Experimental Approach to Quantifying Synthetic Corn Seed Measurement error Using a Ground Penetrating Radar

Abstract

Planter manufacturers' provide open-loop pressure-based active control systems to improve planting depth (PD) control. These systems could be significantly enhanced using realtime seed depth sensors for closed-loop pressure-based control accuracy with respect to PD. The primary objective of this proof of concept study was to use a 1.6 and 2.6 GHz ground penetrating radar (GPR) and evaluate its effectiveness in detection in sandy-loam and loam soils. Synthetic Corn Seeds (SCS) were used to simulate the agricultural corn seeds with a uniform flat profile shape of three different sizes made of stainless steel and wood. Five treatment factors were investigated: soil moisture contents, soil types, salinity, seed materials, and seed sizes. The detection rate varied with soil salinity, type and moisture contents. Highest SCS detection rate was at the lowest salinity and moisture contents. Three methods: (Topp's dielectric, soil mixing, and constant dielectric) were used to estimate PD. The statistical regression analysis showed that the Topp's dielectric model was a useful empirical model to estimate PD. The Topp's dielectric model had Mallows Cp ≤ 3.94 and p = 4.00 for wood seeds measurement error predictions. The GPR measurement error was 7.86% which corresponded to a coefficient of planting depth accuracy value of ±0.005 m relative to the theoretical depth. Under certain workable soil conditions, SCS signatures were of high quality and conversely, in other soil conditions target signatures were sub-par or nonexistent. The implication was the utility of GPR for SCS detection was site-specific, and a detailed understanding of the soil matrix is necessary for successful SCS detection. The SCS material and size had a profound effect on the GPR attainment of readable radargrams.