A soil temperature decision support tool for agronomic research and management under climate variability: adapting to earlier and more variable planting conditions

Soil temperature is important for agro-environmental processes, including germination, plant growth and soil microbial functioning. Here we develop a soil temperature model and show its use for guiding a soil temperature dependent farming decision – when to sow cotton. We used general additive mixed modelling, which allows for non-linear effects and an extensive soil temperature dataset (245,330 observations from 45 sites) from Australia to build the model. The model uses climatic and environmental data as inputs and had good predictive ability (mean cross-validated R2=0.92, RMSE=1.91, percent bias=−0.01). The preceding day’s maximum temperature and temperature difference (daily maximum – minimum) were the two climate variables most strongly correlated with soil temperature. The model was used to develop a real-time web-based spatial decision support tool to inform cotton planting decisions. We also demonstrate the approaches utility for research, by hindcasting our model using high quality historical climate data and demonstrate significant shifts towards earlier (−11 days) and more variable (standard deviation+3.16 days) planting windows relative to the origins of the Australian modern cotton industry in the 1960s.