Enhanced understanding of forest stocks and dynamics can be gained through improved forest measurement, which is required to assist with sustainable forest management decisions, meet Australian and international reporting needs, and improve research efforts to better respond to a changing climate. Integrated sampling schemes that utilise a multi-scale approach, with a range of data sourced from both field and remote sensing, have been identified as a way to generate the required forest information. Given the multi-scale approach proposed by these schemes, it is important to understand how scale potentially affects the interpretation and reporting of forest from a range of data.
¶
To provide improved forest assessment at a range of scales, this research has developed a strategy for facilitating tree and stand level retrieval of structural attributes within an integrated multi-scale analysis framework. The research investigated the use of fine-scale (~1m) airborne Light Detection and Ranging (LiDAR) data (1,125 ha in central Queensland, and 60,000 ha in NE Victoria) to calibrate other remotely sensed data at the two study sites. The strategy refines forest structure mapping through three-dimensional (3D) modelling combined with empirical relationships, allowing improved estimation of maximum and predominant height, as well as foliage and crown cover at multiple scales. Tree stems (including those in the sub-canopy) were located using a height scaled crown openness index (HSCOI), which integrated the 3D density of canopy elements within the vertical profile into a two-dimensional spatial layer. The HSCOI modelling also facilitated the reconstruction of the 3D distribution of foliage and branches (of varying size and orientation) within the forest volume.
¶
Comparisons between forests at the Queensland and NE Victorian study sites indicated that accurate and consistent retrieval of cover and height metrics could be achieved at multiple scales, with the algorithms applicable for semi-automated use in other forests with similar structure. This information has facilitated interpretation and evaluation of Landsat imagery and ICESat satellite laser data for forest height and canopy cover retrieval. The development of a forest cover translation matrix allows a range of data and metrics to be compared at the plot scale, and has initiated the development of continuous transfer functions between the metrics and datasets. These data have been used subsequently to support interpretation of SAR data, by providing valuable input to 2D and 3D radar simulation models. Scale effects have been identified as being significant enough to influence national forest class reporting in more heterogeneous forests, thus allowing the most appropriate use and integration of remote sensed data at a range of scales. An empirically based forest minimum mapping area of 1 ha for reporting is suggested. The research has concluded that LiDAR can provide calibration information just as detailed and possibly more accurately than field measurements for many required forest attributes. Therefore the use of LiDAR data offers a unique opportunity to bridge the gap between accurate field plot structural information and stand to landscape scale sampling, to provide enhanced forest assessment in Australia.
| Identifer | oai:union.ndltd.org:ADTP/204843 |
| Date | January 2008 |
| Creators | Lee, Alex C., alexanderlee@aapt.net.au |
| Publisher | The Australian National University. Fenner School of Environment and Society |
| Source Sets | Australiasian Digital Theses Program |
| Language | English |
| Detected Language | English |
| Rights | http://www.anu.edu.au/legal/copyrit.html), Copyright Alexander C. Lee |
Page generated in 0.0026 seconds