Abstract

CASTRO-FRANCO, Mauricio et al. Predicción de contenido de arcilla superficial utilizando conductividad eléctrica aparente y esquemas de muestreo basados en modelos. Cienc. suelo [online]. 2017, vol.35, n.1, pp.135-146. ISSN 1850-2067.

Spatial prediction of clay content at field scale is needed to implement precision agriculture and hydrological models. However, the lack of techniques that can detect clay content heterogeneity limits the ability to determine its variability. In this study, we tested the use of geostatistical interpolation (ordinary cokriging), apparent electrical conductivity (CEa) as auxiliary information and two model-based soil sampling schemes (EBM) (conditioned Latin hypercube (HCL) and fuzzy K-means (FCM) to predict clay content in an 25.18 ha agricultural field. Results support the underlying assumptions that both HCL and FCM capture adequately the full distribution of CEa; and that clay content was closely related to the CEa. Also, suggested that (i) the type of EBM affects the clay prediction model efficiency; (ii) a considerable soil sample reduction is possible when the proposed methodology is applied; (iii) an independent data set is most adequate to validate the proposed methodology; and (iv) the geostatistical interpolation based on CEa and HCL provided a slight improvement in the clay content prediction (R² = 0.75, RMSE = 1.50%) compared to the geostatistical interpolation based on CEa and FCM (R² = 0.73, RMSE = 1.69%). The proposed methodology provided a significant improvement of information on clay content with respect to soil survey techniques and is easy to replicate in other farm fields. Therefore, it can be significant to implement these findings in site-specific managements or hydrological simulations.

Keywords : Digital soil mapping; precision agriculture; soil texture; VERIS 3100®.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )