Repeatability of commercially available visible and near infrared proximal soil sensors

Authors: CONWAY, LANCE - University Of Missouri; Sudduth, Kenneth - Ken; Kitchen, Newell; ANDERSON, STEPHEN - University Of Missouri

Submitted to: Precision Agriculture
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 12/29/2022
Publication Date: 1/24/2023
Citation: Conway, L.S., Sudduth, K.A., Kitchen, N.R., Anderson, S.H. 2023. Repeatability of commercially available visible and near infrared proximal soil sensors. Precision Agriculture. 24:1014-1029. https://doi.org/10.1007/s11119-022-09985-1
DOI: https://doi.org/10.1007/s11119-022-09985-1

Interpretive Summary: Gathering information about soil properties during seeding or tillage operations has potential to improve equipment performance and overall management of grain crops. Commercially available technologies currently allow integration of optical sensors that measure soil properties such as moisture and organic matter into traditional row-crop seeding or tillage equipment. Evaluation of these sensor systems is needed to determine their accuracy and repeatability across varying environments. Therefore, research was conducted in central Missouri in 2019 to determine how well two commercial sensor systems performed at estimating soil organic matter, and whether the estimates were repeatable. Results found that both systems were able to detect relative differences between low and high soil organic matter. However, one system was unable to capture the entire range in soil organic matter observed in laboratory-measured samples and provided estimates that were more variable in areas of low clay content. The second system provided organic matter estimates that were accurate and consistent across multiple sensing dates, a result attributed to the local (field-specific) calibration recommended by the manufacturer. This information will aid producers when interpreting measurements collected from both sensor systems and will help improve agronomic decisions made at planting that have potential to improve grain-crop yield.

Technical Abstract: Integration of reflectance sensors into commercial planter or tillage components have allowed for dense quantification of spatial soil variability. However, little is known about sensor performance and reproducibility. Therefore, research was conducted in Missouri in 2019 to determine (i) how well sensors can estimate soil organic matter (OM) and (ii) whether sensor output could be repeatable among sensing dates. Soil sensor data were collected across three weeks on an alluvial soil with the Precision Planting SmartFirmer and the Veris iScan. Output layers used in analyses included OM and Furrow Moisture from the SmartFirmer, as well as OM, reflectance, and soil apparent electrical conductivity from the iScan. Ground-truthing soil samples were collected at 0-5 cm on the first date to determine OM and on all dates to determine soil gravimetric water content. Results showed OM estimations by the iScan, which included the manufacturer’s specified field-specific calibration, were reproducible among the three sensing dates, with coefficient of determination (R2) ranging from 0.95 to 0.97. Similarly, root mean square error (RMSE) values were between 1.60 and 2.41 g kg-1. SmartFirmer results showed OM was overestimated in areas of low OM and underestimated in areas of high OM when compared to laboratory-measured data (R2 = 0.34; RMSE = 6.90 g kg-1). Additionally, variability existed in OM estimations between dates in areas that were lower in laboratory-measured OM, soil moisture, and clay content. These results suggest real-time estimations of OM may be subject to variability, and local information is likely necessary for consistent soil reflectance-based OM estimations.