Geostatistical assessment of the spatial variability of soil texture in the coffee plantation

Geostatistical assessment of the spatial variability of soil texture in the coffee plantation
Marcelly da Silva Sampaio1, Marcelo de Carvalho Alves1, Fábio Moreira da Silva2, Edson Ampélio Pozza2, Marcelo Silva de Oliveira2, Luciana Sanches1

1 Universidade Federal de Mato Grosso, Cuiabá, MT – Brasil (arcellysampaio@gmail.com, marcelocarvalhoalves@gmail.com, lsanches@ufmt.br)
2 Universidade Federal de Lavras, Lavras, MG – Brasil (fmsilva@ufla.br, eapozza@ufla.com.br, marcelo.oliveira@ufla.br)

Abstract: The aim of this study was to evaluate the variability of soil texture in coffee plantation by geostatistics, using simple kriging as linear interpolator to analyze the prediction errors. The research was conducted in Cafua Farm where 67 sample points were collected in an area of 6.5 ha. Spherical model variograms were adjusted by the methods of ordinary least squares (OLS), weighted least squares (WLS), maximum likelihood (ML) and restricted maximum likelihood (REML) for the data of clay, sand and silt. The best method of assessment was performed using the Akaike information criterion, where the REML method showed better performance for all attributes. The simple kriging was done for the data and for the standard deviation of the data. It was observed by kriging that the highest percentage of clay and silt were located in the northern area, while the highest percentage of sand concentrate in the south. Analyzing the standard deviation, data higher variation was found in clay, especially in places with sample deficiency. The range of spatial dependence of clay, sand and silt were 143.7m, 245.2m and 141.2m, respectively. The minor variations of the standard deviation are associated with the sampling points.

Keywords: simple kriging, Akaike information criterion, spherical model, prediction errors.

AttachmentSize
SampaioAccuracy2012.pdf165.92 KB