Detecting change in a variable landscape. Monitoring Innamincka Regional Reserve with AVHRR-NDVI and Landsat TM

Abstract

Monitoring arid land condition through field data gathered at remotely located points is expensive and time consuming and heterogeneity within the landscape limits extrapolation of results. Snapshot observations capture vegetation responses to erratic rainfall, obscuring long term trends and variations in observer judgement may further blur results. Such monitoring therefore delivers, at best, low value for money. We explored the potential of broad scale and moderate scale satellite imagery to improve monitoring for the multi-use Innamincka Regional Reserve in north east South Australia. We plotted twice monthly AVHRR-NDVI levels, recorded over 25 years, for the different land systems within the Reserve. The AVHRR NDVI time traces showed that clear differences could be observed between the various land systems in regards to timing and amplitude of response to rainfall. The resulting plots, in conjunction with rainfall data, aided selection of suitable Landsat scenes for image subtraction at dry and wet times 10 years apart. We used Band 3 for subtraction, and interpreted loss in brightness as increase in soil cover. We highlighted the extremes of change in the resulting Landsat subtraction images. This showed where in the landscape vegetation cover losses and gains had occurred. Some of these changes could be linked to on-ground knowledge and observations. This research demonstrated that satellite imagery, combined with rainfall data, overcomes the limitations of time and space that hamper field based monitoring and it can be used to monitor vegetation cover anywhere in the Reserve.