Propagation of Positional Measurement Errors to Field Operations

Abstract

Precision agriculture requires detailed information about numerous factors that determine optimal agricultural management. Recently, the importance of accurate assessment of field boundaries for precision agriculture has also been recognised. It has been claimed that the upcoming targeted approach to managing field operations requires that field boundaries are measured with cm level accuracy, thus avoiding losses such as wasted inputs, unharvested crops and inefficient use of the area. The aim of this work is to develop a method for experimental verification of such claims. The method comprises three main steps. The first step concerns the choice and parameterisation of a geostatistical model defining positional error. Secondly, the model is used to generate a set of possible realisations of field geometry. Next, these realisations are the input to a model computing the effects of positional uncertainty on field operations. The output is compared with that obtained using reference geometry. We illustrate the method using, amongst others, data from the 2005 GPS workshop of the EU Joint Research Centre. Our implementation of the model uses the Data Uncertainty Engine (DUE), which is free software that aids the user in defining probability distributions for uncertain spatial objects and draws random samples from these distributions. The application concerns an irregularly shaped field of approximately 15 ha. Results for three different measurement scenarios (Egnos augmented GPS, RTK-GPS, topographic map) are shown.