Abstract
Suitable methods for measuring and monitoring the condition of riparian environments are being investigated by government agencies responsible for maintaining these environments in Australia. The objective of this work was to compare two riparian condition assessment approaches, the Tropical Rapid Appraisal of Riparian Condition (TRARC) method developed for rapid on-ground assessment of the environmental condition of savanna riparian zones and an image based riparian condition monitoring scheme. Measurements derived from these two approaches were compared and correlated. The sample representativeness of the TRARC method was evaluated and the cost-effectiveness and suitability for multi-temporal analysis of the two approaches were assessed. Two high spatial resolution multi-spectral QuickBird satellite images captured in 2004 and 2005 and coincident field data covering sections of the Daly River in the Northern Territory, Australia were used in this work. Both field and image data were processed to map indicators of riparian zone condition including percentage canopy cover, organic litter on the ground, canopy continuity, tree clearing, bank stability, and flood damage. Spectral vegetation indices, image segmentation, and supervised classification were used to produce riparian health indicator maps. QuickBird image data were used to examine if the spatial distribution of TRARC transects provided a representative sample of ground based estimates of riparian health indicators. Covering approximately 3% of the study area, the sample mean of the TRARC estimates of individual indicators of riparian zone condition were in most cases within 20% of the global mean derived from the whole imaged riparian area. The cost-effectiveness of the image based approach was compared to that of the ground based TRARC method. Results showed that the TRARC method was more cost-effective at spatial scales from 1
km to 200
km of river in relatively homogeneous riparian zones along rivers with only one channel, while image based assessment becomes more feasible at regional scales (200–2000
km of river). A change detection analysis demonstrated that image data can provide detailed information on gradual change, while the TRARC method is less suited for multi-temporal analysis due to the ranked data format, which inhibits precise detection of change. However, results from both methods were considered to complement each other for single date assessment of riparian zones if used at appropriate spatial scales.