Servicios Personalizados
Revista
Articulo
Indicadores
- Citado por SciELO
Links relacionados
- Similares en SciELO
Compartir
RIA. Revista de investigaciones agropecuarias
versión On-line ISSN 1669-2314
Resumen
CORDOBA, M; BRUNO, C; COSTA, J y BALZARINI, M. Variabilidad espacial de suelo a escala de lote y su relación con los rendimientos. RIA. Rev. investig. agropecu. [online]. 2016, vol.42, n.1, pp.47-53. ISSN 1669-2314.
Site-specific management requires delineation of homogeneous zones within the field. Several variables, such as some soil properties, are used for zonification. Fuzzy k-means cluster analysis (FKC) is here used to delimit zones. FKC is applied to original variables and to synthetic variables obtained with regular principal component analysis (PCA). However, PCA does not consider the presence of spatial correlations. We propose to use, MULTISPATI-PCA as an extension of PCA that considers spatial information. The method is also used in a canonical correlation analysis to quantify the magnitude of the linear relationship between crop yields and soil variables. In this paper, we evaluate the capacity of five multivariate procedures to delineate zones: FKC on soil variables, FKC on principal components and FKC on spatial principal components. Finally, we include field-site partitions based on percentiles of canonical variables that correlate yields with principal components or spatial principal components, alternatively. Yield differences between the delineated zones by each method were compared. We worked with apparent electrical conductivity data in two depths 0-30 cm and 0-90 cm, elevation, hardpan depth and soybean and wheat yields. Cluster analysis on spatial principal components, was the best procedure to delineate zones.
Palabras clave : Spatial data; Principal components; MULTISPATI-PCA.