Spatial modelling of soil mechanical-hydraulic behavior for precision agriculture

Area of Science:

• Soil Science
• Precision Agriculture
• Geostatistics

Background:

• Current precision agriculture soil management relies on simplified scalar data or geostatistical models of individual parameters.
• This approach overlooks the inherent functional nature of many soil properties, limiting comprehensive understanding and management.
• Existing methods struggle to capture complex, spatially variable soil behaviors across different scales.

Purpose of the Study:

• To develop and validate a geostatistical framework for spatial modeling and zoning of function-valued soil descriptors.
• To explicitly address data heterogeneity, spatial non-stationarity, and multiscale variability in soil property modeling.
• To analyze the joint mechanical-hydraulic behavior of soil across various spatial scales.

Main Methods:

• Utilized functional responses: soil penetration resistance (vs. soil moisture) and soil water retention (vs. matric suction).
• Employed a workflow combining polygon-based estimation, multivariate geostatistical modeling of functional parameters, and factorial cokriging.
• Applied the framework to a 200-ha field with 100 sampling locations.

Main Results:

• Successfully reconstructed complete functional soil responses and quantified uncertainty at unsampled locations.
• Extracted scale-dependent regionalized factors that synthesize joint soil hydraulic and mechanical behavior.
• Enabled a multiscale field partition into zones with distinct soil conditions, revealing hidden spatial patterns.

Conclusions:

• The proposed framework offers a novel approach for interpreting and zoning functional soil behavior under non-stationary conditions.
• It provides complementary decision-support beyond scalar prediction accuracy, enhancing understanding of complex soil dynamics.
• This method facilitates more nuanced and effective within-field soil management strategies in precision agriculture.