Mr Qazi Muqeet Amir1, Prof. Thomas Bishop1
1Precision Agriculture Laboratory, Sydney Institute of Agriculture, School of Life and Environmental Sciences, Faculty of Science, The University of Sydney, Australia, 2School of Computer Science, The University of Sydney, Australia
Biography:
Qazi Muqeet Amir is a first year PhD candidate at the University of Sydney. Muqeet’s research is focused on improving soil moisture nowcasting and forecasting capabilities in Australian agricultural fields.
Abstract:
A key issue in dryland agriculture is determining inputs and managing risk to optimise yield productivity. Soil water availability is a critical constraint in yield productivity as crops are highly dependent on water available to them in the root zone. Understanding and forecasting soil moisture in a paddock can provide growers with crucial information that can optimise and secure their yield for a season. Forecasts of soil moisture can inform decisions related to irrigation, fertiliser input, and expected yield for the growing season.
This project presents the temporal forecastability of soil moisture across Australian dryland fields. Forecastability is the degree to which soil moisture can be forecasted with respect to accuracy and lead time. The forecastability is understood through the implementation of conventional machine learning and novel deep learning methods. Soil moisture data is sourced from various probe networks, such as the CosmOz, VicAg, and OzNet networks, which span across Australia and represent its diverse soil types and climatic conditions. The probe data is available at multiple depths in the soil profile and along with various in-situ and remote sensed meteorological data, is used to forecast soil moisture.
A key outcome of this study is identifying patterns in forecast accuracy and predictor importance with respect to region, soil type, weather conditions, and season. These patterns create a bigger picture of soil moisture forecastability and the potential for forecasting in areas with no soil moisture probe data available. With a clearer understanding of forecastability, higher accuracy forecasts can be conducted for growers in Australia, which will allow them to adjust growing practices and improve estimates of seasonal yield potential.