Table Of ContentsNext Page

The use of mutli-sourced remote sensing data for mapping soils and key soil properties in Victoria

S. Ryan1, M. Abuzar2, M. Imhof1 and P. Rampant3

1Agriculture Victoria-Werribee, Department of Natural Resources and Environment,
621 Sneydes Road, Werribee 3030 VIC
(03) 9742 8727ph (03) 9742 8700 (fax) steve.ryan@nre.vic.gov.au
2
Agriculture Victoria-Tatura, Department of Natural Resources and Environment,
Ferguson Road, Tatura 3616 VIC
3
Agriculture Victoria-Bendigo, Department of Natural Resources and Environment,
Bendigo Delivery Centre PO Box 3100, Bendigo 3554

Abstract

The potential for using remotely sensed data to facilitate the mapping of soil types and soil properties that may limit plant growth (sodicity, salinity, alkalinity and acidity) is being assessed at three sites in Victoria. These sites are representative of dryland, broadacre cropping and pasture systems in high and low rainfall zones across Victoria. The research aims to evaluate data generated from satellite, airborne and ground surveys, representing a range of spectral and spatial scales, for mapping soils and their properties. This research forms part of a major investigation in Victoria to develop management strategies to reduce the impact of soil constraints on plant growth.

Identification and mapping of soil constraints follows an integrated approach involving;
1. Characterising key sites to understand soil types, soil properties (surface and sub-surface), and the variability of soil properties at the paddock scale.
2. Collection and analysis of remotely sensed data sets that may have potential to identify and discriminate soil properties for each area of study. These include Landsat ETM, airborne radar (AIRSAR), airborne gamma-ray radiometrics, and ground surveys including EM 31 & 38 and ground penetrating radar (GPR).
3. Integration of remotely sensed information with site data and elevation models to produce maps of the distribution of soil constraints from the paddock to catchment scale.

The integrated approach enables a number of remotely sensed data sets to be evaluated in terms of their effectiveness in providing information on soils and their properties. It is anticipated that maps generated as a result of this work will provide valuable information to land managers and other researchers on the distribution of certain soil properties across these sites.

Introduction

A major project in Victoria, funded by NRE’s Science Technology and Innovation initiative (STII), aims to overcome the constraints to plant growth in many Australian soils. The project has three components which aim to integrate a) soil information, b) research data on plant response and adaptation to soil constraints, and c) management and modelling approaches to allow land users to optimise plant growth in relation to soils and environment.

A key part of the research is to investigate relationships between soil properties and remotely sensed data and to use this knowledge to facilitate the identification of soil properties and to understand their spatial variability from the paddock to catchment scale. This is a three year project and in this paper we report on progress in the first year.

Study sites

To date the research has concentrated on three research sites that represent dryland, broadacre cropping and pasture systems of Victoria. The three sites Birchip, Hamilton and Ruffy are shown in Figure 1.

Figure 1: Location of research sites in Victoria.

The Birchip research site (80 ha in size) is in NW Victoria, and represents a low rainfall (<500 mm) broadacre cropping system. Major soil constraints to plant growth include alkalinity and associated boron toxicity, sodicity and salinity. The Hamilton research site, comprises part of the NRE Pastoral and Veterinary Institute farm (1000 ha approx. in size), in SW Victoria, and represents a high rainfall (>600 mm) cropping (wheat) and pasture (perennial ryegrass) system. The main soil constraints to plant growth in this region include waterlogging and compaction. The Ruffy site (50 ha in size) represents a high rainfall (>600 mm) pasture (perennial ryegrass) site in NE Victoria. Soil constraints to plant growth in this area relate to soil acidity and associated aluminium toxicity, and dryland salinity.

Data collation and methods

Identification and mapping of soil properties follows an integrated approach, combining exiting soils data, new soils information generated from the current research and remotely sensed data sets captured for each of the three research sites. This approach involves:

1. The use of point based sampling to characterise the soils and their properties across key trial sites.

2. Obtaining high resolution remotely sensed data sets that have potential to identify and discriminate between soils and key soil properties for each area.

3. Analysis of these remotely sensed data sets to extract relevant soils information.

4. Integration of remotely sensed information with site data and elevation models to produce maps on the distribution of soil properties from the paddock to catchment scale.

Soil characterisation of trial sites

To understand the spatial variability of soils and soil properties both at the surface and down the profile, a soil survey was conducted across each research site. This involved a reconnaissance soil-coring program, which targeted changes in landform and geology, and resulted in approximately 20 cores being collected for each site. In the field, typical soil information was described, including horizon depth, texture, colour, structure and field pH. Representative samples were also collected for detailed chemical analysis. Based on the results of this information, soil pits were excavated to better understand the spatial distribution of soil properties down the profile and to classify the soils according to the Australian Soil Classification (Isbell, 1996).

Figure 2: Collecting soil samples by coring at the Birchip site. Note the gilgai microrelief at the site.

Remotely sensed data

High resolution remotely sensed data that have potential to identify and discriminate between soils and their properties were obtained for each area. This involved obtaining data, not only from surface measuring sensors, but also from sensors with potential to collect soil information down the profile.

Ground based surveys

Ground penetrating radar (GPR) was evaluated over the three sites in association with Geo-Eng Aust. Pty Ltd to assess its capabilities to map soil horizons and other impeding layers at depth that may limit plant growth (Freeland et al., 1998). A differential GPS collected elevation and location data at the same time as the GPR survey.

Figure 3: Acquisition of GPR data across the Ruffy site.

Electromagnetic induction (EM31 & 38) surveys were conducted to measure bulk soil conductivity, which is influenced by soil porosity, moisture, salinity, clay and organic matter content. EM38 measures soil conductivity to a depth of approximately 1.5 metres, whilst EM31 measures conductivity over a depth of approximately 5 metres.

A field based gamma ray spectrometer survey was also undertaken to measure the potassium, uranium and thorium concentrations in the soil. Information on the spatial variability of these elements will assist in the interpretation of airborne radiometric data.

Airborne and spaceborne surveys

Airborne synthetic aperture radar (PACRIM 2 AIRSAR) data was acquired by NASA over the three research sites in August 2000. AIRSAR is fully polarimetric, captured in 3 wavelength bands C (5.6 cm), L (25 cm) and P (68 cm), and has a spatial resolution of 6 metres. AIRSAR data was combined with the synchronous field measurements for surface roughness, soil moisture, soil salinity and vegetation cover. Radar data has been used successfully to map soil moisture content (Bindish & Baross, 2000; Ulaby et al., 1996) and has potential for mapping salinity distribution (Taylor et al., 1996) and surface soil texture mapping.

Previously captured airborne gamma ray spectrometery (radiometrics) data for regional geological and mineral mapping, held by the Geological Survey of Victoria, is also being utilised in this research. Radiometric surveys detect the abundance of potassium (K), thorium (Th) and uranium (U). Radiometric data are useful for distinguishing the geochemical properties of soils and near-surface rock types (30-45 cm depth) (Pickup & Marks, 2000).

Landsat ETM+ data was obtained for each research site to assist with identifying surface soil properties such as salinity and texture. Landsat ETM+ measures the electromagnetic spectrum in 7 spectral bands in the visible to thermal infra-red, and has a spatial resolution of 25 m. Imagery was obtained over two different times of the year; the first in January 2000 to optimise soil information when vegetation cover is reduced and the second in August 2000 to coincide with the AIRSAR data collection.

Preliminary Results

Results at this stage of the project relate to the soil characterisation of the research sites and some initial interpretation of the data acquired from ground penetrating radar, EM and radiometrics. The AIRSAR and Landsat ETM+ data are still being processed prior to further analysis.

Soil characterisation.

Detailed characterisation for each of the research sites has provided valuable information on the soil types and soil variability at the sub-paddock scale. This information is also important to the other components of the overall project. Results indicate that soil characteristics can vary considerably within and between the sites, and that the spatial variability of some of the plant limiting constraints can be significant at the sub-paddock level.

Soils across the Birchip site are Vertic Calcarosols. The vertical and horizontal distribution of boron, a potential constraint to plant growth when levels become toxic (>3 mg/kg - Bell, 1999), has an interesting spatial distribution. Boron levels in the subsoil are closely associated with gilgai microrelief (Figure 2), with levels being significantly higher in the upper profiles of soils associated with the gilgai mounds and lower in the associated gilgai depressions (Figure 3). A high resolution laser altimetry survey is planned over this site to map this microrelief and to explore this relationship away from the trial site.

Figure 3: Boron levels from soil pits at the Birchip site, variation in boron levels is closely related to landscape position.

For the Hamilton site, several different soil types, based on previous soil mapping (Newell, 1962), have been analysed in detail as part of this research. Soil variability closely reflects the plains and rises associated with Late Tertiary to Quaternary basalts on which they form. Soils on the rises are texture contrast soils comprising of Brown and Yellow Sodosols and Chromosols, with varying amounts of ferruginous nodules (up to 80%) in subsurface horizons. On the plains soils, are mainly Hydrosols, Organosols, and Vertosols. Waterlogging is related to landscape position, with topographically lower areas on the plains often having water lying on them over the winter months. The massive, ferruginous A2 horizons in the soils on the rises are also likely to impede water movement and growth of plant roots.

Soils across the Ruffy research site are classified as Yellow and Brown Kurosols and consist of sandy loam to sandy clay loam surface horizons overlying medium clay subsoils. These soils have developed on Devonian granite, and the depth to this granite varies considerably. The major soil constraint identified at this site is acidity, with all soils tested having a pH (water) below 5.5. At this level there is potential for aluminium toxicity to limit plant root development (Slattery et al., 1999). Variation in pH appears to be fairly uniform across the site and future work is planned to investigate any relationship between soil pH and the depth to the granite.

Remotely sensed data

Ground based sensors were evaluated over each of the three research sites to assist with extrapolating between point based soils information.

Ground penetrating radar (GPR)

Initial processing of the GPR data by Geo-Eng Aust Pty Ltd indicates that results from the survey varied from site to site. The 50 and 100MHz frequencies provided better results than the 200MHz frequency for all sites, as the penetration of the radar signal was limited by the presence of heavy clays in many of the soil profiles.

At the Birchip site the GPR was significantly affected by the presence of light to medium clays throughout the soil profile. However the 50MHz frequency was able to detect a strong reflection at a depth of between 20-25 cm. This reflection may be due the soil structure. Soils at this depth in the gilgai depressions are light to medium clays with a weak blocky structure, whilst on the gilgai mounds the soils are of the same texture but with a strong prismatic structure. For the Hamilton site the best results from the GPR were confined to the texture contrast soils (Chromosols and Sodosols). The depth to a dense layer at 50-70 cm depth was mapped at across on locality, and output from the GPR shows this layer becomes shallower toward the west (Figure 4). The layer is perhaps medium clay above deeply weathered basalts but further pit excavation is required to confirm this. For much of the Hamilton site penetration of the radar signal was significantly reduced due to the high percentage of clay in the soils. The best results for the GPR survey were at the Ruffy site, where it was possible to detect the presence of both outcrop and sub-crop (possible large floaters) of granite (Figure 5).

Figure 4: Example of GPR along one traverse at the Hamilton site. The base (or near the base) of the blue amplitude is possibly the depth to medium clay.

Figure 5: Results from part of the GPR survey of the Ruffy site, highlighting granite outcrops and likely granite sub-crops.

Electromagnetic (EM) induction surveys (EM31 and 38).

The EM surveys have proved useful to delineate soil variability at the sub-paddock level, particularly with the EM31 sensor which measures bulk soil conductivity over a depth of 5 metres. The spatial distribution of EM31 values at Birchip is likely to reflect the composition of underlying geology. High values correspond to a north-south trending rise of Tertiary quartz and ferruginous sandstone, and lower values occur off this rise (Figure 6). Further work is required to interpret the data from the other sites, and all sites will require deep augering to assist in explaining some of the spatial patterns delineated by these surveys.

Figure 6: Spatial distribution of EM31 values over the Birchip site.

Airborne radiometrics.

Airborne radiometrics is being readily incorporated into current soil mapping programs in Victoria and is being utilised in this project to assist with the extrapolation of soils information collected at the paddock level to the sub catchment and catchment levels. To aid the interpretation of the airborne radiometrics, and to identify potential correlations between soil properties and the radiometric data, field based gamma ray spectrometer data is also being collected across all three sites.

Figure 7: Colour composite image of airborne radiometrics captured over the Hamilton region by AGSO and the Geological Survey of Victoria. Red = Potassium, Green = Thorium, Blue = Uranium. Black lines on the image correspond to previous soil/landscape mapping (Maher & Martin, 1987).

Discussion and Future work

The first stage of the project; to characterise the soils across the research sites, has been essential to the project, through identifying soil types and soil properties, and providing important information on the spatial distribution of these properties across each site. The next stage of the research will involve further development of relationships between point based soils information and measurements made by remote sensing systems. This knowledge will be used in the interpretation of the remotely sensed data, and to extrapolate soils information from paddock to catchment scale. Outcomes of the research will include the evaluation of several remote sensing techniques to aid soil mapping and generation of maps on the distribution of soil types and key soil properties across these sites, to provide valuable information to land users and land managers.

Acknowledgments

Funding for this research is provided by the Victorian Department of State and Regional Development through the Department of Natural Resources and Environment’s Science Technology, Innovation (STI) Initiative.

References

Bell, R.W. (1999). Boron. In K.I Peverill, L.A. Sparrow and D.J. Reuter (eds) Soil Analysis: an interpretation manual. CSIRO Publishing, Melbourne.

Bindish, R. and Barros, A.P. (2000). Multifrequency soil moisture inversion from SAR measurements with the use of IEM. Remote Sensing of Environment, 71(1), 67-88.

Freeland, R.S., Yoder, R.E. and Ammons, J.T. (1998). Mapping shallow underground features that influence site-specific agricultural production. Journal of Applied Geophysics, 40, 19-27.

Isbell, R.F. (1996). The Australian Soil Classification. CSIRO Publishing, Melbourne.

Maher, J.M and Martin, J.J. (1987). Soils and Landforms of south-western Victoria. Part 1. Inventory of soils and associated landscapes. Research Report series no. 40. Department of Agriculture and Rural Affairs.

Newell, J.W. (1962). The soils of the Hamilton Pastoral Research Station. Technical Bulletin 15. Department of Agriculture.

Pickup, G. and Marks, A. (2000). Identifying large-scale erosion and depositional processes from Airborne Gamma Radiometrics and Digital Elevation Models in a Weathered Landscape. Earth Surface Processes and Landforms, 25, 535-557.

Slattery, W.J., Conyers, M.K. and Aitken, R.L. (1999). Soil pH, Aluminium, Manganese, and Lime Requirement. In K.I Peverill, L.A. Sparrow and D.J. Reuter (eds) Soil Analysis: an interpretation manual. CSIRO Publishing, Melbourne.

Taylor, G.R., Mah, A.H., Kruse, F.A., Kierein-Young, K.S., Hewson, R.D. and Bennett, B.A., (1996). Characterization of saline soils using airborne radar imagery. Remote Sensing of Environment, 57, 127-142.

Ulaby, F.T., Dubois, P.C. and van Zyl. (1996). Radar mapping pf surface soil moisture. Journal of Hydrology, 184, 57-84.

Top Of PageNext Page