on 
er 
of 
ss 
Barrett, Rachel 
  
AGRICULTURAL CROP IDENTIFICATION USING SPOT AND LANDSAT IMAGES IN 
TASMANIA 
Rachel BARRETT, Paul CROWTHER”, Rowland LAURENCE “, Ross LINCOLNE 
University of Tasmania, Australia 
“ School of Agricultural Science 
Rachel.Barrett(@utas.edu.au 
"Tasmanian Institute of Agricutlural Research 
Rowland.Laurence(@utas.edu.au 
"School of Computing 
P.Crowther@utas.edu.au 
a 
Central Science Laboratory 
R.Lincolne@utas.edu.au 
KEY WORDS: Agriculture, Remote sensing, Multi-spectral data, Image processing, Recognition 
ABSTRACT 
Recording precise crop location and paddock rotational history is becoming an increasingly important part of crop 
management and product quality assurance. Advances in satellite imagery now allow economies of scale in the 
calculation of paddock position, area measurement and mapping, when compared with aircraft-based photography. 
Crop recognition, growth and health monitoring are now also considered achievable goals using satellite imagery. This 
paper reports on the ability of current technology to recognise and differentiate between horticultural crops in Tasmania, 
Australia. The methodologies used for recognising crops, the testing methods used and the results of two years of 
interpretation are discussed. 
1 INTRODUCTION 
Tasmania is a source of many high value, commercial crops, including poppies for the pharmaceutical industry, 
pyrethrum for the production of natural insecticides, potatoes, onions and peas. The cool, temperate climate, winter 
dominant rainfall and variable soils enable a wide variety of intensive horticultural crops to be grown in areas from 
about 2 up to 20 hectares. 
In 1996, the University of Tasmania initiated a project in response to a request from local industries. The aim was to 
find a simple and cost effective method of recognising crops in agricultural areas and record their exact location. 
A pilot study indicated that distinction between some crops was possible with data from good sequential satellite 
imagery. In October 1997 funding was secured to investigate the ability of current technology to identify and 
differentiate between the many horticultural crops grown using commercially available, multi-spectral satellite images 
for the growing seasons of 1997-98 and 1998-99. The first stage of this project was to develop a methodology for the 
systematic recognition of individual crops. If this were achieved the establishment of growth and yield patterns could 
lead to large scale monitoring of agronomic influences. In addition the basis for a simple and cost-effective paddock 
inventory system would be provided. 
1.4 Study Areas 
Two representative but distinct study areas of intensive horticultural production within Tasmania, the North West Coast 
and the Coal River Valley were selected, since different crops grown in distinct climatic and edaphic environments may 
have dissimilar growth signatures (Nieuwenhuisef al, 1996). This paper will report on the North West study area alone. 
The North West Coast study area lies between 41? 07" 13" S and 41° 18’ 45" S and from 146? 03? 58" E to 146? 34^ 
2" E (Figures 1& 2) Average annual rainfall in the study area ranges from 750mm to 1000mm, of which 
approximately two thirds occurs in winter. With mean daily maximum temperatures rarely exceeding 32 °C, summers 
are mild. The study area receives seven to eight hours of direct sunlight in mid-summer and about four hours in mid- 
winter. 
  
International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part B7. Amsterdam 2000. 133