A Method to Investigate Within‐Field Variation of the Response of Combinable Crops to an Input

Precision agriculture is based on the hypothesis that the optimum rate of inputs to a crop varies spatially within fields. Evidence for this hypothesis is scarce due to the practical and theoretical difficulties of designing appropriate experiments. This paper proposes a procedure for testing the hypothesis of precision agriculture for crops that may be harvested with a combine harvester equipped with a yield monitor. An input is applied according to a randomized block design. Yield monitor data may be treated as a convolution of yield with a function that characterizes the smoothing effect of processes in the combine on the mass flow rate at the sensor. The input rates, determined by the experimental design, are transformed using the combine's smoothing function and a preselected yield response function. The parameters of the response function for the whole field or a local neighborhood can be estimated from these transformed rates and the yield monitor data. A null hypothesis, that the spatial variation in one of these parameters (that determines the local optimum rate) may be attributed to random yield variation about a uniform response function, may be tested. A wheat crop ( Triticum aestivum cv. Consort) was treated with varying rates of N fertilizer in a case study in the south of England. Analysis of the yield data showed that the observed variation in the response could not be explained as random fluctuation around the field‐scale response function. The economic optimum rate of N varied from zero to greater than 200 kg ha −1