Machine learning has achieved great success in computer vision, natural language processing and other fields, especially in the accuracy of prediction and some may have exceeded the human capabilities. Nevertheless, users still need to understand the reasons for their conclusions in a more detailed and tangible way in application scenarios. Giving strong explainability to the model also helps to ensure its robustness, and usability of the method. This proposal focuses on developing interpretable models for early smoke wild-fire detection from satellite images.

Detecting fires at their early stages is essential to prevent fire caused disasters. Research has been conducted to detect smoke in satellite imagery for fire detection. Unfortunately, the imagery data used in previous research have low spatial resolution and only contain the RGB bands, which are ineffective for early fire detection. Our team (Data Analytics Group at UniSA) has been working on early fire smoke detection with multispectral multi-sensor satellite imagery for one year and the accuracy of wild fire detection can reach more than 90% accuracy. An AI framework of deep learning neural networks that identifies wild fires has been developed. It is necessary to present users with understandable reasons for the detection so the users can validate the detection and assess its severity. Current detection models are deep learning-based and have a black box detection kernel. This project aims to make the detection transparent to users so users can use and interact with the model easily. In the following, we will introduce the main techniques for explaining predictions made by deep learning / black-box
machine learning models.

The objective of this project is to assess the suitability of the Canadian WildFireSat mission for Australian Bushfire Management.

The key innovation of this project is the development of robust methods for the integration of radar-based EO data into current and emerging systems for monitoring the impact of fire on vegetation. Rapid fire extent mapping, including fire progression of large wildfires, will be based on dense time-series of synthetic aperture radar (SAR), optical data and machine learning. The research will also explore the capabilities of SAR and LiDAR data, integrated with optical data, for distinguishing structural characteristics of post-fire recovery dynamics.

The overarching focus of the research is on the integration of multi-sensor EO data to fill key gaps in operational monitoring of the impacts of wildfire. The project aims to support land and fire managers to make more informed decisions, by developing more accurate and timely measures of burnt area extent and tools for monitoring post-fire recovery.

Flooding is a common and extremely impactful event within Australia and around the world. For example, the March 2021 Australian floods are a series of floods that began from 18 March 2021 which have affected New South Wales, from the North Coast to the Sydney metropolitan area in the south, in a disaster described as a “one-in-100-year event”. Additionally, far-south and far-southeast communities in Queensland were also greatly affected by flooding and heavy rainfall.

The aim of this project is to develop and operationalise smart analysis of SAR and optical satellite imagery (primarily NovaSAR and Sentinel missions) to address time-critical applications such as flood mapping (2D) and floodplain water harvesting (3D), based on many years of research in this area by the project team since 2009. Project activities include feasibility studies, remote sensing software
development (analytic toolbox) and extensive case studies.

The expected outcomes are:

1. A suite of near real-time, cross platform, scalable and operational tools for mapping floods with satellite remote sensing, ready for flood management agencies to takeover and/or private sector to commercialise, and improve volume estimate during the floodplain harvesting event for the Murray-Darling Basin states;
2. A comprehensive report on feasibility studies to inform a Phase 2 project; and
3. A comprehensive report on the case studies, targeting a range of users and promoting SmartSat CRC research through the mass media. The project brings together core partners such as UNSW, NSW Department of Planning and Environment, and Geoplex / Nova Systems, an ideal mix of academia, end user and geospatial service provider. The proposed project has also attracted strong support from the Federal Department of Agriculture, Water and Environment (DAWE) because of its significant national benefits (Letter of Support attached), as well as other key players such as Airbus and Geospatial Intelligence Pty Ltd.

The aim of this research is to improve and advance upon signal processing algorithms and functions developed for geostationary satellites. The results obtained will be used to improve the positional detection and tracking of fire, and its spatial characterisation, to inform emergency services where to allocate and deploy fire combat resources. This final output is aligned with SmartSat’s “Disaster and Emergency Management” application area.

The proposed research project “Monitoring changes in water quality in response to landcover disturbance with Earth Observations in Australia” aims to develop the methods to link in situ and satellite information to monitoring transport processes and recovery of aquatic ecosystems, namely inland waterbodies, after landcover disturbance events to further understand causal relationships for the purposes of reducing the impact on protected water bodies against future natural disasters for Australia.