The Agriculture Data Science research group focuses on creating opportunities for innovation in big data to support the global efforts towards sustainable intensification of agriculture.
Modelling, machine learning and meta-analysis
To achieve efficient agricultural nitrogen (N) management, our group applies process-based agroecosystem simulation models to simulate, predict and evaluate the impacts of management practices on N losses through different pathways, thereby developing decision-support tools. More recently, we integrated agricultural big data with machine learning to revamp existing parameters of agroecosystem models and improve the reliability of model prediction in N processes. We have also applied meta-analytic techniques to national and international datasets on soil-plant carbon and N dynamics in agricultural systems and their mitigation and adaptation to climate change.
Sustainable nitrogen indices
Australia has a reputation globally for production of food that is ‘clean and green’, but this branding is neither meaningfully defined nor readily verified. To substantiate this claim scientifically, our group has ongoing research, integrating vast datasets and resource-use models, for building the world’s first evidence-based N indices for (Australian) agricultural products that are embedded with an environmental cost-benefit analysis. The indices will help consumers to choose food products with low environmental footprints, incentivise farmers to adopt more sustainable N management practices and benchmark Australian agricultural production against international practices. Our group has also developed anew indicator, reactive nitrogen spatial intensity (NrSI), to identify Nr emission hotspots, indicate the potential environmental impacts of Nr, and assist management recommendations.
Remote sensing, GIS and precision agriculture
Rapidly advancing sensor systems on ground, air and space-borne platforms delivering hyperspectral and thermal remote sensing imagery at high spatio-temporal resolutions provide massive volumes of data to be analysed in novel ways, unveiling detailed insights into crop performance and their variability across large areas. Simultaneously, modern agricultural production systems generate large volumes of spatially referenced information such as yield, grain quality, soil moisture and weather data. This represents rich opportunities to develop smart farm systems that will allow growers make increasingly precise decisions to maximize yield potential while optimizing water and nutrients. Our group is working on developing new techniques and algorithms using hyperspectral and thermal remote sensing technology to detect biotic and abiotic crop stress under drought, heat, frost and nutrient deficiencies before symptoms are visible.