AbstractThe quantification of Above Ground Biomass (AGB) plays a major role in issues related to greenhouse gas emissions and carbon sequestration, pertaining to global warming and the effects of climate change. In Eucalyptus miniata/tetrodonta dominated open-forest in Australia’s northern tropical savanna, AGB mapping is challenging, due to the complex structure of the canopy stand, highly dynamic woody cover, vast spatial extent, vulnerability to climatic effects and the impacts of extensive fire.
Remotely sensed data allow for the mapping, quantification and monitoring of AGB at various scales. Quantification of AGB in tropical savanna by common medium and coarse spatial resolution optical sensor data is inappropriate to monitor finer-grained ecological processes responsible for measuring carbon stocks at an individual tree level and are limited in detecting vertical vegetation structure. There is limited research on the utility of airborne LiDAR (Light Detection and Ranging) and alternative high resolution (< 0.5 m) remote sensing tools for Australian savanna structural assessment. The main goal of this research is to evaluate the efficiency of small footprint airborne LiDAR and determine whether Unmanned Aerial Systems (UAS) and Very High Resolution (VHR) satellite stereo remote sensing data can be used to extract tree biophysical and vertical structural parameters for the purposes of accurately estimating biomass stocks in Australian mesic savannas.
This study utilized a two-phase LiDAR analysis procedure integrating both Individual Tree Detection (ITC) and Area-Based Approaches (ABA) to better understand how the uncertainty of biomass estimation varies with scale. Regression analysis was applied on remote sensing data to develop biomass estimation models based on tree height allometry. This study demonstrated that where field-plot data are spatially limited, it is possible to use a hierarchical integration approach based on AGB uncertainty calculation and calibration to upscale AGB estimates from individual trees to broader landscapes.
Although airborne LiDAR provided higher tree detection rates and accurate estimates of tree aboveground biomass, this research found that a 3D point cloud obtained from light-weight optical UAS imagery by an image dense matching technique is an adequate low-cost alternative for the detection of dominant and co-dominant tree stands, at least at a local scale in Australian tropical savanna.
This study offers some insight into factors causing the poor dense image matching by high-resolution stereo satellites. The structural complexity of Eucalypt crowns, represented by clumped-leaf-grain structure and erectophile foliage, are the main factors determining the efficiency of tree/canopy detection using stereo satellite imagery.
|Date of Award||May 2019|
|Supervisor||Shaun Levick (Supervisor), Stefan Maier (Supervisor), Andrew Edwards (Supervisor) & Jeremy Russell-Smith (Supervisor)|