Soil Organic Matter Mapping Using Hyperspectral Imagery and Elevation Data

Abstract

Knowledge of the spatial distribution of soil organic matter (OM) is highly important in many of today's applications of digital agriculture. Spatial OM awareness provides important insights to agronomists and allows the application of farm management practices in the right locations. Soil organic matter, which accumulates as a result of plant and animal decomposition, affects soil reflectance properties. The concentration of soil organic matter can be detected using aerial imagery, where darker images are correlated with higher OM concentration. In this paper we evaluate different approaches to OM analysis using hyperspectral imagery. By utilizing bare earth fields we can analyze a single image for the field without the need to acquire multiple images during the growing season or to time image acquisition before plant senescence and vegetation indice saturation. Here we report on a study in which bare earth spectral images, together with soil samples were collected in various counties in the Midwest. Two meter LIDAR digital elevation models were obtained from public sources. Soil reflectance, elevation, and topographic attributes and indices were used to describe the correlation between soil organic matter and various bands. Automated machine learning methods were used to analyze the relationship between soil spectral reflectance and soil organic matter. From a remote sensing and soil sampling perspective, in-field OM data collection is costly and time consuming. For this reason, more significant band identification using hyperspectral imagery is important for future sensor and OM prediction model development.