State-wide inter-annual changes to foliage projective cover: better products from higher resolution satellite imagery and improved processing methodologies

摘要

Landsat and SPOT5 imagery have been used to derive foliage projective cover (FPC) over the states of New South Wales (NSW) and Queensland (QLD). Year-to-year changes have been analysed based on satellite observations as far back as 1986. The products created provide stakeholders and decision makers with a solid information base relevant to vegetation management planning and compliance. This study focuses on the continuing improvements that have lead to more refined FPC products. The original FPC products were based on medium resolution Landsat TM and ETM+ imagery (30m resolution), but since 2005 SPOT imagery with a resolution of 10m or 2.5m pan-merged has become available. The signal recorded by the satellite detector is the radiation received at the top of the atmosphere at a specific viewing angle convolved with the optical filter function. But to reliably compute FPC we require the NADIR viewed reflectance on flat ground. Past image processing on ortho-rectified imagery included corrections for the optical filler function, cloud masking, and a practical atmospheric correction based on the 6S algorithm (Vermote, 1997). Recent improvements to the cloud masking are described in a separate presentation (Fisher 2011). The improved atmospheric correction includes a new Bidirectional Reflectance Distribution Function (BRDF) module that accounts for topographic differences including slopes. As a result the scene-to-scene variability of the final FPC product has been reduced, while at the same time providing much finer detail.