Predicting soil organic matter content in southwestern Ontario fields using imagery from high-resolution digital cameras

Abstract

The spatial pattern of soil organic matter (SOM) content may provide information for variable-rate fertilizer nitrogen recommendations, but often requires intensive soil sampling to be properly characterized. This study evaluated whether imagery of bare soils obtained using a high-resolution digital camera system may be used to predict SOM content in southwestern Ontario fields. Using the camera system, image intensity was measured in near-infrared (0.70-0.80 μm) and visible red (0.60-0.70 μm), green (0.50-0.60 μm), and blue (0.40-0.50 μm) wavebands under both controlled and field conditions for soil samples representative of the range in SOM typically found in southwestern Ontario fields. Under controlled conditions, SOM content did not relate well to image intensity measured in any waveband when multiple soil types were used (r2 ≤ 0.39). Without multiple soil types, image intensity in all wavebands related to SOM content for soil samples taken from two of the sites (r2 ≥ 0.53 for both sites). Image intensity measured under field conditions related to SOM content for only one site (r 2 ≥ 0.54 for all wavebands). Variability between SOM content and image intensity shown in this study is most likely attributed to the relative variability in SOM content and confounding factors such as surface residue.