Gerald Page

The cumulative impact of 150 years of grazing has led to widespread degradation in the West Australian rangelands, with the future impacts of increasing temperature and changing precipitation regimes associated with climate change uncertain. Monitoring the ecological condition of these seasonally dynamic rangelands is a difficult task owing to the vast size and remoteness of these ecosystems. Disturbances are not uniform across the landscape nor through time and are further complicated by the episodic nature of weather systems that have a strong influence of the year-to-year changes in condition. Therefore, methodological approaches to monitoring that combine field measurements with remote-sensing tools represent a powerful and cost-effective toolset to conduct spatial analyses of changes in ecological condition that can incorporate topography, weather and management interventions at the landscape scale. While there are existing continental-scale vegetation cover-change products, here we aimed to test a regionally specific plot to landscape scale method to predict projected foliar cover (PFC, %) using Landsat imagery and a digital elevation model in semi-arid shrublands in Western Australia calibrated using either ground-based data collection methods or image-processing based methods to extract training data from high-resolution (50 cm) satellite imagery. We then tested the utility of predicting PFC from an annual time-series of Landsat imagery between 1989 and 2019 to detect changes in vegetation cover over time. We will present an analysis of long-term (30 year) changes in woody vegetation cover over a 350,000 hectare area as well as the intra-annual dynamics of canopy and ground-cover species.