Estimating pasture biomass with Planet Labs CubeSats

Dairy farms require efficient monitoring of pasture biomass. However, physical proxy or destructive measurements are time-consuming, subject to sampling and collection error, and cannot characterise spatial biomass within entire paddocks. Until recently, spatial resolution and temporal frequency of satellite data have been insufficient to achieve effective management. We developed a model for biomass prediction based on data from new Planet Labs satellites to evaluate its utility in the context of commercial dairy farming. While the model had minimal residual error on the farm with the highest quality data (RMSE 260 kg/ha), validation of the model on new farms was poor. Accurately estimating biomass for the calibration farm is promising, but further work is necessary to develop a model which extrapolates robustly.