Precision Agriculture, 2, 333–346, 2000
© 2001 Kluwer Academic Publishers. Manufactured in The Netherlands.
Use of Wavelets to Compare Simulated Yield
Patterns for Precision Agriculture at the
A. (JAN) VERHAGEN firstname.lastname@example.org
Plant Research International, PO Box 16, 6700 AA Wageningen, The Netherlands
Wageningen University, Mathematical and Statistical Methods Group, PO Box 9101, 6700 HB Wageningen,
Wageningen University, Dept. of Environmental Sciences, PO Box 37, 6700 AA Wageningen, The Netherlands
Abstract. In this paper spatial patterns ofsimulated water limited potato production for a farm ﬁeld in the
Netherlands are analyzed using wavelets. The simulated yield patterns are decomposed using wavelets into
crystals with a varying resolution. These crystals are used to relate the simulated spatial patterns to weather
conditions. We compare simulated yield patterns from 10 successive years to arrive at a prototype patterns
that can be used as a basis for site-speciﬁc management. Typical patterns for dry and wet years are quantiﬁed
using wavelets. Wavelets provide a mechanism to distinguish between spatial patterns and allow a quantitative
approach for classiﬁcation of these spatial patterns.
Keywords: patterns, wavelets, simulation, geostatistics, precision farming
When dealing with precision agriculture soil variability is a key element. Soil and yield
maps are already widely used to guide farm management at the sub-ﬁeld level. Sensors
either mounted on farm equipment or airborne will in the near future provide additional
layers of spatial information. The information obtained using such techniques reveals
the integrated effect of the physical, chemical and biological processes under speciﬁc
weather conditions for one speciﬁc crop. When reacting to variable conditions in farm
management we need to distinguish between the different sources of variability. Apart
from understanding and quantifying the processes a need for pattern comparison tech-
niques is obvious.
In this paper we use a dynamic simulation model to calculate water limited crop pro-
duction ignoring yield reduction resulting from nutrients, pests and diseases. Generated
yield patterns are the result ofcrop-soil interactions related to weather variability or,
more speciﬁcally, rainfall variability. General patterns ofsoil variable are ofa much
larger explanatory power than ordinary, point tied comparisons.