Determination of soil type boundaries using electromagnetic induction scanning techniques

Information on the spatial distribution of soil texture and soil water is vital in understanding crop yield variation. Such information allows improved management of all agricultural inputs. One of the limiting factors in the mapping of soil texture information, however, is cost. Confusion matrix analysis was used to determine whether bulk apparent electrical conductivity (ECa) data derived from electro-magnetic induction (EMI) scanning at field capacity, and clustered using a k-means algorithm, accurately delineates soil textural boundaries in a field containing clay loam and sandy loam soils. The ECa map was compared to two soil surveys, the first conducted at one sample per hectare and the second at four to eight samples per hectare. Using confusion matrix analysis a significantly stronger relationship was measured between the ECa map and soil units of the more intensive soil map, than with the lower sampling density soil survey. This relationship was between two classes of soil with a difference in clay content of 12% and two clusters with a difference in mean ECa of 16(.)9mSm(-1). (C) 2003 Silsoe Research Institute. All rights reserved Published by Elsevier Ltd.