Improve predictive understanding of soil carbon cycling by characterizing composition, distribution, and persistence of soil organic matter using 3-D nanoscale mass spectrometric imaging and machine learning
EMSL Project ID
60685
Abstract
Machine learning (ML) methods will be used to quantitatively determine relationships between soil organic matter (SOM) composition, mass fraction and reactivity and SOM-mineral interactions characterized by advanced 3-dimensional (3-D) nanoscale mass spectrometric imaging. A data base will be developed, which can improve the mechanistic representation of carbon cycling in watershed and Earth System Models (ESMs). Mineral-SOM interactions are essential for stabilizing soil nutrients that influences carbon (C) and nitrogen (N) biogeochemical cycling in soil. Current ESMs treat SOM–mineral interactions as a “black-box”, leading to large uncertainties and bias in predictions. SOM-mineral-microbe interactions are complex because they occur majorly at various surfaces, while most analysis tools used in this field cannot provide molecular information at the surfaces. We will use EMSL’s state-of-the-art 3-D mass spectrometric imaging tools that are highly surface sensitive to characterize the SOM composition and identify their co-existence along with various mineral particles. Meanwhile, ML methods can be used to leverage these experimental data, along with massive data available in EMSL user programs, EMSL 1000 Soil Project, and other open-source, community database, to generate reaction parameters that consider the SOM-mineral interactions derived from those micro-scale measurements and can be incorporated into ESMs, ultimately minimizing the uncertainty and bias in predicted carbon emission/sequestration.
Project Details
Start Date
2023-01-27
End Date
N/A
Status
Active
Released Data Link
Team
Principal Investigator
Team Members