Mapping Intra-Field Variability of Soil Horizons Using Ground Penetrating Radar.

Recent advancements in mobile ground penetrating radar (GPR) technology showed promise in characterizing spatial dynamics of soil profiles. The objective of this study was to evaluate the efficiency of GPR to map soil horizon thickness in two potato fields from New Brunswick, Canada. A soil survey was completed in February 2016 using a GPR (GSSI model SIR-3000; 400 MHz antenna; 30 scan sec^-1) antenna attached to a receiver and a DGPS receiver with submeter accuracy on two 12-ha fields (referred to as SVP and SVS fields). Data (density about 20 416 points ha^-1) were collected on parallel transects approximately 10-m apart (28 transects in SVP and 25 in SVS fields, respectively). The dielectric contrast of the GPR was adjusted using the maximum depth penetration of the VERIS P4000® probe (Veris Technologies Inc., Salina, Kansas) at 41 and 37 sampling sites in the SVP and SVS fields, respectively. A field-specific GPR calibration was completed using the signal of a metallic plate installed at a depth of 0.75 m in the soil. In both fields, two soil layer thickness (SLT) (referred to SLT surface and SLT subsurface limited by rock depth) showed contrasting relative amplitude of the GPR waveform (linescan). Interpolated maps of SLT surface and SLT subsurface were obtained using ordinary kriging (exponential model semivariogram). In the SVP field, SLT surface varied from 5 to 25 cm (average=14 cm) and the SLT subsurface varied from 24 to 122 cm (average=73 cm). In the SVS field, SLT surface varied from 8 to 31 cm (average=19 cm) and the SLT subsurface varied from 18 to 121 cm (average=72 cm). The RMSE of the cross-validation (0.019 and 0.01 for SLT surface and 0.075 and 0.10 for SLT subsurface for fields SVS and SVP, respectively) clearly showed efficiency of GPR to quantify soil layer thickness.