. Using linear regression, exchangeable [Mg.sup.2+] accounts for more of the variation in E[C.sub.a] than any other exchangeable cation or any other single soil property. It was noted that Mg values, which were moderate to high (>1.0 cmol(+)/kg) (Fig. 5), tended to increase down the profile, suggesting Mg originates in the parent material, rather than as a fertiliser addition to the soil surface. It is therefore likely that its origin is chlorite clay minerals weathering to illite. Chlorite and illite are the dominant clay minerals in this soil series (Shepherd 1992), being most dominant at 18-28 cm depth. This may explain why the weighted mean value to 50 cm was predicted more strongly than for the whole profile. When chlorite weathers to illite, it not only provides a clay surface with increasing charge, but it also releases [Mg.sup.2+] ions to the soil solution. It is the increased charge on the soil surfaces as well as the cations in soil solution that will increase measured E[C.sub.a].

Cation exchange capacity, and the proportion of it occupied by exchangeable bases, is a useful indicator of fertility. Regression of E[C.sub.a] against CEC explained ~50% of the variability of CEC (Table 1 and Fig. 5). The CEC of these soils is dominated by [Ca.sup.2+], which contributes 5-90% of the exchange surfaces in all soils sampled. The calcium is added to the soil in fertiliser and dung, as well as being contributed by clay minerals. The CEC of these soils relates less well to E[C.sub.a] values than exchangeable [Mg.sup.2+] does. This shows that E[C.sub.a] best reflects [Mg.sup.2+], which is the exchangeable ion that is actively weathering from clay minerals with accompanying increasing charged surfaces as well as ions released to the soil solution.

Regression of E[C.sub.a] against Olsen E an indicator of plant-available P, explained ~60% of the variability in Olsen P (Table 1 and Fig. 5). This suggests E[C.sub.a] readings by the EM38 sensor are significantly affected by adsorbed phosphate ions in the top 10 cm of the soil profile, but probably also reflect the concentration of cations accompanying P fertiliser and dung.

In this study E[C.sub.a] did not relate well to pH, total C, total N, and extractable N (Table 1). Total C and N levels were consistently low over the whole study area (mean value for the 24 sites: total C, 1.11% [+ or -] 0.21; total N, 0.12% [+ or -] 0.02). The pH range was very small (6.54 [+ or -] 0.21). Extractable N values, a measure of ammonium and nitrate ions in soil solution, were highly variable. The extractable N being controlled not only by soil organic matter and N fertiliser applied, but also by urination from grazing animals (mean value for the 24 sites: extractable N 9.56 [+ or -] 16.26 mg/kg). The results therefore suggest that, in this study, E[C.sub.a] best predicts soil properties that relate to charged soil clay surfaces rather than ionic strength of the soil solution. This is supported by the prediction of Olsen P; the measurement providing an indication not only of phosphate in solution but also of phosphate weakly adsorbed onto soil particle surfaces.

It is important to understand how EC is related to combinations of soil characteristics (Brevik and Fenton 2002). Therefore linear models were developed using step-wise regression analysis. As previously discussed E[C.sub.av] correlated best with per cent clay ([R.sup.2] = 0.72). When exchangeable magnesium is added to the regression the [R.sup.2] is improved to 0.77 and regression with each parameter is significant ([P.sub.clay] = 0.008; [P.sub.exch Mg] = 0.026). The addition of Olsen P to the regression, further improves the [R.sup.2] to 0.83, with regression of each parameter being significant ([P.sub.clay] = 0.043; [P.sub.exch Mg] = 0.038; [P.sub.Olsen P] = 0.017). The resulting equation is:

E[C.sub.av] = 0.252a + 1.808b + 0.209c + 22.705 ([R.sup.2] = 0.83)

where a is per cent clay (weighted for the whole profile), b is exchangeable Mg (weighted for the whole profile), and c is Olsen P (in top 10 cm of soil).

The same model was applied for E[C.sub.ah]. Regression of E[C.sub.ah] against per cent clay and exchangeable magnesium (to 50 cm) has an [R.sup.2] of 0.81, with the relationship between clay and exchangeable magnesium being significant and highly significant respectively ([P.sub.clay] = 0.018; [P.sub.exch Mg to 50 cm] = 0.000). The addition of Olsen P improves the [R.sup.2] to 0.84, with levels of significance being [P.sub.clay] = 0.058; [P.sub.exch Mg to 50 cm] = 0.003; [P.sub.Olsen P] = 0.066. The resulting equation is:

E[C.sub.ah] = 0.127a + 2.579b + 0.110c + 22.120 ([R.sup.2] = 0.84)

where a is per cent clay (weighted for the whole profile), b is exchangeable Mg (weighted for the top 50 cm), and c is Olsen P (in top 10 cm of soil).

These equations indicate that E[C.sub.a] values reflect a component of each of these soil characteristics in this study area.

Relationships of soil E[C.sub.a] and soil units

Canonical discriminant analysis was used to illustrate the relationship of per cent clay and E[C.sub.a], using soil unit as a grouping variable, for the 24 sample sites. The 2 linear functions, for discriminating between soil units, of per cent clay and E[C.sub.a] were plotted against each other in the canonical scores plot (Fig. 6). Factor (1) is the function providing best discrimination of soil units by E[C.sub.a] and clay. Factor (2) provides the small remaining discrimination. Figure 6 illustrates that soil phases fall into 2 distinct groups: soil phases K1, K2, and K3 fall into one group; and soil phases K4, K5, and K6 into a second group.


In addition, when only E[C.sub.a] is used to predict soil unit, discriminant analysis shows the data will predict these 2 groups of soils with 75% accuracy (K1-K3) and 100% accuracy (K4-K6) (Jack-knifed classification matrix). If, however, the 2 major soil groupings (A = K1 and K2; B = K3-K6) are used then E[C.sub.a] predicts both of them with 100% accuracy. In this study, therefore, E[C.sub.a] can be used to predict 2 major groupings of the soil phases within 1 soil type. This information provides a useful management tool because Group A [K1 (clayey phase) and K2 (clayey over loamy phase)] is characterised by an impeded clayey horizon at depth that impedes water flow and reduces root growth (Shepherd 1992), whereas Group B (K3-K6) does not have this limitation to use. E[C.sub.a] was able to predict the presence of this clayey horizon but could not predict its actual depth, which, at sites sampled, varied between 30 and 54 cm. Doolittle et al. (1994) reported a relationship between E[C.sub.a] and depth to clay pan, but it is suggested that in their study, the compacted layer, a clay pan developed in glacial till and dominated by smectitic clays, was significantly more developed than the compacted clayey horizon of our study area.

This study shows the ability of E[C.sub.a] to predict within-paddock changes in soil properties, implying that it is a powerful tool for adding detail to existing soil maps with a precision consistent with the agricultural management of the land.

Relationships of E[C.sub.a] and soil management zones

Soil E[C.sub.a] and soil strength

Soil strength results showed no obvious trend with soil unit or E[C.sub.a] value (Table 2). Soil strength values for the 0-10 cm depth of each soil unit were not significantly different, except for K6, the coarsest soil textural phase, which had a significantly lower soil strength of 1.55 MPa. Small differences were found between mean soil strength values for the top 80 cm of each soil unit. However, differences in mean soil strength could not be related to E[C.sub.a]. Mean values greater than 2 MPa (Sudduth et al. 2000), which occur in all soils except K6 (0-10 cm), indicate that these soils do inhibit root growth to some extent.

Table 3 shows 2 localised zones with elevated E E[C.sub.a] values (Site 3a and 4a in Fig. 3) near gateways and along fencelines in pastoral sites had greater soil strength values, weighted for the top 50 cm, than a paired site within the same soil unit. This indicates soil compaction has occurred at these sites of elevated E[C.sub.a] values, most likely due to stock treading. At Site 2a (Table 3 and Fig. 3), soil strength values tended to be higher, but not significantly higher, where elevated E[C.sub.a] values were noted. At Site 1, the cropped site, no increase in soil strength was noted at the site of elevated E[C.sub.a].

Soil E[C.sub.a], Visual Soil Assessment, and aggregate size distribution

Localised zones of elevated E[C.sub.a] within a soil unit were investigated using paired site methodology as discussed above. These zones typically occurred along fencelines and in gateways. Table 3 indicates these 'hot spots' of elevated E[C.sub.a] (see Fig. 3) can be related to differences in soil physical condition, as shown by highly significant differences between VSA scores for all 4 paired sites. Elevated E[C.sub.a] therefore occurs where soil structural units are larger, indicated by a lower VSA score and a larger aggregate mean weight diameter. This is exemplified in Fig. 7, which shows the VSA assessment of soil at paired sites 4a and 4b. Site 4a, which occurs close to a gateway, has an elevated E[C.sub.a] value of 44 mS/m, compared with the paired site that has an E[C.sub.a] value of 31 mS/m. Its aggregate mean weight diameter is 72 mm, compared with 33 m at the paired site, and its VSA score is 8.2 compared with 17.5 for the paired site. The changes in aggregate size and VSA scores show that the site with elevated E[C.sub.a] has a degraded soil physical condition, which will result in reduced air and water movement through soils and will impede root growth, thus reducing pasture production and crop yield. It is proposed that the elevated E[C.sub.a] is related to improved conducting pathways between soil particles due to compaction resulting from management practices such as cultivation and stock treading.



Electromagnetic induction surveys of soilscapes provide a rapid, affordable approach to assessing soil variability. Zones on the EMI map primarily delineate textural differences in this non-saline situation, but can also be related to certain soil fertility indicators such as exchangeable Mg, Olsen P, and CEC, E[C.sub.a] values, determined using the EM38 sensor both in the vertical and horizontal mode, gave good correlations with per cent clay for the whole soil profile ([R.sup.2] = 0.72 (E[C.sub.av]) and 0.65 (E[C.sub.ah])), as well as Olsen P ([R.sup.2] = 0.62 (E[C.sub.av]) and 0.61 (E[C.sub.ah])), and exchangeable Mg ([R.sup.2] = 0.71 (E[C.sub.av]) and 0.76 (E[C.sub.ah])). Exchangeable [Mg.sup.2+] for these soils correlated better than the other exchangeable cations ([Ca.sup.2+], [K.sup.+], [Na.sup.+]), and this is thought to reflect the clay mineralogy, dominated by chlorites weathering to illites, releasing [Mg.sup.2+] into the soil solution and increasing surface charge on the clay surfaces.

Compaction and reduced soil physical condition in localised areas such as gateways and along fencelines are associated with elevated E[C.sub.a] values.

The E[C.sub.a] map has been used to predict accurately two major groupings of the soil phases (K1 and K2; K3-K6) within the 1 soil type, with management implications relating to the clayey horizon occurring in soil phases K1 and K2, which reduces water flow and impedes root growth in comparison to the other mapped soil phases. EMI technology is able to detect subtle changes in soil characteristics across a soilscape, and can be used to predict textural groupings of soils, as well as to identify those areas of compaction within a paddock that require corrective management. It is indeed a useful tool in site-specific management.
Table 1. Relationships of measured soil properties with E[C.sub.a]
values obtained in the September 2002 survey, expressed as [R.sup.2]
(the proportion of variance accounted for by linear regression)

CEC, Cation exchange capacity. E[C.sub.av] and E[C.sub.ah] are
apparent EC measured in the vertical and horizontal dipole mode,

Strongest relationships for each soil property are shown in bold.
[R.sup.2] > 0.29 are significant, n = 22, P = 0.01

 Whole profile

Soil property E[C.sub.av] E[C.sub.ah]

Per cent clay 0.72 0.65
Vol. water content 0.42 0.42
CEC 0.49 0.48
Exchang. Ca 0.28 0.33
Exchang. Mg 0.68 0.67
Exchang. K 0.15 0.12
Exchang. Na 0.10 0.06
pH (water) 0.01 0.01
Total C 0.01 0.00
Total N 0.04 0.03
Extract. N (2 M KC1)
Olsen P (for top 10 cm soil)

 To 50 cm

Soil property E[C.sub.av] E[C.sub.ah]

Per cent clay 0.51 0.46
Vol. water content 0.13 0.14
CEC 0.53 0.59
Exchang. Ca 0.07 0.14
Exchang. Mg 0.71 0.76
Exchang. K 0.25 0.25
Exchang. Na 0.23 0.28
pH (water) 0.12 0.11
Total C 0.02 0.05
Total N 0.00 0.01
Extract. N (2 M KC1) 0.02 0.00
Olsen P (for top 10 cm soil) 0.62 0.61

Table 2. Mean soil strength (MPa) and E[C.sub.a] values (September
2002 survey) for each phase of Kairanga silt loam

Soil unit Soil E[C.sub.a] Soil strength
 map (vertical) (MPa)
 symbol (0-80 cm)

Clayey phase K1 43.4 2.33
Clayey over loamy phase K2 42.0 2.58
Fine silty phase K3 34.2 2.36
Fine silty over sandy phase K4 32.0 3.07
Fine loamy phase K5 34.9 2.45
Coarse loamy phase K6 34.0 2.52

Soil unit s.e. Soil strength s.e.
 (0-10 cm)

Clayey phase 0.07 2.30 0.12
Clayey over loamy phase 0.13 2.48 0.24
Fine silty phase 0.14 2.18 0.11
Fine silty over sandy phase 0.11 2.39 0.33
Fine loamy phase 0.18 2.26 0.32
Coarse loamy phase 0.13 1.55 0.17

Table 3. A comparison of paired sites to investigate management
effects on E[C.sub.a] values

Site 'a' is site with elevated E[C.sub.a] and likely management
effect; site 'b' is paired site in same soil phase. VSA, Visual
Soil Assessment

Site Manag. Soil Soil Present Soil depth
 effect map phase land use assessed
 symbol (cm)

1a Fenceline K2 Clayey over Cropping 10-20
1b K2 Clayey over Cropping 10-20
2a Fenceline K3 Fine silty Pasture 0-20
2b K3 Fine silty Pasture 0-20
3a Fenceline K3 Fine silty Pasture 0-20
3b K3 Fine silty Pasture 0-20
4a Near K3 Fine silty Pasture 0-20
4b gateway K3 Fine silty Pasture 0-20

Site E[C.sub.a] VSA P No. of P
 (mS/m) score worms/
 [m.sup.2] (A)

1a 48.5 12.5 0.0003 8 0.2
1b 37.5 16 25
2a 48.5 13.5 0.002 475 0.2
2b 37.5 20 667
3a 44 13 0.0008 333 0.02
3b 31 17.5 1083
4a 44 8.2 0.00002 108 0.0007
4b 31 17.5 1083

Site Agg. size P Soil P
 (mean wt diam.) strength (B)
 (mm) (MPa)

1a 53 0.1 2.12 1
1b 33 2.12
2a 49 0.06 2.89 0.3
2b 25 2.75
3a 58 0.03 3.49 <0.00001
3b 33 2.63
4a 72 0.04 2.96 0.03
4b 33 2.63

(A) 5-min earthworm count of soil aggregates obtained from a
200-mm block of soil.

(B) 0-50 cm soil depth.

Fig. 4. Variograms to compare correlation of E[C.sub.a] with
distance at each survey in an east-west direction (Plot A) and
a north-south direction (Plot B), with variogram parameter

Variogram details

Time of survey November February September
 2001 2002 2002

General mean 27.99 29.98 37.26
General variance 6.32 4.85 10.27
No. of observations 1524 990 713
Maximum lag 100 100 100
Step length 5 5 5


The authors would like to thank AgResearch, owners of the experimental site and Steve Lees, the farm manager for providing farm records. They also acknowledge the highly skilled contributions of John Dando, Rob Murray, and Warren Woodgyer in the field; and Brian Daly and Keitha Giddens in the laboratory. Finally we acknowledge the NZ Foundation of Science, Research and Technology who funded this research.


Anderson-Cook CM, Alley MM, Roygard JKF, Khosla RB, Noble RB, Doolittle JA (2002) Differentiating soil types using electromagnetic conductivity and crop yield maps. Soil Science Society of America Journal 66, 1562-1570.

Beecher HG, Dunn BW (2002) Using electromagnetic induction technology to determine spatial variability and sampling sites for soil acidity assessment. In 'Proceedings of Australian Society of Soil Science Conference'. Perth, WA, Abstract, pp. 26-27. (ASSSI: Perth, W. Aust.)

Blakemore LC, Searle PL, Daly BK (1987) Methods for chemical analysis of soils. New Zealand Soil Bureau Scientific Report 80, Wellington.

Bramley RGV (2001) Progress in the development of precision viticulture--variation in yield, quality and soil properties in contrasting Australian vineyards. In 'Precision tools for improving land management'. Occasional Report No. 14. (Eds LD Currie, P Loganathan) pp. 25-43. (Fertiliser and Lime Research Centre, Massey University: Palmerston North, NZ)

Bramley RGV, Proffitt APB (1999) Managing variability in viticultural production. Grapegrower and Winemaker 427, 11-16.

Brevik EC, Fenton TE (2002) Influence of soil water content, clay, temperature, and carbonate minerals on electrical conductivity readings taken with an EM-38. Soil Survey Horizons Spring 2002, 9-13.

Cook SE, Bramley RGV (1998) Precision Agriculture--opportunities, benefits and pitfalls of site-specific crop management in Australia. Australian Journal of Experimental Agriculture 38, 753-763.

Cowie JD (1978) Soils and agriculture of Kairanga County, North Island, New Zealand. New Zealand Soil Bureau Bulletin No. 33.

Doolittle JA, Sudduth KA, Kitchen NR, Indorante SJ (1994) Estimating depths to claypans using electromagnetic induction methods. Journal of Soil and Water Conservation 49, 572-575.

Eigenberg RA, Doran JW, Nienaber JA, Ferguson RB, Woodbury BL (2002) Electrical conductivity monitoring of soil condition and available N with animal manure and a cover crop. Agriculture Ecosystems and Environment 88, 183 193. doi:10.1016/S01678809(01)00256-0

Godwin RJ (2001) Field condition mapping technologies. In 'Smart Farming II: Workshop on Automation for Agriculture'. University of Putrajaya, Malaysia. (Ed. MSM Amin) p.20. (MSAE/UPM: Serdang Selangor, Malaysia)

Hewitt AE (1998) 'New Zealand Soil Classification. Landcare Research Science Series No.1.' 2nd edn (Manaaki Whenau Press: Lincoln, New Zealand)

Johnson CK, Mortensen DA, Wienhold BJ, Shanahan JF, Doran JW (2003) Site-specific management zones based on soil EC in a semiarid cropping system. Agronomy Journal 95, 303-315.

Kachanowski RG, Gregorich EG, van Wesenbeeck IJ (1988) Estimating spatial variations of soil water content using non-contacting electromagnetic inductive methods. Canadian Journal of Soil Science 68, 715-722.

Lund ED, Colin PE, Christy D, Drummond PE (1999) Applying soil electrical conductivity technology to precision agriculture. In 'Proceedings of 4th International Conference on Precision Agriculture'. St Paul, Minnesota, 19-22 July 1998. (Eds PC Robert, RH Rust) pp. 1089-1100. (Soil Science Society of America: Madison, WI)

McBride RA, Gordon AM, Shrive SC (1990) Estimating forest soil quality from terrain measurements of apparent electrical conductivity. Soil Science Society of America Journal 54, 290-293.

McNeill JD (1992) Rapid, accurate mapping of soil salinity by electromagnetic ground conductivity meters. In 'Advances in measurement of soil physical properties: bringing theory into practice'. SSSA Special Publication No. 30. pp. 209-229. (Soil Science Society of America: Madison, WI)

Olsen SR, Cole CV, Watanabe FS, Dean LA (1954) Estimation of available phosphorus in soils by extraction with sodium bicarbonate. USDA Department Circular No. 939.

Plant RE (2001) Site-specific management: the application of information technology to crop production. Computers and Electronics in Agriculture 30, 9-29. doi:10.1016/S01681699(00)00152-6

Rhoades JD, Corwin DL (1981) Determining soil electrical conductivity-depth relations using an inductive electromagnetic soil conductivity meter. Soil Science Society of America Journal 45, 255-260.

Shepherd TG (1992) Soils of the pasture-cropping systems unit, Aorangi Research Station, Manawatu District. DSIR Land Resources Scientific Report No. 30, New Zealand.

Shepherd TG (2000) 'Visual soil assessment. Vol 1. Field guide for cropping and pastoral grazing on flat to rolling country.' p. 84. (Horizons. mw & Landcare Research: Palmerston North, New Zealand)

Shepherd TG, Saggar S, Newman RH, Ross CW, Dando JL (2001) Tillage-induced changes to soil structure and organic carbon fractions in New Zealand soils. Australian Journal of Soil Research 39, 465-489. doi: 10.1071/SR00018

Shepherd TG, Sparling GP, Todd MD (2002) Visual soil assessment: can we see what we measure? In 'Proceedings of Soil Quality and Sustainable Land Management Conference'. (Eds P Stephens, J Callaghan, A Austin) pp. 117-122. (Landcare Research: Palmerston North, NZ)

Shepherd TG, Park SC (2003) Visual soil assessment: a management tool for dairy farmers. In 'Proceedings of the 1st Dairy Conference'. Dexcel's Ruakura Dairyfarmers Conference. (Ed. IM Brookes) pp. 111-123. Continuing Massey University Dairyfarming Annual (Vol 55).

Smith JL, Doran JW (1996) Measurement and use of pH and electrical conductivity for soil quality analysis. In 'Methods for assessing soil quality'. SSSA Special Publication 49, pp. 169-185. (Soil Science Society of America: Madison, WI)

Soil Survey Staff (1998) 'Soil Taxonomy--A basic system of soil classification for making and interpreting soil surveys.' 2nd edn. USDA Agriculture Handbook No. 436. (USDA)

Sudduth KA, Drummond ST, Kitchen NR (2001) Accuracy issues in electromagnetic induction sensing of soil electrical conductivity for precision agriculture. Computers and Electronics in Agriculture 31, 239-264. doi: 10.1016/S0168-1699(00)00185-X

Sudduth KA, Hummel JW, Kitchen NR, Drummond ST (2000) Evaluation of a soil conductivity sensing penetrometer. In '2000 ASAE Annual International Meeting'. Midwest Express Center, Milwaukee, WI, paper no. 001043.

Triantafilis J, Huckel AI, Odeh IOA (2001) Comparison of statistical prediction methods for estimating field-scale clay content using different combinations of ancillary variables. Soil Science 116, 415-427. doi: 10.1097/00010694-200106000-00007

Triantafilis J, Ahmed MF, Odeh IOA (2002) Application of a mobile electromagnetic sensing system (MESS) to assess cause and management of soil salinisation in an irrigated cotton-growing field. Soil Use and Management 18, 330-339. doi: 10.1079/SUM2002139

Waine TW, Blackmore BS, Godwin RJ (2000) Mapping available water content and estimating soil textural class using electro-magnetic induction. In 'EurAgEng 2000'. Warwick, UK, paper no. 00-SW-044.

Manuscript received 17 October 2003, accepted 19 April 2004

C. B. Hedley (A), I. J. Yule (B), C. R. Eastwood (B), T. G. Shepherd (A), and G. Arnold (A)

(A) Landcare Research, Private Bag 11052, Palmerston North, New Zealand.

(B) New Zealand Centre for Precision Agriculture, Massey University, Palmerston North, New Zealand.
COPYRIGHT 2004 CSIRO Publishing
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2004 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Hedley, C.B.; Yule, I.J.; Eastwood, C.R.; Shepherd, T.G.; Arnold, G.
Publication:Australian Journal of Soil Research
Geographic Code:8NEWZ
Date:Jul 1, 2004
Previous Article:Development of an equation to relate electrical conductivity to soil and water salinity in a Mediterranean agricultural environment.
Next Article:Micromorphology and electron microprobe analysis of phosphorus and potassium forms of an Indian Black Earth (IBE) anthrosol from Western Amazonia.

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters |