The feasibility of using satellite SAR images to monitor pasture in Australia

Abstract

Pasture is a vital land use type in Australia. Farmers need near real-time information at whole-farm and within-paddock scale to assist management decisions including grazing rotations and feed budgeting. The application of optical remote sensing for pasture is limited by weather condition such as clouds and index saturation problem. Synthetic Aperture Radar (SAR) satellite remote sensing has the potential to solve clouds and saturation problem and is sensitive to moisture of grassland, because of the use of microwave. Its high spatial resolution makes it possible to pasture monitoring at paddock scale. Thus, SAR has the advantage in providing frequent and accurate measurements of multiple pasture properties (e.g. biomass, plant water content and soil moisture) of at both whole-farm and within-paddock scale. This thesis aims to evaluate the feasibility of using multiple satellite SAR data to estimate temporal and spatial variation of pasture properties at paddock scale in four different sites (Muradup, Gippsland, Ipswich and Otway) in Australia. A large number of ALOS PALSAR (L-band), ENVISAT ASAR (C-band), TerraSAR-X (X-band) and COSMO-SkyMed (X-band) images were used. SAR backscatter was correlated with NDVI, NDWI and soil moisture index (M.I), to evaluate the capability of SAR for estimating pasture biomass, plant water content and soil moisture. With NDVI, NDWI and M.I as proxy of biomass, plant water content and soil moisture, the method used here saves lots of field work cost. This study was the first attempt to analyse, at both the temporal and spatial domains, multiple pasture properties in Australia, by integrating SAR images at X-, C- and L-band with optical images. The results demonstrated that SAR overcomes saturation problem of optical and delivers more accurate measurements than optical. The ability of SAR backscattering to estimate pasture parameters changes with sensor (wavelength, polarisation, incidence angle) and grassland properties (vegetation type, biomass, plant water content, soil moisture). The advent of multiple SAR satellites and short revisit interval of new sensors (e.g. TerraSAR-X and COSMO-SkyMed) makes it possible to observe pasture densely. Integration of SAR and optical will eventually achieve near real-time monitoring of pasture to solve the problem of lack of optical data.