Variability and uncertainty in spatial, temporal and spatiotemporal crop-yield and related data

Application of the theories of precision agriculture to the practicalities of broad-acre farming relies on successful handling of the ramifications of uncertainty in information, i.e. information pertaining to the spatial and temporal variation of those factors which determine yield components and/or environmental losses. This paper discusses the uncertainty of yield and related variables as measured by their spatial and temporal variance. The magnitude of these two components gives a suggestion as to the appropriate scale of management. Simultaneous reporting on spatial and temporal variation is rare and the theory of these types of process is still in its infancy. Some brief theory is presented, followed by several examples from the Rothamsted classic experiments, yield-monitoring experiments in Australia, a long-term barley trial in Denmark, and a soil moisture monitoring network. It is clear that annual temporal variation is much larger than the spatial variation within single fields. This leads to the conclusion that if precision agriculture is to have a sound scientific basis and ultimately a practical outcome then the null hypothesis that still remains to be seriously researched is: 'given the large temporal variation in yields relative to the scale of a single field, then the optimal risk aversion strategy is uniform management.'