Using EM and gamma maps to map soil types and help locate subsoil constraints for management

Key Messages

  • Soil maps improve the targeting of sampling for constraint and nutrients management, as well as variable rate management.
  • EM and Gamma farm maps can be overlayed to make simple soil type maps.
  • Using low, medium and high gamma K or TC counts classes and low and high EM classes 6 soil types can be mapped to define areas of sand, saline sand, gravels, duplex soils, gravelly duplex and clay soils.
  • Point data, such as soil texturing and soil chemistry and farmer knowledge were used to adjust and evaluate the gamma and EM classes.

Aims

The aim is to create simple paddock and farm scale soil maps, using ground based geophysics measurement of EM and gamma, for variable rate management such as nutrient or soil constraints.  The main soils mapped, were defined after discussions with advisors and farmers about which soils they manage differently such as light versus heavy soils (i.e. sands vs. clays), gravels, and duplex soils (sand over clay, or loam over clay) where they would like to identify the depth to clay (pers com David Hall) (Table 1).

Previously EM and gamma maps were used separately, and the measurements were correlated to soil chemical properties such as cation exchange capacity (CEC), organic matter (OM), % clay and electrical conductivity (ECe) (Cook et al. 1996). Soil property zones were then created from these correlations based on statistical approaches, fuzzy classification or decision trees (McBratney et al. 2003).  These maps were difficult for famers to understand, and not useful for management of soil constraints as farmers and advisors wanted soil units which related to farm management options.

Combining EM and gamma, to map four soil types was tested on a paddock in WA (Wong et al. 2008). We expanded this approach to map six soil types (Table 2) and tested the approach on 29 paddocks and one whole farm in WA (Figure 1).

Table 1. Commonly recognised and managed WA soil types (DPIRD MySoil types (van Gool et al, 2018), with descriptions, area in the Wheatbelt, soil constraints and common management practices.

DPRID MySoil type

Soil description

Area of WA wheatbelt (%)

Constraint

Management

Sands
&
Sandy Earths

Sandy texture throughout

Sand grading to loam

24%

Low pH, compaction, non-wetting, nutrient leaching

Liming, ripping, mouldboard, non-wetting agents, nutrient timing

Gravels

Ironstone gravel (>20%) within 15cm from surface, greater than 20cm thick.

7.8%

Nutrient leaching, poor water holding capacity, non-wetting

Manage differently and care with deep ripping, nutrient timing

Duplex soils *

Sandy or loamy surface texture. Abrupt change to sandy clay loam or clay.

32%

Low pH, compaction

Shallow or deep duplex*, depth to duplex

Liming, ripping, spading, mouldboard, delving (depth dependant)

Gravelly duplex

Gravel over rock or clay at < 80cm

8.2%

Low pH, Non-wetting
Clay subsoil may have salinity or sodicity

Liming, nutrients
Gypsum, OM amendment
Care with management

Clay
&
Shallow loamy duplex

Loamy or clayey surface texture and clayey subsoil

Loamy surface texture, abrupt change to clayey subsoil

15%

Salinity, Sodicity, Boron

Gypsum, OM amendments, slotting, water harvesting

* Shallow duplex = abrupt texture change at 30cm or above, deep duplex =abrupt texture change between 30 and 80cm

Method

The ground based measurement of EM and Gamma as well as point data of soil descriptions, soil chemistry and PAWC measurements were sourced from the GRDC invested Subsoil Constraint project (DAW00242) and additional sites from previous GRDC invested precision agriculture projects.  The case study sites included 16 farmers, 29 paddocks (2733ha) plus 1 whole farm (>20 paddocks, 2466 ha) and 355 soil sampling points (Table 3). Sites were spread across the WA wheatbelt (Fig 1). In this paper we present results from a case study in the Northern region (N1) to describe the methodology developed.

Image of south west of western australia

Fig 1. Location of the case study paddocks with geophysics and point soil measurements in WA.

Soil point data was assigned to a WA soil class (van Gool et al 2018, MySoil), using available data which could include: hand texture, gravel, particle size analysis (PSA), plant available water capacity (PAWC), soil chemistry and visual descriptions in layers up to 1 m depth. Soil descriptors from hand texture used:

Sandy soils  - fine, medium and coarse sand, loamy sand and clayey sand  - 5-10% clay

Loamy soils - sandy loam, loam, silty loam, sandy clay loam and clay loam – 10-35% clay

Clayey soil - light, medium and heavy clay, sandy and silty clay – >35% clay

Duplex soils – have an abrupt texture change (>20% difference in clay content), with shallow duplex change occurring at <30cm and deep duplex change between 30 and 80cm

Gravel soils – greater than 20% gravel Ironstone gravel within 15cm from surface, greater than 20cm thick.

Gamma and EM measurements were smoothed to a 10m x10m grid spacing using kriging. The EM readings are related to water content, salts in the soil (EC) and clay content. The EM values, were separated into low and high classes and mapped using red and blue colours (Table 2). The gamma (K counts or total counts) correlates with clay and gravel content. The gamma counts were separated into 3 classes, low, medium and high and assigned a dark grey, light grey or yellow colour respectively (Table 2) and mapped with a 50% transparency. The gamma and EM maps were then overlayed to produced 6 colour classes which roughly correspond with broad soil types, eg. MySoils (Table 2).  The EM and gamma cut-off values were based on the point data for the various locations.

The combination of EM and gamma can improve the mapping of the major soil differences, compared to just using them individually. This is due to the combination of sands, clays, gravels and salts in the WA soils. For example, low gamma is associated with sandy soils with the EM used to determine if the sand is saline or not.  High gamma is related to clay and clay/gravel soils and EM can separate these two soils as clay soils usually have a higher EC compared to gravel. Medium gamma means there is some gravel or some clay but there is also sand present, usually associated with duplex soils (sands over clay) and sandy gravels. In this case the EM is able to separate these to soils as the gravels have low EC, while the duplex soils has a higher EM due to the clay subsoil (Table 2).

Table 2. Overlaying the gamma (with 50% transparency) on the EM layer produces 6 colours, which are correlated with the major WA soil groups

Gamma

EM

WA MySoil type and Map colour

Low

Low

Sand and Sandy earth

Low

High

Coloured sand – saline/ loams

Medium

Low

Gravel

Medium

High

Duplex soils

High

Low

Gravelly duplex

High

High

Clay and shallow loamy duplex

Results

We show one of the case study sites (3 paddocks of N1) where using the soil point data the gamma K classes were low <40 counts, medium 40-70 counts and high >70 counts, and the EM values were low at <20 units and high at > 20 units.  The table shows the point ID with the soil description from hand texturing and the colour of the cell indicates the colour/soil type from the mapping. The only site where the description did not match that from the mapping was site 5, a deep duplex soil was mapped as a gravel.

Table 3. Soil points with their ID number, soil description and mapping soil type (colour of cell) for three paddocks at Site N1 (Figure 2.).

Point ID

Soil Description

Point ID

Soil Description

Point ID

Soil Description

1

Sand

3-20

shallow gravel duplex  @ 30cm

5

deep duplex @ 60cm

2-10

Sand

14

shallow gravel duplex @ 25cm

11

deep duplex

2-20

Sand

15

deep duplex @55cm

8

deep duplex @80cm

12

sand

13-10

deep gravel duplex @40cm

10-10

shallow  duplex @25cm

9

sand

13-20

deep gravel duplex @40cm

3

shallow duplex @25cm

6

sand

3-10

deep gravel duplex @ 50cm

10

clay

7-10

sand

    

4

sand

    

7

sandy earth- loamy

    

In this example (N1), the EM layer separated the high gamma areas into gravels and clays (Figure 2). The medium gamma class was separated into gravelly sand and duplex soils by the EM cut-off level used. This example shows that the combination of EM and gamma is required to map soils on this farm.

Image of gamma k separated in 3 classes in the farm

Figure 2. The gamma K separated into 3 classes, the EM separated into 2 classes and the subsequent soil map from overlaying the two layers for Case study N1.

For the northern sandplain cases, the classes required some refinement (Table 4). The gamma was low as the paddocks consisted of mostly sandy soil types and did not have ironstone gravels or clay soil types.  For these sites, the gamma was split into 2 classes, with no high gamma class. For NS1, the low gamma was the quartz gravel soils and the medium gamma with low EM the sand, and medium gamma with high EM the deep sandy duplex. For NS2, the low gamma with high EM was a sandy earth, and the medium gamma with low EM the sand (there was only 2 soil classes on this paddock). It is important to note that the divisions between soil types are relative, using these mapping techniques, and will not always correlate precisely with formal soil classes.

We applied this methodology across the remaining case studies and found the cut-off value for the high gamma class was similar to the mean of the gamma for the paddock (Table 4). For 2 farms, at Kellerberrin (E1, E2) and Bodallin (E4 and E5), the individual paddocks did not have the full range of the gamma but when they were combined, the average gamma was near the cut-off values chosen. The low gamma cut-off value is similar (but not quite) to the mean subtracting 0.6 x Standard deviation. At some of the sites, the mean was also useful as a first estimate to separate the low and high EM classes. Other spatial information such as terrain attributes e.g. elevation above sea level, slope, etc. could also be introduced to further differentiate likely differences in soil types.

Table 4. Description of available data across all sites, paddock size, number of soil samples and gamma and EM (mean & standard deviation (SD) results for estimating mapping cut off values.

Site

Nearest town

No pdk

Area (ha)

No points

Gamma

EM

     

type

mean

SD

Mean-0.6*SD

Cutoffs

Low, High

mean

SD

Cutoffs

Northern

N1

Coorow

3

243

32

K

68

49

39

40    70

22

 

20

N2

Buntine

3

168

26

K

82

38

59

50    80

32

35

20

N3

Buntine

3

265

41

K

67

34

47

40    70

21

11

25

Northern sandplain

NS1

Mingenew

2

87

13

TC

148

20

 

135    --

15

2

15

NS2

Erudu

1

59

4

TC

183

12

 

170    --

1.2

0.5

1.9

Eastern

E1

Kellerberrin

1

67

9

K

40

22

  

9

8

15

E2

Kellerberrin

1

69

6

K

85

52

  

22

21

 

E1+E2

  

136

  

62

37

40

40    70

   

E3

Greenhills

2

142

8

K

61

17

51

30    60

50

30

15

E4

Bodallin

1

150

6

K

114

9

  

111

29

82

E5

Bodallin

1

115

21

K

56

17

     

E4+E5

  

265

  

83

 

77

45    70

   

E6

Corrigin

3

89

15

K

18

13

10

13    20

7

4.5

10

South

S1

Dumbleyung

3

332

21

Kconc

1.27

0.3

1

0.75  1.25

146

53

110

S2

Dumbleyung

>20

2466

68

Kconc

1.27

0.52

1

0.93  1.25

74

61

45

South Coast

SC1

Myrup

1

79

15

K

11

3.6

9

-      12

46

25

85

SC2

Scaddan

1

325

26

K

27

8

22

15   25

128

48

40

SC3

Munglinup

1

195

17

K

14

5

11

8     13

131

56

55

SC4

Howick

1

164

16

K

13.7

2.5

12

9     14

44

14

50

SC5

Salmon Gums

1

183

11

K

17

6

13

12    15

111

48

100

Conclusion

Simple soils maps which can be used for management of nutrient and subsoil constraints can be created by overlaying farm scale EM and gamma spatial layers. Splitting the gamma into 3 classes and EM into 2 classes enabled differentiation of up to six soil types. Using the mean and standard deviation of gamma to estimate the class cut off values and the mean for EM is a good starting place for mapping different soils for practical management applications and can be refined with farmer knowledge and soil description sites. Using one soil type at a time, we can use the good areas to benchmark the yield the poor areas should be achieving. Farmer knowledge, visual observation and soil sampling of the poor areas of a soil type can then identity the likely nutrient and soil constraints.  Once the constraint and area of the constraint is identified the economic analysis tools such as ROSA (Ranking options for Amelioration) (Petersen et al. 2018) can be used to decide which of these poor areas to target for future removal of soil constraints.

Future work on determining depth to clay using on-the go sensors would improve this mapping to enable the duplex soils to be managed particular for delving and other soil mixing amelioration strategies.

Acknowledgments

We would like to thank the farmers who provided farm data and allowed access to their farms for soil sampling.

References

Cook SE, Corner RJ, Groves PR, Grealish GJ (1996). Use of airborne gamma radiometric data for soil mapping. Aust. J. Soil Res. 34:183–194.

McBratney A, Mendon¸ca Santos M, Minasny B (2003) On digital soil mapping. Geoderma 117, 3–52.

MYSoil   

Petersen E, Lemon J,  Xayavong V. (2018) Decision tool to estimate the economic benefits from soil amelioration at a paddock and industry scale.  2018 Grains Research Updates, Perth 26-27th Feb 2018.

van Gool D, Stuart-Street A, Tille  P (2018) Distribution of classified soils in south-west Western Australia, Resource management technical report 401, Department of Primary Industries and Regional Development, Perth.

Wong MTF, Asseng S, Robertson MJ, Oliver Y (2008). Mapping subsoil acidity and shallow soil across a field with information from yield maps, geophysical sensing and the grower. Journal Precision Agriculture 9, 3-15.

GRDC Project Number: Subsoil Constraints: Understanding and Management DAW00242

GRDC Project code: DAW00242