Fundamental to the philosophy of precision agriculture is the concept of matching inputs to needs. Recent research in precision agriculture has focused on use of Management Zones which are field areas possessing homogeneous attributes in landscape and soil conditions. There are several methods for delineating management zones, depending on the available resources and the characteristics of the field being mapped. Existing traditional clustering techniques do not account for the spatial correlation between observations and take little account of gradual change, either from one class to another or within any one class. Differently, geostatistics treats variables as continua in a joint attribute and geographic space. Therefore, in geostatistical applications clusters are unnecessary, nevertheless in precision farming it may be sensible to divide the field into a restricted number of practical management zones. It then needs to develop an algorithm of clustering that is also spatially constrained, in order to ensure spatial contiguity. The methods based on nonparametric density estimation are the ones which allow clusters of unequal size and dispersion and with highly irregular shapes to be detected (Castrignan: et al., 2006). The objectives of this work are to propose a combined approach to aggregate soil and crop properties into contiguous management zones, based on multivariate geostatistics and a non-parametric density algorithm of clustering, and to use visualization for displaying data and statistical analysis.
A combined approach to delineating management zones for precision agriculture
Pisante, Michele;
2008-01-01
Abstract
Fundamental to the philosophy of precision agriculture is the concept of matching inputs to needs. Recent research in precision agriculture has focused on use of Management Zones which are field areas possessing homogeneous attributes in landscape and soil conditions. There are several methods for delineating management zones, depending on the available resources and the characteristics of the field being mapped. Existing traditional clustering techniques do not account for the spatial correlation between observations and take little account of gradual change, either from one class to another or within any one class. Differently, geostatistics treats variables as continua in a joint attribute and geographic space. Therefore, in geostatistical applications clusters are unnecessary, nevertheless in precision farming it may be sensible to divide the field into a restricted number of practical management zones. It then needs to develop an algorithm of clustering that is also spatially constrained, in order to ensure spatial contiguity. The methods based on nonparametric density estimation are the ones which allow clusters of unequal size and dispersion and with highly irregular shapes to be detected (Castrignan: et al., 2006). The objectives of this work are to propose a combined approach to aggregate soil and crop properties into contiguous management zones, based on multivariate geostatistics and a non-parametric density algorithm of clustering, and to use visualization for displaying data and statistical analysis.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.