Table Of ContentsNext Page

Matching veris data to soil properties on vertisols in northern Australia

R. Kelly1, T. Jensen1, J. Cooper1, D. Butler1, W. Strong1 and B. Town2

1Farming Systems Institute, QDPI, Toowoomba
2
Wesfarmers Landmark, Dalby
Phone 07 4688 1524; Fax 07 4688 1192
Email: kellyrm@dpi.qld.gov.au

Abstract

The Veris 3100 cart, an on-the-go sensor that records soil electrical conductivity (ECV), has been successfully used to identify spatial patterns in soil within fields. We used the cart to identify spatial variation in soil EC and thus, on a full moisture profile, use this as a surrogate indicator of soil water supply. This covariable could then assist with the interpretation of coincidental yield and protein maps. We mapped ECV in 3 fields near Dalby characterised by deep (>150 cm), cracking, heavy-clay soils. Ten soil cores to a depth of 1.2 m were sampled from each field, using directional soil sampling. and analysed for a range of soil properties; these included pH, EC, Cl, moisture content, N, S, P, total N, and OC. Early indications suggest that few robust relationships were found for the 3 fields. Data from the sensor were related most closely to pH and Cl. While further statistical analyses may identify other relationships, the poor discrimination could be related to the low variation in ECV present in this soil type, or inadequate sampling. Further work on soil type and what soil properties are actually being measured by the Veris is needed.

Introduction

Mapping variation in soil properties is one strategy to understand which factor, or factors, contributed to site-specific variation in grain yield. Soil mapping can be conducted via remotely-carried sensors, such as airborne or satellite-derived imagery, or via direct recording sensors that capture variation while moving across the field. The Veris 3100 cart (http://www.veristech.com/) is a direct sensor capable of measuring variation in soil electrical conductivity (EC) to 30 and 90-cm depth. The device is made up of pairs of coulter electrodes that penetrate the soil surface. One coulter pair directs an electrical current into the soil, and the other pairs measure the voltage drop. The readings, then, are essentially an average of the EC found throughout that profile depth (ACPA 2001).

In addition, the EC data has been successfully used to highlight potential yield variation under certain weather patterns. Researchers at ACPA have used the EC data as a surrogate, when the soil profile is full, to identify soil textural variation within grain fields which, under water-limiting conditions, closely align with resultant yield variation (ACPA 2001).

Our objective was to identify variation in EC for a number of deep, self-mulching Vertisolic fields on the Jimbour floodplain, near Dalby, and to see if these values correlate with selected soil properties and with resultant yield variation.

Materials and methods

Locations and site selection

Three fields on the Jimbour floodplain, near Dalby (27.18°S, 151.26°E), were selected for evaluation of the Veris 3100. Each field had been summer-cropped with sorghum or corn, and was under fallow at the time of measurement. The fields have been managed under minimum-tillage with traffic set on 3-m widths to provide a controlled traffic system. Soil type is a black, self-mulching cracking clay classified as a Norillee Vertisol and characterised by a deep profile with uniform clay texture throughout the profile (Harris 1998).

Measurement of soil EC

In September 2000, the Veris 3100, kindly provided by APCA, was towed behind a tractor across the fields in a skip-row configuration with 18-m between runs. The coulters were lowered to penetrate 3-5 cm depth. Positions were provided by a differentially-corrected global positioning system accurate to 1-m. Electrical conductivity was then determined every second for 0-30 cm and 0-90 cm depth, as indicated in Figure 1.

Figure 1. Sample locations in 3 fields on the Jimbour floodplain, Dalby. Fields are drawn to scale, but have been re-drawn side-by-side.

Validation of Veris data to soil properties

After filtering out nonsensical data, Veris maps were interpolated, using the localised kriging routine in Vesper (Minasny et al. 1999), to provide EC surfaces for each field. The maps were zoned, and 10 sample sites per field were chosen that reflected the variation in EC. The locations were loaded into an AgLeader box to locate each site, and soil cores to 1.5-m were collected at each site. Surface soil was collected as a composite sample by bulking 5 step-cores to 10-cm. Each sample was labelled to depth, site, and field.

Profile soil (0-120 cm) was analysed for pH/EC, chloride (Cl), nitrate-N (N), and sulfate-S (S) for each 30-cm increment. Surface soil (0-10 cm) was analysed for pH/EC, S, total N (TN), phosphorus (P), and organic carbon (OC). Percent clay content of the surface soil of 6 cores taken from field A was also determined. Gravimetric moisture contents (θg), after drying, were calculated for all depths. Analytical methods used by the Leslie Research Centre’s Analytical Section for analyses were obtained from Rayment and Higginson (1992) (Table 1).

Table 1. Methods for analysis of soil samples from 3 fields near Dalby. Methods refer to sections in Rayment and Higginson (1992).

Soil property

Extractant

Method

Profile samples (0-120 cm)

pH/EC/Cl

1:5 soil/water extract

3A1

N

2 M KCl

7C1

S

0.01 M Ca(H2PO4)2

10B1

Surface samples (0-10 cm)

OC

Walkley-Black procedure

6A1

P

0.5 M NaHCO3

9B1

TN

Kjeldahl steam distillation

7A1

Statistical analysis

Soil analytical data were compared with the Veris data to highlight dependencies using a simple correlation model and applying a linear regression to the data. Dependencies were validated by matching Veris data (ECV), averaged to 20-m around the sample site in the direction of travel, for the equivalent depth to each soil property.

Results and discussion

Variation in ECV

Variation in ECV was apparent in all fields to differing extents (Figs. 2, 3, 4). The greatest variation in ECV was found in Field 3 for the 0–90-cm range, while the least variation was found in Field 1 at the 0–30-cm range (59 v. 30 dS/m). For all fields, the range in ECV within the field to 30 and 90-cm never exceeded 40 and 60 dS/m, respectively, while standard deviation was rarely > 15 dS/m (Table 2).

The marked difference within Field 1, which had just been sown with spring mungbeans, was related to previous crop rotation. The eastern third was sown after a long-fallow while the remainder was sown immediately following a crop (Figure 2).

Figure 2. Veris maps interpolated from ECV to 30-cm (a) and 90-cm (b) from Field 1 near Dalby.

Figure 3. Veris maps interpolated from ECV to 30-cm (a) and 90-cm (b) from Field 2 near Dalby.

Figure 4. Veris maps interpolated from ECV to 30-cm (a) and 90-cm (b) from Field 3 near Dalby.

Table 2. Non-spatial variation in ECV (dS/m) to 30 or 90-cm for 3 fields near Dalby. Data had been filtered prior to this collation.

Fields

10% quartile

90% quartile

Mean

Standard deviation

Veris 30

Field 1

49

79

64

14

Field 2

96

128

114

14

Field 3

106

143

125

15

Veris 90

Field 1

67

104

86

17

Field 2

109

152

131

18

Field 3

106

165

137

23

Statistical analysis of ECV

Based on surface analysis, ECV data to 30-cm was positively correlated to all soil properties except for pH, and was most closely correlated with S (r2 = 0.42***) and OC (r2 = 0.19**) (Tables 3, 4). In contrast, ECV data to 90-cm was most closely correlated with OC and P, although the linear regression models were poor (r2 = 0.09*; r2 = 0, respectively). Sulfate levels in the surface soil were able to explain 42 and 22% (P<0.001) of the variation in the 30 and 90-cm ECV data, respectively. The higher correlation between ECV at 30 and 90-cm and clay content (0.44 v. 0.46) was largely due to the smaller number of analyses used to derive the relationship (n = 16).

Table 3. Correlations between ECV to 30 or 90-cm and surface (0–10-cm) and profile (0–30 or 0–90-cm) soil properties determined from directed soil sampling in 3 fields near Dalby.

Surface soil properties

 

Profile soil properties

Soil properties

Units

Veris 30

Veris 90

 

Soil properties

Units

Veris 30

Veris 90

pH

 

-0.56

-0.30

 

pH

 

-0.60

-0.34

Clay

%

0.44

0.46

 

Cl

mg/kg

-0.54

-0.20

EC

dS/m

0.58

0.08

 

EC

dS/m

-0.34

-0.38

OC

%

0.85

0.74

 

N

mg/kg

0.22

0.13

P

mg/kg

0.68

0.56

 

N

kg N/ha

0.22

0.12

TN

mg/kg

0.57

0.18

 

S

mg/kg

0.09

0.03

N

mg/kg

0.62

0.21

 

θg

%

0.42

0.42

S

mg/kg

0.86

0.39

         

Table 4. Percent variance accounted for by linear regressions between ECV to 30 or 90-cm and soil properties of surface soil (0–10-cm) determined from directed soil sampling in 3 fields near Dalby.

Soil properties

Veris 30

Veris 90

Variance

F-ratioa

Variance

F-ratio

pH

30

***

14

**

Clay

16

<0.10

19

*

EC

0.4

n.s.

1

n.s.

OC

19

**

9

*

P

0

n.s.

0

n.s.

TN

10

*

8

*

N

4

n.s.

3

n.s.

S

42

***

22

***

aF-ratio is significant at 5, 1 or 0.1% (*, **, *** respectively); n.s. indicates the F-ratio is not significant.

Table 5. Percent variance accounted for by linear regressions between ECV to 30 or 90-cm and soil properties of profile samples (0–30, 0–90-cm) determined from directed soil sampling in 3 fields near Dalby.

Soil properties

Veris 30

Veris 90

Variance

F-ratioa

Variance

F-ratio

pH

34

***

9

*

Cl

27

***

1

n.s.

EC

9

*

12

*

N (mg/kg)

2

n.s.

0

n.s.

N (kg N/ha)

2

n.s.

0

n.s.

S

0

n.s.

0

n.s.

θg

15

**

15

**

aF-ratio is significant at 5, 1 or 0.1% (*, **, *** respectively); n.s. indicates the F-ratio is not significant.

Based on the profile data, the ECV data to 30-cm was most correlated with pH and Cl, while ECV data to 90-cm was most correlated with pH and EC (Table 3). The variation in the ECV data to 30-cm was most closely accounted for by pH and Cl (P<0.001), but the percent variance was <35% (Table 5).

Validation of ECV

Soil conductivity, and that measured by the Veris, is a function of soil salinity, clay content, and water content, and provision must be made for the effects of the other non-estimated properties on the conductivity measurement. In non-saline soils, spatial variation in soil moisture content is the major factor in determining variations in soil conductivity (Suddoth et al. 1997). Researchers at ACPA suggest then that when soil conductivity is measured in non-saline soils with a full moisture profile, such as at the end of a long-fallow, the ECV values are a surrogate of soil texture (APCA 2001).

Although there were some reasonable correlations between measured soil properties and the ECV data, some of these relationships were unexplainable. Variation in soil texture was low, and all 16 samples had >65% clay content. This may have led to a similarly narrow range in ECV data within each field. In spite of small variations in these soil properties, associations between ECV and soil water content and/or OC do suggest potential to use ECV as a surrogate to determine spatial patterns of soil water supply for these clay soils.

The sampling methodology may also have limited the ability to obtain dependencies between ECV data and soil properties. The use of a single core at each site, and added errors in site location, may have contributed to discrepancies. Multiple regression analysis may highlight related dependencies between ECV data and a number of soil properties.

Acknowledgements

We wish to thank Jamie Grant and Rob Taylor for access to their fields. We are also grateful to Brett Whelan (ACPA) for making the Veris 3100 available for this study. Staff at DPI’s Leslie Research Centre, Toowoomba, conducted the soil analysis. This work was made possible through the provision of funds by GRDC and QDPI, and in-kind support from Wesfarmers Landmark and RDS Technologies.

References

Australian Centre for Precision Agriculture (ACPA) (2001). “Preliminary results with the VERIS soil electrical conductivity instrument.” (Australian Centre for Precision Agriculture: The University of Sydney). URL: www.usyd.edu.au/su/agric/acpa/veris/.

Harris, P.A. (ed.) (1998). “Central Darling Downs Land Management Manual.” (Resource Management, Department of Natural Resources: Brisbane).

Minasny, B., McBratney, A.B., and Whelan, B.M. (1999). ‘VESPER version 1.0.’ (Australian Centre for Precision Agriculture, The University of Sydney: Sydney). URL: www.usyd.edu.au/su/agric/acpa.

Rayment, G.E., and Higginson, F.R. (1992). ‘Australian Laboratory Handbook of Soil and Water Chemical Methods.’ (Inkata Press: Melbourne).

Suddoth, K.A., Hummel, J.W., and Birrell, S.J. (1997). Sensors for site-specific management. In ‘The State of Site-Specific Management for Agriculture’. (eds. F.J. Pierce and E.J. Sadler.) pp. 183-210. (ASA/CSSA/SSSA: Madison, WI.)

Top Of PageNext Page