| Background |
Weeds cost Australian agriculture
$3.3 billion per annum through lost productivity and control costs. Under
the extensive agricultural systems in Australia, they are of major economic
importance.
However, accurate information on weed distribution, abundance
and temporal change is lacking. Such information is essential to determine
their rate of spread, economic importance, to evaluate the effectiveness
of control strategies and to justify funding for control programs.
Remote sensing offers an objective, low cost and repeatable
means of mapping weeds at a range of scales. It involves the acquisition
of information about a target, using a sensor positioned at some distance
from the target. Remote sensing has value for vegetation survey and assessment,
and its capability for detecting weeds has been investigated in a number
of preliminary studies.
For remote sensing to be able to detect and map weed species,
there are two major requirements. There must be sufficient differences
in spectral reflectance between the target weed and background vegetation,
stubble or soil, and the sensor must have an appropriate spatial and spectral
resolution to detect the target weed. |
| Outcomes |
It was possible to discriminate
both target weeds using remote sensing techniques. Dense to moderate infestations
of both weeds could be detected using Landsat TM imagery, although Scotch
thistle was mapped more reliably (and at lower infestation levels) than
serrated tussock. Landsat imagery was suitable for mapping at a regional
to catchment scale.
The analysis of field spectral data captured across the
growing season proved essential. This was for understanding the best times
of year and areas of the spectrum to discriminate the target weeds from
other vegetation, and the potential of indices/ratios to improve discrimination.
In the future, the difficulties encountered in this study
in the use of airborne imagery should be fewer. More advanced systems will
have better on-board GPS equipment which should reduce locational errors,
and better radiometric calibration will reduce brightness differences.
Airborne imagery should be able to map moderate to light weed infestations,
and at a farm to subcatchment scale.
As hyperspectral systems become available, the potential
to map individual plant species will improve, although the need for good
field spectral data will increase and new techniques to analyse the huge
amount of data produced by such sensors.
The recommended methodology for mapping weeds is to investigate
their reflectance characteristics over the growing season to determine
the best times of year for discrimination and how differences in reflectance
could be maximised using band ratioing. The next step is to use a combination
of multi-date imagery with the ratios for analysis.
This study was successful in mapping areas infested by,
or likely to be infested by both serrated tussock and Scotch thistle. Such
information is of use for a range of purposes, and is in demand by Local
Control Authority Weeds Officers, Landcare groups and coordinators, researchers
and Government bodies. Such information for particular years can be used
as a basis for comparison of change in infestations in the future. |
| Implications |
Studies need to be made on the
utility of the techniques used in this study, and other remote sensing
procedures, to map other important weeds. Such studies should be coordinated
and overseen through bodies who have experience in the area and through
the Cooperative Research Centre for Weed Management Systems. This will
ensure duplication of effort does not occur, and intensively monitored
field sites can be used for both assessment of weed biology, the effectiveness
of biological control agents and remote sensing for weed mapping.
The remapping of areas mapped in this study after several
years is suggested to attempt to quantify change in infestation and the
potential of new remote sensing systems which should be available in the
future. |
| Publications |
McGowen, I.J. (1995). Remote sensing
techniques for mapping and monitoring of weed infestations. In Survey
Workshop, Proceedings of a workshop, Wagga Wagga, December 1995. Cooperative
Research Centre for Weed Management Systems. December, 1995. p 21.
McGowen, I.J. (1997). Weed mapping using GPS, GIS and
remote sensing. In: Proceedings of the 9th Biennial Noxious
Weeds Conference, 16th – 18th September, Dubbo
NSW Australia. NSW Agriculture. pp. 33-42.
McGowen, I.J. (1998a). Remote sensing for mapping of herbaceous
weeds - the background and a case study. In: Proceedings of the 9th
Australasian Remote Sensing and Photogrammetry Conference, 20-24th
July, University of NSW, Sydney, NSW, Australia. (CDROM, Paper No. 115).
McGowen, I.J. (1998b). Remote sensing - background to
the technology and opportunities for mapping of pasture weeds. In Precision
Weed Management of Crops and Pastures, ed. R.W. Medd and J.E. Pratley.
Proceedings of a workshop, 5th – 6th May 1998, Charles
Sturt University, Wagga Wagga, NSW, Australia. Cooperative Research Centre
for Weed Management Systems, Adelaide. 154pp.
McGowen, I.J. (1998c). Remote sensing for mapping of herbaceous
weeds. Remote Sensing and Photogrammetry News 7 (2), 12. |