Using aerial video data to verify satellite data
The spatial complexity of the large areas analysed within Region 102 meant that it was very difficult to check the results of grazing gradient analyses using conventional ground-based measurement. Instead, the research team tested the feasibility of using airborne videography to validate results obtained using satellite imagery. A four-camera airborne video system was used to collect data in the visible and near-infrared spectrum. [Details of this work are in the aerial video section of the CD-ROM.]. Click on the example images below to see full-size
 |
 |
 |
 |
| (a) 20cm resolution | (b) 1 metre resolution | (c) 1 metre resolution | (d) classified frame |
There were two main tasks involved with checking grazing gradient analyses:
1. Relating ground measurements of vegetation cover to the aerial video data. To do this, we needed good information about vegetation present in the video images. Images at 20 cm pixel resolution provided this level of detail (example (a) above). There was good agreement between cover levels obtained from classifying the video data (example (d) above) and ground-based measures of the same vegetation components.
2. We then classified a transect of video images covering 100 ha into components of vegetation cover in order to calibrate the PD54 index derived from Landsat TM data. Before doing this, it was necessary to spatially register the video images acquired at 1 m resolution (example (b) above) to the satellite data. There was good agreement in cover levels obtained from both the classified video data and TM-derived PD54 index. Based on this agreement, we used the video data to convert the PD54 index to percentage cover.
Conclusion: The results show that video data, when classified into components of vegetation cover, relate well to ground-based measures of cover. Further, the video data can be used to verify the index of vegetation cover calculated from Landsat TM data by converting the index to percent cover of vegetation.
