Use of Satellite Imagery for the Assessment of Rehabilitated Land in Central Queensland

Each mine in Queensland has or will negotiate criteria relating to successful revegetation with regulatory authorities and, prior to lease surrender will be required to demonstrate that such criteria have or are likely to be met. Revegetation after open cut coal mining may be influenced by many factors resulting in a mosaic of vegetation. Demonstration of success would be expected therefore to require a survey and monitoring of revegetation, and the authorities encourage regular monitoring prior to this time. Guidelines exist in Australia for natural resource surveys essentially requiring an observation at a density of one site for each one to four square centimetres at the map publication scale. Such intensities for large scale mapping (1:10,000 scale for example) can be resource intensive. Satellite imagery was primarily investigated in this project to assess its value in firstly mapping vegetation and secondly reducing the required field effort for a survey.

Five mines consented to provide test sites for this project these being Callide, Curragh, German Creek, Norwich Park and Yarrabee. The project was limited to satellite imagery available commercially at the commencement of the project. Imagery purchased for this project was ETM (all mines), SPOT (all mines), IKONOS (Callide only) and a digital orthophoto (Yarrabee only). Field data from 1996 to 2002 was available for analysis.

Effectiveness of mapping vegetation with satellite imagery

• Processing of satellite imagery using both spectral and texture information (context) was found to separate transects into vegetation types at an accuracy commensurate with published guidelines for map accuracy.
• The context classification consistently improved agreements with vegetation by 10% or more over spectral classification alone.
• Using imagery, the number of field sample sites could be reduced to one half of that recommended by survey guidelines while identifying more than 90% of vegetation types present. Further sample reduction to one quarter would identify at least 60% (average of 78%) of vegetation types present.

Sample site selection

• Satellite imagery can be used to map an area into different cover units. The quantitative nature of the digital data allows similar units to be grouped for any chosen level of detail. Within each group, the largest cover unit can be selected to represent that group. Consequently for any number of samples, the largest possible area representing the different cover types present is selected for sampling.
• Within any cover unit, the distribution of spectral classes can be used to identify atypical patches of cover and transects placed to avoid them.

Direct measurement of vegetation cover

• Regressions derived from image bands were shown to have the potential to predict both green and dry pasture cover at sites to within 10% of measured values even though imagery was found to be more sensitive to green cover than dry cover.
• Pasture cover is more likely to be satisfactorily predicted than tree parameters.
• Different vegetation types and measurements taken at different times may result in different regressions. It is important therefore that, prior to making predictions, the vegetation is stratified (mapped) into types that behave similarly with respect to the regressions with imagery.

Benefits of alternative satellite imagery

• There was little difference in the performance of ETM and SPOT imagery with ETM producing slightly better agreements with vegetation.
• There was also little difference in performance between SPOT and IKONOS at Callide, SPOT being slightly better in some circumstances.
• Both ETM and SPOT were found to separate vegetation types substantially better than digital aerial photography at Yarrabee.
• ETM and SPOT imagery costs between $700 and $900 to cover 2,000 square kilometres, while IKONOS costs more than $5,000 for 125 square kilometres. As there were no benefits derived from mapping using IKONOS imagery, ETM or SPOT imagery is more cost effective unless IKONOS is to be used for detailed visual interpretation.

ETM and SPOT imagery costs between $700 and $900 to cover 2,000 square kilometres, while IKONOS costs more than $5,000 for 125 square kilometres. As there were no benefits derived from mapping using IKONOS imagery, ETM or SPOT imagery is more cost effective unless IKONOS is to be used for detailed visual interpretation.