Location: Crop Production Systems ResearchTitle: Machine learning mapping of soil electrical conductivity in mississippi
Submitted to: Agricultural Sciences
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/18/2023
Publication Date: 7/21/2023
Citation: Fletcher, R.S. 2023. Machine learning mapping of soil electrical conductivity in mississippi. Agricultural Sciences. Vol.14 No.7 July 2023. https://doi.org/10.4236/as.2023.147061.
Interpretive Summary: Open-source and free tools are readily available to the public to process data and assist producers in making management decisions related to agricultural landscapes. On-the-go soil sensors are being used as a proxy to map soils because of the amount of data they can collect and their ability to cover a large area quickly. Machine learning is a form of artificial intelligence that makes predictions from data. More information is needed on integrating these technologies for developing digital soil maps in Mississippi at a field scale. A scientist at the USDA-ARS, Crop Production Systems Research Unit, Stoneville, MS, demonstrated that a freely available open-source machine learning tool could be used to analyze apparent electrical conductivity data and derive maps to use at the field scale. The apparent electrical conductivity data were acquired with an on-the-go soil apparatus. Measurement depth of the apparent electrical conductivity readings and site location appeared to influence the machine learning tool’s ability in developing models to process the data.
Technical Abstract: Open-source and free tools are readily available to the public to process data and assist producers in making management decisions related to agricultural landscapes. On-the-go soil sensors are being used as a proxy to map soils because of the amount of data they can collect and their ability to cover a large area quickly. Machine learning is a form of artificial that makes predictions from data. The intermixing of open-source tools, on-the-go technologies, and machine learning may vastly improve Mississippi soil mapping and crop production. This study aimed to evaluate machine learning as a tool for mapping apparent soil electrical conductivity at two sites on a research farm in Mississippi. Site one (MF2) contained multiple soil types; site two (MF9) consisted of one single soil type. The apparent electrical conductivity readings were collected with an on-the-go soil system. Machine learning tools incorporated in Smart-map, an open source machine learning tool, was used to evaluate the sites and derive the apparent electrical conductivity maps. Autocorrelation of the shallow (ECas) and deep (ECad) readings were statistically significant at both sites (Moran’s I, p 0.001); however, the spatial correlation was greater at MF2. The best models according to the leave one out cross validation results were developed at MF2. Spatial patterns were observed for the ECas and ECad readings in both fields. The patterns observed for the ECad readings were more distinct than the ECas measurements. The research results indicated that machine learning was a valuable tool to use for deriving apparent electrical conductivity maps in two Mississippi fields. Location and depth played a role in the machine learner’s ability to develop the maps.