Australia and South Australia’s push into SPACE provides opportunities to develop Earth Observation sensor componentry, image processing algorithms and rapid delivery of information to end users. This presentation explores these opportunities in two application areas: forest fire detection and forest harvest monitoring. The former application is extremely time critical and the latter application requires automation of the processing of significant quantities of satellite image data. Both applications require multi-disciplinary professional teams to generate innovative solutions.
Australia and South Australia’s push into SPACE provides opportunities to develop Earth Observation sensor componentry, image processing algorithms and rapid delivery of information to end users. This presentation explores these opportunities in two application areas: forest fire detection and forest harvest monitoring.
In forest fire detection the critical issues are a) early, within 15-minute, detection of fire commencement and the transmission of fire location to a local fire management agency, and b) the development of a sensor and processing algorithms with sufficient detectivity of the indicators of fire (smoke, rising land surface temperature and flame visibility). Research at the University of SA by PhD student Parwati Sofan has provided the satellite based answers to some of these issues, but additional opportunities exist to expand this work leading to a system that can be implemented globally. For the time constraints of this application to be met the ideal solution is on-board satellite processing of images with downlink of potential fire location direct to fire management agencies. On-board satellite processing generates additional issues related to compute power, electrical power budget and thermal management, as well as information transmission via either relay to geostationary satellites, or direct downlink to Earth based tracking antennae.
In the second application, public and private managers of forests across Australia and New Zealand seek a web-based delivery tool that will provide monitoring, on a monthly basis, of three forest states (cleared, juvenile and mature). Initial research at University of South Australia by David Bruce has shown that existing and free near polar orbiting satellite imagery from ESA’s Sentinel 1 and 2 satellites provide an adequate source of data at a reasonable temporal resolution suitable to generate moderately accurate results for this application. The quantities of data to download, pre-process (orthorectify, radiometrically calibrate, de-noise and de-speckle), process, generalise and distribute via a web-based tool, is a significant computing task which requires automation, potentially through the application of artificial intelligence (AI). In this work the change in the forest state can be detected in both optical and radar imagery more sensitively than absolute classification of satellite images. This requires all images to be well ortho-rectified and this still remains a challenge for the C Band SAR images from Sentinel 1.
The above applications require multi-disciplinary teams from a variety of professional backgrounds including sensor technology, remote sensing, AI, Wed IT, electrical and telecommunication engineering – just the sort of teams South Australia should be assembling to take advantage of the focus on SPACE that is currently being experienced in this state.
Time & room
2.30 pm–2.45 pm in City Room 3
Prof David Bruce
University of South Australia; Flinders University; International Space University
Date & venue
Friday, 25 October 2019 at the Adelaide Convention Centre