Precision Agriculture, 4, 193±201, 2003
#
2003 Kluwer Academic Publishers. Manufactured in The Netherlands.
Is the Soil Variability within the Small Fields of
Flanders Structured Enough to Allow Precision
Agriculture?
MARC VAN MEIRVENNE Marc.VanMeirvenne@ugent.be
Department of Soil Management and Soil Care, Ghent University, Coupure 653, 9000, Gent, Belgium
Abstract. The average area of agricultural fields in Flanders (Belgium) is about 1.7 ha, being very small
compared to fields where precision agriculture is currently applied. Therefore this paper addresses the question
whether the within-field variation of soil properties in such fields is structured enough to motivate precision
agriculture. To answer this question, 9 soil properties determined on 380 soil samples located in 77 agricultural
fields situated in the 5 most dominant pedoscapes of Flanders were used to analyze their spatial variation over
intervals ranging from 5 to 900 m. The data set was subjected to a principal component analysis which identified
two principal components (PCs) explaining more than 78% of the total variance. The first PC represented the
chemical soil properties and the second the physical and biological properties. A variogram analysis of the
scores on these two PCs showed that the micro-scale and random variation dominated (82%) the within-field
variability of the first PC. The within-field variability of the second PC was dominantly spatially structured
(only 37% micro-scale and random variation). Therefore, it was concluded that mainly for soil physical and
biological properties (like soil textural fractions and organic matter), the average within-field variation in the
small fields of the investigated landscapes is structured enough to allow precision agriculture.
Keywords: within-field variability, precision agriculture, variogram analysis, geostatistics
Introduction
Precision agriculture (PA), being the adaptation of management to site specific
conditions, has triggered new attention to soil spatial variation since this is considered
to be the key element to its successful implementation (Robert, 1999; Verhagen and
Bouma, 1997). In particular the within-field scale is of importance for PA since
traditional agriculture focuses mainly on the between-field variation of yield controlling
properties. However, within-field scale does not define a particular order of dimension
since the size of agricultural fields can vary considerably and it will be clear that large
fields could be expected to benefit more from PA than small fields. Moreover, it is not
sufficient to encounter an important within-field variation to motivate PA. This varia-
tion must be spatially structured to allow accurate mapping. Micro-scale and random
variation cannot be mapped, they just add uncertainty to the cartographic information.
Geostatistical tools, like variogram analysis, allow differentiation between structured
variation and micro-scale and/or random variation (Cressie, 1991). Consequently these
tools have been used intensively to map soil properties to guide PA. Mostly however,
only a few soil properties sampled within one, or a few neighboring, fields have been
considered (e.g. Geypens et al., 1999; Mulla, 1993). Therefore, these results apply only