Spatial Prediction of Soil Salinity Using Electromagnetic Induction Techniques: 2. An Efficient Spatial Sampling Algorithm Suitable for Multiple Linear Regression Model Identification and Estimation

Abstract

In our companion paper we described a regression-based statistical methodology for predicting field scale salinity (ECe) patterns from rapidly acquired electromagnetic induction (ECa) measurements. This technique used multiple linear regression (MLR) models to construct both point and conditional probability estimates of soil salinity from ECa survey data. In this paper we introduce a spatial site selection algorithm designed to identify a minimal number of calibration sites for MLR model estimation. The algorithm selects sites that are spatially representative of the entire survey area and simultaneously facilitate the accurate estimation of model parameters. Additionally, we introduce two statistical criteria that are useful for selecting optimal MLR variable combinations, describe a technique for identifying faulty signal data, and explore some of the differences between our recommended model-based sampling plan are some more commonly used design-based sampling plans. Survey data from two of the fields analyzed in the previous paper are used to demonstrate these techniques. able selection criteria that are useful for selecting optimal combinations of regression variables in the MLR model and show how the principal components analysis can be used to detect faulty signal data. We demonstrate these different techniques using the salinity survey and calibration data already introduced in the previous paper.The details concerning our analysis are, by necessity, specific to the MLR salinity modeling approach already described. However, the techniques we employ are quite general and 387

Apparent soil electrical conductivity mapping as an agricultural management tool in arid zone soils

“…The statistical approach relies on targeted sampling strategies and either geostatistical or spatial regression calibration models (Lesch et al, 1995a(Lesch et al, , 1995bLesch, 2005). This approach can also be effectively used to estimate other soil properties from EC a survey data, provided that calibration soil samples are acquired and the survey data are found to be well correlated with the target soil property of interest.…”

Mapping clay content variation using electromagnetic induction techniques

“…The sampling approach discussed in Lesch (2005) and Lesch et al (1995b) is specifically designed for use with ground-based EM signal readings. In this model-based sampling approach, a minimum set of calibration samples are selected based on the observed magnitudes and spatial locations of the EC a data, with the explicit goal of optimizing the estimation of a regression model (i.e., minimizing the mean square prediction errors produced by the calibration function).…”

“…The basis for this sampling approach stems directly from traditional response surface sampling methodology (Box and Draper, 1987). Due to this direct relationship, Lesch et al (1995b), referred to this site selection process as a "spatial response surface sampling" design.…”

Método do hipercubo latino condicionado para a amostragem de solos na presença de covariáveis ambientais visando o mapeamento digital de solos

“…Diferentemente do mapeamento convencional, no mapeamento digital de solos muita atenção tem sido dada para otimização da amostragem, utilizando diferentes métodos como: amostragem aleatória (Gessler et al, 1995;Howell et al, 2007); amostragem aleatória estratificada (McKenzie & Ryan, 1999;Roecker & Thompson, 2010); análise multivariada (Lesch et al, 1995;Hengl et al, 2003); lógica fuzzy (Brus et al, 2007;Zhu et al, 2008); e geoestatística (Brus & Heuvelink, 2007;Vašát et al, 2010).…”

“…Para tanto, esses autores propuseram o método cLHS, que é de uso livre, e que se caracteriza como ferramenta robusta para a alocação de pontos amostrais pela utilização de um conjunto de covariáveis contínuas e, ou, categóricas. O cLHS utiliza como base o método de amostragem do hipercubo latino (LHS, sigla em inglês), proposto por McKay et al (1979). Minasny & McBratney (2006) compararam o método cLHS com a amostragem aleatória e a aleatória estratificada (Brus et al, 2007) e demonstraram que sua utilização com as covariáveis declividade, índice topográfico combinado (CTI, sigla em inglês), NDVI (índice de vegetação por diferença normalizada, sigla em inglês) e uso do solo representou a distribuição dessas covariáveis, com um número de amostras relativamente pequeno, melhor do que os outros dois métodos utilizados.…”