Unravelling Patterns and Processes of Terrestrial and Aquatic Carbon Fluxes under Hydroclimate Extremes

Li Li, The Pennsylvania State University (Principal Investigator)
Rodrigo Vargas, University of Delaware (Co-Investigator)
Ben Bond-Lamberty, Pacific Northwest National Laboratory (Unfunded Co-Investigator)

This project seeks to determine the patterns and processes that drive vertical (terrestrial) and lateral (aquatic) carbon flux response to hydroclimate extremes across gradients of climate, vegetation, and watershed characteristics. Vertical effluxes are defined as the soil CO₂ effluxes to the atmosphere and lateral fluxes as the carbon (C) export to rivers and streams, including dissolved organic and inorganic C (DOC and DIC) and particulate organic C (POC). Hydroclimate extremes, including storms, floods, and droughts, are projected to become more frequent and intense, and can cause substantial changes in the carbon (C) reactions and fluxes. Such changes can potentially transform terrestrial ecosystems into net carbon sources, exert long-lasting effects on surface water quality and aquatic ecosystems, and threaten water supplies. Extreme events are rare and difficult to track and replicate in space and time. The responses of C to extremes are complex and nonlinear, such that small shifts in event characteristics can substantially alter responses. Existing studies have mostly adopted approaches in separate disciplines, demanding more systematic and integrated understanding of how, how much, and to what extent C reactions and fluxes respond to extremes.

The project seeks to answer these questions by leveraging interdisciplinary data synthesis and harnessing the power of deep learning and process-based reactive transport modeling. At the Continental United States scale, the team plans to synthesize existing C concentrations and soil CO₂ efflux data, that are often measured separately in ecosystems, soils, and aquatic biogeochemistry. Deep learning approaches will be used to “reconstruct” consistent data to identify patterns of C flux response to hydroclimate extremes. At the watershed scale, the team plans to focus on two ESS-supported sites with intensive, coordinated data and use reactive transport models to understand the processes that drive C fluxes. One site is the Upper East River site in Colorado, which is experiencing rapid warming and frequent droughts; the other is the TEMPEST site in Maryland, where controlled experiments and a rich dataset can help understand threats of storms and floods that loom large in the coastal U.S.. The proposed work forges a new collaboration across hydro- and terrestrial biogeochemistry, thus breaking discipline silos. The rich datasets will set the stage for new ideas that advance multiple fields. The proposal aligns with the DOE mission to “incorporate AI into models, analytics, and data generation”, and will support women and minority students in STEM, thus helping diversify the machine learning field and the geosciences.