A Conceptual and Numerical Framework for Multiscale Data-Model
Integration in Plant-Microbe-Soil Systems
Abstract
The rhizosphere is a complex system in which many diverse and
heterogeneous small-scale components (e.g, plant roots, fluids,
microbes, and mineral surfaces) interact with one another, often in
nonlinear ways, giving rise to emergent system behaviors.
Ecosystem-scale perturbations, such as nitrogen limitation or drought,
drive changes in micro-environments through a cascade of complex
interacting processes, leading to a bidirectional feedback across scales
between microbial and plant habitats at the microscale and ecosystem
function at the macroscale. We are developing a conceptual and numerical
framework for multiscale simulation of organic carbon transport,
transformation, and disposition in the soil-microbe-plant continuum. The
conceptual model comprises a set of directed graphs, with nodes
representing system processes and states and edges representing
process-state relationships. The graphs are coded in the graphviz syntax
enabling dynamic web visualization. Graph nodes are hyperlinked to
metadata pages summarizing current understanding of each process or
state and its representation in current numerical codes. This conceptual
model is available via a git repository and can guide identification of
opportunities for coupling (data exchange) between codes operating at
different length scales. The numerical implementation of the conceptual
model is based on execution of integrated data processing and multiscale
modeling scientific workflows. The numerical framework is enabled by a
recent development in information technology known as orchestration, a
class of solutions to problems of deployment and execution of
cloud-oriented software. Orchestration technology is well-suited to
automating complex scientific workflows, both in model-coupling efforts
and experimental analysis pipelines. Here it is used to flexibly define
workflow steps based on precedent events (such as arrival of a new model
output in the data repository). It is being applied to integrate several
community software packages spanning scales from molecules to
ecosystems, linked to experimental data from the Environmental Molecular
Sciences Laboratory (a national scientific user facility), to address
critical scientific questions related to soil nutrient cycling.