Implications of Precision Agriculture for soil resource assessment

1996

A.B. McBratney, B.M.Whelan and R.A. Viscarra Rossel
Australian Centre for Precision Agriculture, McMillan Building A05, The University of Sydney NSW 2006, Australia.

ASSI & NZSSS National Soils Conference, Melbourne, Volume 2, pp.161-162.

INTRODUCTION

The spatial variation of agronomically significant soil attributes has become a subject of importance to the farming and wider communities (Larson and Robert 1991; Robert et al. 1995). The notion that productive land can be treated as a relatively homogeneous resource at the "within-field scale" is now being questioned (McBratney and Whelan 1995a). Such an assumption, and the subsequent uniform application of planting material, chemicals, or tillage effort may result in zones within a field being under or over-treated. Arising from this are problems associated with the inefficient use of input resources and the environment, such as economically significant yield losses, excessive chemical costs, gaseous or percolatory release of chemical components, unacceptable long-term retention of chemical components and a less than optimum growing environment. The production of accurate continuous maps of field variation in important soil variables and subsequent crop yields may significantly improve the quality and quantity of information on site variation available to land managers. Such information forms the basis of precision agriculture. This approach delineates the various production potentials of the site and highlights where management actions can be varied to attain optimum production. By providing a more thorough understanding of the variation in a site, the efficiency of resource use, such as fertilizers and irrigation water, may be increased. It is expected that differential management strategies will be regarded as 'best practice' in the not-too-distant future and therefore this places the onus on soil and land scientists to provide precise but economical statements of soil resources within individual paddocks.

IMPLICATIONS

The implications of precision agriculture are to provide:

1. Mapping strategies to satisfy an unprecedented demand for precise soil and other land information to guide site-specific management; and
2. Assessment procedures that can translate soil data into site-specific management decisions.

Mapping

For precision agriculture, estimates of a range of soil properties are required at scales of 10 metres or less. Routine soil maps, especially the 1:100 000 maps available in Australia are clearly inadequate for this purpose. These maps may be useful for stratification for calibration purposes. An alternative is to combine grid or random sampling in association with digital elevation models and other environmental data in a generalised geostatistical approach. Soil sampling may still be too expensive, however. Therefore scanning and sensing technologies must be developed in the long run (McBratney and Whelan 1995a) and these may be soil-invasive or not. Non-invasive methods provide estimates of soil condition at various depths from above-ground observations. Electromagnetic induction instruments used for salinity mapping and ground-penetrating radar offer non-invasive methods of soil mapping and such techniques need to be tested for their ability to provide agronomically significant information. The development of invasive sensors for a whole range of soil properties also needs to be investigated. Particularly targeted is the construction of sensors to be applied in studying the continuous variation in clay content, moisture content, soil colour/organic matter and soil nitrogen levels. Soil physics moved into the field in the 1980s. One of the imperatives of precision agriculture is for soil analytical chemistry to become field-based.

Assessment

Once obtained, information on yield, topographic, soil and other agronomic factors can be integrated into a simple Geographic Information System for the integrative study of the causes of yield variation. This together with statistical and simulation models will suggest appropriate management strategies. A method of analysing the resulting maps is required that will allow informed management decisions to be made based on the spatial pattern of variation present at the site (McBratney and Whelan 1995b). The essential key to this understanding is the prediction of yield or yield components at a point in the growing season. Two approaches need to be investigated separately- although a combination of the two seems a strong possibility. The first is statistical - yield is predicted at each site as a function of measured soil and other agronomic properties. The effect of changing these properties (e.g. soil nitrate) on yield can then be evaluated. This would allow, for example, prediction of optimal local fertiliser requirement. Secondly, the effects of soil and other agronomic properties can be gauged through a simulation model which takes into account the known physiology and agronomy of crops. The outcome of this evaluation process for each proposed treatment is a treatment map which guides the differential management. This is a form of continuous direct land evaluation.

EXPERIMENTAL

Experiments were conducted to design an invasive sensor for simultaneous on-the-go measurement of clay, organic matter and moisture content from reflectance of a suitable wavelength or combination of wavelengths in the near infra-red (Viscarra Rossel 1995). Soil materials were prepared with varying amounts of clay, soil moisture and organic matter according to a response-surface design and the reflectance spectra measured at 2nm intervals from 1300 nm to 2500 nm. Response surface models were fitted to the reflectance data at specified wavelengths. Reflectance showed significant responses to clay content and moisture but not to organic matter. A thorough selection procedure using non-linear modelling and root mean square error of prediction was used to derive the four most suitable wavelengths for simultaneously measuring soil moisture and clay content. In a simulation experiment clay content was more accurately predicted than moisture content. This technique could therefore be implemented as a soil-invasive measurement tool by mounting source and sensor into a tine similar to an American organic matter sensor (Borgelt 1993). Conclusion Precision agriculture, largely the application of information and communications technologies to in-field data gathering and management, may be regarded as 'best practice' for crop growth in the future because of its twin goals of maximising economic returns whilst concurrently minimising environmental impact. The practice of precision agriculture, whether it be to differentially apply fertiliser, seed, pesticide, or tillage requires detailed knowledge of the spatial and temporal variation of crop yield components, weeds, soil-borne pests and attributes of physical, chemical and biological soil fertility. Such detailed information at the scale of a few metres has not previously been required for soil management. This talk reviewed the possibilities for supplying such information from using existing soil maps combined with digital elevation models, to intensive soil sampling combined with geostatistics, to real-time invasive measuring and non-invasive scanning technologies. The implications for soil science in general, and soil resource assessment in particular, are profound. Precision agriculture provides a clear focus and rationale for localised soil resource assessment. Assessment models will require much more focus and, to achieve accurate predictions, the degree of localisation required in the inherent calibrations must be ascertained.

REFERENCES

Borgelt, S.C. (1993). Sensing and measurement technologies for site-specific management. In 'Proceedings of Soil-Specific Crop Management: a Workshop on Research and Development Issues'. (Eds P.C. Robert, R.H. Rust and W.E. Larson.) pp. 141-57. (American Society of Agronomy: Madison, Wisconsin.)

Larson, W.E., and Robert, P.C. (1991). Farming by soil. In 'Soil Management for Sustainability'. (Eds R. Lal and F.J. Pierce.) pp.103-12. (Soil Water Conserv. Soc.: Ankeny, Iowa.)

McBratney, A.B., and Whelan, B.M. (1995a). The potential for site-specific management of cotton farming systems. Co-operative Research Centre for Sustainable Cotton Production, Discussion Paper No. 1, Narrabri.

McBratney, A.B., and Whelan, B.M. (1995b). Continuous models of soil variation for continuous soil husbandry. pp. 323-38. 'Site-Specific Management of Agricultural Systems'. (Eds P.C. Robert, R.H. Rust and W.E. Larson.) pp.323-38. (American Society of Agronomy: Madison, Wisconsin.)

Robert, P.C., Rust, R.H., and Larson, W.E. (Eds) (1995). 'Site-Specific Management for Agricultural Systems.' (American Society of Agronomy: Madison, Wisconsin.)

Viscarra Rossel, R.A. (1995). Aspects of the concept of site-specific soil management. Unpublished BScAgr Thesis, Department of Agricultural Chemistry & Soil Science, The University of Sydney.