ABSTRACT
Growing water-scarcity under climate change makes cost-effective water saving interventions in agriculture crucial for sustaining crop yields. Forecasting crop canopy temperature (Tc) is a widely used, cost-effective method of precision irrigation applied to over 60% of Australian cotton farms and across six U.S. cotton belts. Tc forecasting uses in-situ sensors to predict the number of hours Tc exceeds optimal plant temperature (stress-hours) to plan irrigation up to 7 days in advance.

Despite its widespread adoption, the industry forecasting model is understudied, lacking crucial time-series methodology and stress-hour forecast evaluation for limited locations. Evaluating newer Tc and time-series ML models across 3 irrigation treatments shows improvements in elevated Tc and stress-hour forecasts by 30%. Improving and expanding forecasting on 1000 in-situ Tc sensors (~20 fold data) shows a consistent increase of 12% in forecasting. In addition, methods for forecasting Tc sensors at unobserved locations were developed, improving 20% over current methods, and enabling cost-effective Tc management by filling in locations without sensors or those with missing data. These improvements across Australia and the USA are comparable to a decade's improvement in Bureau of Meteorology air-temperature forecasting. 

Developed `point' forecasts are problematic as the placement of the in-situ Tc sensor on the farm may change the total number of irrigations recommended. To improve this, a spatial forecasting model utilising 3 in-situ sensors and remote-sensing Tc images is developed. This UAV-linear model can produce spatial forecasts on a single farm as effectively as if each location on the farm has an in-situ sensor monitoring it. 

These results present significant forecasting developments for canopy temperature, covering point, national and spatial scales. Applying these results for current Australian and USA markets can largely improve accuracy, cost and effectiveness of Tc forecasting management, leading to significant water-and-cost savings under the growing threat of climate change. 

BIOGRAPHY
I am a PhD student working jointly across the Australian National University (ANU) and CSIRO since 2023. I completed a Bachelor of Philosophy (Science) at ANU in 2023 and commenced the Plant Science PhD program in the same year.

Working across ANU plant science and Data61 at CSIRO has provided exposure to experts across both fields providing invaluable experience across disciplines. I have enjoyed my PhD studies and look forward to completing it and future research.