Using a combined deductive and inductive approach to assess catchment attributes and water availability effects on stream water composition across the continental U.S.

The Critical Zone (CZ) is the part of the Earth where the biosphere, hydrosphere, and atmosphere interact to sustain life on Earth through the provision of important ecosystem services. Water of suitable quantity and quality is one of these important services and is closely related to CZ function. However, the impact of the CZ attributes on stream water composition, and possible trickle-down effects across several interconnected parts of this system, are not yet deeply understood. To investigate connections between CZ structure, climatic parameters and stream water composition I use a combination of inductive (i.e. initiated by observations) and deductive (i.e. initiated by hypothesis) approaches. For this, I use CAMELS-Chem, a dataset which contains hydrologic, geologic, climatic, and vegetative attributes, as well as geogenic and biogenic solute concentrations for 506 catchments across the continental United States (CONUS). My deductive approach included visualizing climate, stream composition and select CZ attributes using the Budyko framework to test hypotheses on the role of root depth and bedrock permeability. While my results did not confirm my hypothesis, they clearly indicated the importance of the types of subsurface materials and bedrock for stream water concentration, informing next steps with the inductive analysis. To accomplish this, I used a random forest analysis for a data-driven approach to identify catchment attributes that drive stream water solute concentration. Results indicated distinct variation by subsurface materials and indicated important, possibly confounding effects between water availability, vegetation, and stream composition. Importantly, ecoregions need to be considered when analyzing data in the context of a geographically, ecologically and climatically diverse context such as the CONUS.