The Centre for Remote Sensing conducts cutting-edge interdisciplinary research and provides quality education for future scientists, educators, and decision-makers. Southern Earth Observatory (SEO) is a research and educational collaboration that combines the strengths of the School of Science, RMIT University and Victorian Department of Environment, Land, Water and Planning (DELWP). Our mission is to address fundamental questions about the functioning of the biosphere and its implications for the environment and human welfare in a rapidly-changing world. Our approach focuses on the development and application of remote sensing technologies, including: satellite earth observation, aerial and UAV (RPAS) based hyperspectral and thermal observations, LiDAR, in situ observation systems and associated modelling to measure, map and monitor the biophysical attributes of terrestrial environments. We also specialise in metric measurements from imagery: vision metrology. Through research, education, and communication, we seek to enhance basic understanding of critical earth system processes, develop practical tools for monitoring and predicting change, and facilitate sustainable management of ecosystems, societies, and the biosphere.
Southern Earth Observatory is situated at RMIT University in a purpose built 240m2 laboratory space that houses a team of three post-doctoral fellows, nine PhD scholars and a technical officer. It has access to the latest image processing and photogrammetrical software in a bespoke adjacent advanced imaging laboratory. SEO currently owns and operates two full-range (0.4-2.4um) ASD spectral-radiometers, three aerial camera systems (including a 6-band vegetation camera), a FLIR A325 thermal camera, two UAVs: incl. a DGI S1000 equipped with Tetracam MINI-MCA6, four synchronised Panasonic high speed video cameras, Canon 5D mk-2, Panasonic Lumix GH4, Stereo GoPros). It also operates two Terrestrial Laser Scanning Systems, as well as co-owning (as part of an ARC funded consortium LE0560930) a full waveform airborne LiDAR system. We have also built, from scratch, a network of in situ wireless sensors used for image validation.
• Two Integrating spheres incl. Labsphere HELIOS 20″ Uniform Source for Luminance and Radiance Calibration USS-2000C
• Two ASD full width spectral radiometers (0.4-2.4um)
• Three aerial camera systems including Redlake / Duncantech aerial camera
• FLIR thermal camera A325
• Canon EOS 5D, 20D and Nikon D50 and Calibration Frames
• 9m helium filled remote controlled airship for mobile platform observations
• DGI S1000 & Phantom-II UAS
• 10m hydraulic mast
• Climate controlled chamber for growing botanical samples
• Terrestrial Laser Scanning System (Trimble CX)
• Riegl LMS-Q560 laser scanner aerial LIDAR system (co-owned as part of a consortium)
• 9 dedicated Image processing workstations and two 25 seat laboratories.
• Software: ENVI-IDL, ERDAS IMAGINE, Intergraph Image station, Definiens-eCognition, Terrascan, Matlab, Agisoft
Landsat for Sustainable Forests (LandFor) CRCSI Project 4.104
Victoria’s Department of Environment, Land Water and Planning (DELWP) has built Australia’s only state level public land based forest inventory via the Victorian Forest Monitoring Program (VFMP). Initiated in 2010, the VFMP provides baseline data for long term trend detection and prediction of future changes, supporting statutory reporting obligations as well as forest policy and management decisions, and Victoria’s performance towards sustainable forest management. It was built on an extensive network of plots and includes three tiers of data collection: ground plot information, aerial imagery, and satellite imagery. Ground and photo plots, which represent less than 1% of the public forest estate, are revisited every 5 years; while satellite imagery are used to support estimation processes. The Landsat for Sustainable Forests (LandFor) project enhances the VFMP and supports its strategic development by using the freely available Landsat image archive to build an annual stable image base and create an archive of disturbance events (e.g., logging, fire) that allows trends to be consistently detected and tracked for the last 30 years. By integrating LandFor outputs with ground based data, LandFor is building a toolkit that allows annual state-wide VFMP maps to be generated, showcased through an exemplar variable (biomass) that is validated with LiDAR data.
Fuels3D is a new tool for capturing information about fuel hazard in sensitive and fire-prone areas. Fuels3D combines everyday smartphone technology with the latest cloud-based 3D analytics, to generate real-time information about the level of fuel hazard. Monitoring and quantifying the level of fuel hazard in our natural environment has critical implications for safety, biodiversity, and resilience. For example, information about fuel hazard is fundamental to understanding the success of fuel reduction interventions. Conventional approaches to monitoring fuel hazard either require high levels of human expertise and high costs (such as manual visual assessments and 3D laser-scanning) or are unsuited to small-scale assessments in canopy-covered areas (such as airborne or satellite remote sensing). In contrast, Fuels3D enables non-expert users to make accurate, low-cost, quantitative, and real-time fuel hazard assessments via their smartphone.
The Fuels3D project team of Dr Karin Reinke, Dr Luke Wallace, Prof. Simon Jones, Samuel Hillman, Daisy San Martin, Christine Spits and Bryan Hally was recently awarded the Victorian Spatial Excellence Award for Environment and Sustainability.
Fire surveillance and hazard mapping
The study has been improving the accuracy of vegetation monitoring for flammability, as well as saving critical man hours, through the development of a beta smartphone application (Fuels3D).
Outcomes will enable satellite measures of fire activity to be made, which in turn have the potential to inform or support efforts in bushfire response planning and fire rehabilitation efforts. A particular focus is on the analysis of data obtained from Himawari-8, which is able to provide updated imagery on a 10 minute basis.
The project is currently using simulations and real world experiments to determine the accuracy with which fires can be detected, their temperature and shape determined, for a range of landscapes.
The project is also creating new techniques and protocols for the rapid attribution of fire landscapes (pre- and post-fire). These techniques seek to add quantitative vigour to existing fuel hazard estimation practices.