DEVELOPMENTS IN MONITORING RANGELANDS USING REMOTELY-SENSED CROSS-FENCE COMPARISONS

This paper presents a new method for the use of earth-observation images to assess relative land condition over broad regions, using a cross-fence comparison methodology. It controls for natural spatial and temporal variables (e.g. rainfall, temperature soils, ecosystem) so that we can objectively monitor rangelands and other areas for the effects of management. The method has been tested with small and large scale theoretical models, as well as a case study in South Australian rangelands. This method can also be applied in other systems and experiments such as field trials of crop varieties as a robust spatial statistic. METHODS Cross-fence sample pairs are often used in field ecology to control spatio-temporal variation allowing a direct comparison of grazing pressures. In our method, we extract a very large number of cross-fence pairs of pixels from a vegetation index (or land cover) raster (figure 1). The average ratio of vegetation cover from this cross-fence pair is determined for each fenceline in a system, and this cross-fence ratio is used to generate an equation for the value of each paddock. The resulting series of homogenous linear equations is solved in order to rank each paddock objectively