Degree Type


Date of Award


Degree Name

Doctor of Philosophy


Agricultural and Biosystems Engineering


Agricultural and Biosystems Engineering

First Advisor

Stuart J. Birrrell


Measuring planting depth is a challenge in precision seeding. Planting depth is commonly determined after seedling emergence and through a conventional method as described in DuPont Pioneer (2016), and ISO (1984), respectively. These approaches are disruptive to the soil seedbed, labor-intensive, time-consuming and large areas are not practical to survey. Therefore, a quality control technique to measure the planting depth after planting need to be established. The primary objective of this research work was to investigate the performance of nondestructive Ground Penetrating Radar (GPR) technology to detect buried corn seeds and measure corn planting depth. The GPR is a geophysical tool that employs material physical properties in a nondestructive manner to detect anomalies. GPR has been used in numerous agricultural applications with substantial success. Its application to detect corn seed is novel and may contribute significantly to precision seeding. Thus, a five-part research was conducted to meet the overall objective.

A theoretical review was conducted to identify subsurface sensing candidates and describe the techniques that could have application to the detection of corn seeds. Based on the merits, operational principles, and the suitability criteria of each technique, the GPR was recommended as the viable technology. The possibility and accuracy of using the GPR were investigated using numerical simulations, laboratory, and field experimental tests. An open-source software was used for numerical simulations. GPR antenna center frequencies of 1.6 and 2.6 GHz were examined. The effects of varied soil composition, bulk density, soil moisture contents, corn seed size, and corn dielectric properties were evaluated. The simulation results indicated that the GPR reflections provided substantial responses to locate synthetic and natural corn seed models in different soils with diverse conditions except in higher soil bulk densities and clay. At low soil moisture models, it was observed that amplitudes were extremely low which was attributed to the lack of dielectric contrast between the soil and the natural seed model. As expected, low clay content, bulk density, variable workable field soil moisture content was suitable for GPR wave propagation and detection. Higher antenna frequency provided higher target amplitude responses. Higher soil moisture relative to the natural seed model produced higher amplitude compared to drier soil.

Laboratory tests were conducted with four treatments: two soil types, salinity, soil moisture contents, corn seed simulants (stainless steel and plywood) and three simulant sizes. The simulants were buried at a constant depth of 7.62 cm. The 1.6 and 2.6 GHz antenna center frequencies were tested. Metal seeds were readily detectable in different soil conditions despite the seed size mainly using a 2.6 GHz antenna. However, at higher soil salinity and moisture contents, weaker hyperbola responses signatures were observed. Neither wood size was detected at higher soil salinity and moisture contents. However, the detection of wood seeds was possible in soil conditions that were below field capacity, using the 2.6 GHz antenna. The simulant sizes profoundly affected the detectability. The assessment of the 2.6 GHz antenna effectiveness was based on the percent Coefficient of Planting Depth Accuracy (CPDA) �0.5 cm. The CPDA was determined to be �7.86% for the wood seeds.

Further laboratory tests were carried out using the Pioneer P0339AMXT PDR variety with an average moisture content of 10.1% using the 2.6 GHz GPR antenna frequency. In dry (less than 5% soil moisture content) soils some corn seeds were detected, while in intermediate (within 5% and 13%) and moist (greater than 13%) soils it was challenging to detect corn seeds. The dielectric permittivity used for calculating the depths were computed using Topp dielectric, soil mixing, and Topp-Mixing models. The Topp-Mixing model had the best CPDA of �8.37%. Corn seeds were detected in controlled environment.

Field tests were conducted using the Pioneer P8542AMX-NJ02 PDR variety with an average moisture content of 13.2%. Corn seed spacing and planting depths were measured manually and with the 2.6 GHz antenna. The experimental design had three parameter treatments consisting of; two downforce pressures, three populations, and three planting depths. The downforces significantly affected the planting depths and variability. The planting depth variability was more than the nominal resolution of the planter. Corn seeds were not reliably detected in the field. Hyperbolic features observed within the radargrams were random and inconsistent with the seed depth, and it was challenging to identify individual corn seed. There were horizontal features that were consistent and had similar two-way travel times to the bands observed in parallel and perpendicular radargrams. If the two-way travel times were the same, then it is highly likely that the bands in the parallel and perpendicular radargrams are from the same feature. Using the perpendicular to matchup with the parallel A-scans features that could represent the furrow were effectively identified. The work described shows that GPR could be employed to identify furrow features. The work presented in this dissertation necessitates more future research to establish the use of GPR as a quality control tool in precision seeding.

Copyright Owner

Kenneth Obrien Mpho Mapoka



File Format


File Size

191 pages