Assessment at field scale of soil organic carbon (TOC) is of primary interest for agronomic management, particularly, in the precision agriculture framework. The knowledge about the spatial distribution of TOC is invaluable to implement strategies for improving the crop yield. However, the assessment of TOC spatial distribution requires the collection and the analysis of a large number of samples that is a costly and timeconsuming activity. In the present work, a strategy is proposed to optimize a sampling scheme of the considered soil property by means of an indirect auxiliary variable coming from the proximal geophysical sensing survey (GPR data). This variable has a greater spatial continuity than the soil organic carbon, then can be straightforwardly modelled in the geostatistical fashion. This allows to apply the spatial simulated annealing as an efficient mean for reducing optimally the sampling scheme. In addition, two different variogram models are derived by an automatic method. Such models are compared for assessing i) which is more suited as a descriptor of the indirect variable spatial behaviour and ii) allows the efficient reduction of the sampling scheme discarding all the redundant sampling points and saving those truly informative.
IMPACT OF DIFFERENT VARIOGRAM MODELS OF TOTAL ORGANIC CARBON ON SAMPLING SCHEME OPTIMIZATION AND POTENTIALITY OF COVARIATE INFORMATION IN THE PRECISION AGRICULTURE FRAMEWORK
Daniela De Benedetto;Anna Maria Stellacci
2020-01-01
Abstract
Assessment at field scale of soil organic carbon (TOC) is of primary interest for agronomic management, particularly, in the precision agriculture framework. The knowledge about the spatial distribution of TOC is invaluable to implement strategies for improving the crop yield. However, the assessment of TOC spatial distribution requires the collection and the analysis of a large number of samples that is a costly and timeconsuming activity. In the present work, a strategy is proposed to optimize a sampling scheme of the considered soil property by means of an indirect auxiliary variable coming from the proximal geophysical sensing survey (GPR data). This variable has a greater spatial continuity than the soil organic carbon, then can be straightforwardly modelled in the geostatistical fashion. This allows to apply the spatial simulated annealing as an efficient mean for reducing optimally the sampling scheme. In addition, two different variogram models are derived by an automatic method. Such models are compared for assessing i) which is more suited as a descriptor of the indirect variable spatial behaviour and ii) allows the efficient reduction of the sampling scheme discarding all the redundant sampling points and saving those truly informative.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
