Spatial prediction of soil salinity using electromagnetic induction techniques. 1. Statistical prediction models: A comparison of multiple linear regression and cokriging

• Published in: Water resources research 1995-02, Vol.31 (2), p.373-386
• Main Authors: Lesch, Scott M., Strauss, David J., Rhoades, James D.
• Format: Article
• Language: English
• Subjects: agricultural soils; analisis del suelo; analisis estadistico; analyse de sol; analyse statistique; california; californie; conductividad electrica; conductivite electrique; electrical conductivity; encuestas; enquete; forecasting; measurement; medicion; mesure; propiedades fisico quimicas suelo; propriete physicochimique du sol; salinidad; salinite; salinity; soil chemicophysical properties; soil testing; sol agricole; statistical analysis; suelos agricolas; surveys; technique de prevision; tecnicas de prediccion
• ISSN: 0043-1397; 1944-7973
• DOI: 10.1029/94WR02179

We describe a regression-based statistical methodology suitable for predicting field scale spatial salinity (ECe) conditions from rapidly acquired electromagnetic induction (ECa) data. This technique uses multiple linear regression (MLR) models to estimate soil salinity from ECa survey data. The MLR models incorporate multiple ECa measurements and trend surface parameters to increase the prediction accuracy and can be fitted from limited amounts of ECe calibration data. This estimation technique is compared to some commonly recommended cokriging techniques, with respect to statistical modeling assumptions, calibration sample size requirements, and prediction capabilities. We show that MLR models are theoretically equivalent to and cost-effective relative to cokriging for estimating a spatially distributed random variable when the residuals from the regression model are spatially uncorrelated. MLR modeling and prediction techniques are demonstrated with data from three salinity surveys.