Improving spatial representation of vital biogeochemical data using hyperspectral and color imaging of soil cores
EMSL Project ID
60630
Abstract
Our objective is to develop an EMSL capability using hyperspectral imaging to generate biogeochemical maps (e.g. concentrations of: organic carbon: OC, nitrogen: N, iron: Fe) of soil cores, providing a rapid and non-destructive method to characterize and quantify elemental composition at the core scale. We will generate these elemental maps for the soil cores from EMSL’s 1000 Soils Pilot program. The spatial representation of select biogeochemical features (hyperspectral images) will provide insights to the heterogeneity of soil chemistry, including physicochemical gradients and hotspots. We will use hyperspectral cameras to image a cross section of the cores. Small subsamples (1 – 5 mg) will be collected from points across the core surface and will be analyzed for C, N, and Fe concentration using TOC/TN by combustion, and Fe from extractions. The chemical data will be used to train the image classification of the soil core using multivariate techniques which we will compare with state of the art machine-learning and deep-learning models.
Project Details
Start Date
2022-11-30
End Date
N/A
Status
Active
Released Data Link
Team
Principal Investigator
Team Members