Previous PageTable Of ContentsNext Page

Remote sensing as a potential precision farming technique for the Australian rice industry

Sarah Spackman1,2,4, David Lamb1,2 , John Louis1,2 and Gary McKenzie1,3

1 CRC for Sustainable Rice Production, Charles Sturt University, Locked Bag 588, Wagga Wagga, NSW 2678, Australia.
2
Farrer Centre, Charles Sturt University, Locked Bag 588, Wagga Wagga, NSW 2678, Australia.
3
Spatial Analysis Unit, Charles Sturt University, Locked Bag 588, Wagga Wagga, NSW 2678, Australia.
4
Corresponding author: (02)69 32552, fax (02)69 332737, dlamb@csu.edu.au

Abstract

The rice crop growth model ‘maNage Rice’ is commonly used by NSW Rice farmers to estimate mid-season nitrogen application rates for their fields based on a suite of basic plant measurements completed at panicle initiation. Currently most rice growers treat their crops as being spatially uniform at the bay scale. This paper describes the use of multispectral airborne imaging to provide spatially referenced input data into the model as a potential means of quantifying within-field variability in nitrogen demand. As a first step, an evaluation of errors associated with the process of estimating crop dry-weight biomass from remotely-sensed imagery and subsequent incorporation of this data into the model, will be outlined.

Introduction

New South Wales, Australia, has one of the highest yields or rice per hectare in the world. In Australia, rice is grown in a temperate climate under irrigated conditions. Rice yields average 10 t/ha, whereas countries such as USA and India yield 6 t/ha and 2.9 t/ha respectively (Rice Facts Index 1999). Even so, researchers speculate that NSW yields have not reached their full potential due to limiting factors including poor root-zone aeration, low-tiller survival, low-minimum temperatures and water stress associated with high evaporative demands (Horie et al. 1994; Williams 1997). Levels of nitrogen (N) in fields also has a major influence on rice productivity and this is still considered as one of the most effective, and easiest, variables that growers can directly influence to maximise yields.

At the bay-level, the majority of rice growers treat their crops as being spatially uniform. However, harvest yield maps acquired since 1998 reveal significant variations in yield can occur within them; up to 7-fold. Given the importance of N in rice productivity, an notion of spatial variations in N-demand, coupled with detailed knowledge of soils and other important biophysical parameters, may open the way for more effective management of crop productivity.

The temperate rice yield crop growth model (TRYM) incorporated in the decision support system of ‘maNage Rice’ is one example. The TRYM model helps NSW rice growers estimate mid-season N application rates required to achieve maximum yields for a range of expected climatic conditions. The TRYM model predicts yield in response to various mid-season N application rates (Beecher et al. 1995). The model is based on in-situ measurements of plant N, water level, shoot density and dry weight biomass (DW) at panicle-initiation (PI), the date of sowing and when PI occurred. The model also requires information about daily solar radiation and temperature, all of which can be obtained from historical meteorological data.

Remote sensing, in particular airborne multispectral imaging, is becoming more widely used in Australia as an aid to precision agriculture activities (Lamb, 2000). More recently, the technology is also being applied to row crops such as grape vines (Lamb et al. 2001). The inherent ability of remote sensing to detect and map regions of differing crop biomass within single fields makes in an important source of metre-resolution data suitable for initialising the TRYM model, thereby allowing it to be executed on individual zones characterised by certain levels of initial plant biomass.

In extracting the necessary parameters from remotely-sensed imagery, and then executing a crop growth model based on this data, care must be taken to understand the generation, and subsequent propagation of errors from the beginning of the process through to the final model output. This paper investigates, and quantifies these errors.

Acknowledgments

The authors gratefully acknowledge the Cooperative Research Centre for Sustainable Rice Production (CRC Rice) for provisions of the research funding and a postgraduate scholarship (SLS), the staff of the Farrer Centre (CSU) for acquiring the airborne imagery, G. Batten for the measurements of DW. Helpful discussion with S. Black and J. Medway throughout the conduct of the project is also acknowledged.

References

Beecher, H. G., McLeod. G. D., Pritchard, K. E. & Russell, K. (1995). Benchmarks and Best Practices for Irrigated Cropping Industries in the Southern Murray Darling Basin. NSW Agriculture: Murray-Darling Basin Commission.

Horie, T., Ohnishi, M., Angus, J. F., Lewin, L. G. & Mantano, T. (1994). Physiological characteristics of high yielding rice inferred from cross-location experiments. Temperate Rice Conference Proceedings, Vol. 2, "Temperate Rice - achievements and potential". 21-24 Feb. Yanco, Charles Sturt University, Australia, pp 635-650.

Lamb, D. W. (2000). The use of qualitative airborne multispectral imaging for managing agricultural crops– A case study in south eastern Australia, Aust.J.Exp.Ag. 40 (5): 725-738.

Lamb, D.W., Hall, A. & Louis J. (2001). Airborne remote sensing of vines for canopy variability and productivity, Australian Grapegrower & Winemaker, Annual Technical Edition. (In Press).

Rice Facts Index 1999. Rice Facts. http://www.riceweb.org/ [Accessed 10 October 2000]

Williams, R.L., Angus, J. F., 1997. maNage rice: a software package to assist Riverina ricegrowers with decisions about topdressing nitrogen fertiliser. CSIRO Australia, NSW Agriculture.

Previous PageTop Of PageNext Page