A Quantitative Appraisal of Woody Shrub Encroachment in Western New South Wales.

Abstract

Landsat Multispectral Scanner (MSS) digital imagery was used to estimate the distribution, density and change in woody shrub cover over time in western New South Wales. The purpose of the project was to derive maps of woody cover which can be used as a basis for regional planning and property planning. Assessment of woody vegetation cover using satellite imagery enables regions which are more susceptible to shrub encroachment to be targeted for control strategies. Dry season images which had minimal green vegetation were used, because the spectral signatures of scrubby ground cover interfered with the proper classification of woody vegetation. For each region, multidate imagery was classified using a pixel unmixing algorithm to derive data sets which showed woody canopy cover. These data were then rescaled to percentage values using aerial photography sampled throughout each region. A geographic information system (GIS) was used to derive changes in woody cover between both dates and to present the data in map form. Most current woody cover in the study area occurs at less than 20% cover, whilst higher levels (40 to 80%) occur in the eastern parts of the Louth and Barnato regions. At least 20,3 10 km2 of the 120,000 km2 study area is already affected by woody vegetation cover levels of greater than 40%, which significantly reduces carrying capacity and pastoral productivity. Changes in woody cover over a 10 to 20 year period were varied. Approximately 24% (26,041 km2) was relatively stable, whilst 20% of the Barnato region had moderate decreases (1 1 to 30%) due to wildfires, and increases of 11 to 30% cover occurred on 'hard red' soils in the east. Emerging woody vegetation of less than 10% cover occurred over 1816 km2 of Sandplains and Stony Lowlands in the Louth and Barnato regions, whilst woody vegetation levels of more than 40% cover occurred in the Barnato region. Considerable 'infilling' of previously unwooded areas was noted for regions which already had high levels of woody cover. A minimal amount of prescribed clearing was apparent from the change data, which suggests that effective control of shrubs is difficult to achieve and that future scenarios will see continued encroachment. The findings suggest that the southern Louth and Barnato regions are most at risk of further shrub encroachment, and that these areas need to be targeted for shrub control. The data provide a quantitative estimate of woody shrub cover which is useful for economic assessments, as well as providing an information base upon which woody shrub management strategies can be developed. Key words: Landsat Multispectral Scanner, remote sensing, geographic information system, change detection, rangeland, monitoring, land cover.