Source Relationships and Model Structures Determine Information Flow Paths in Ecohydrologic Models

In a complex ecohydrologic system, vegetation and soil variables combine to dictate heat fluxes, these fluxes may vary depending on the extent which drivers are linearly or non-linearly interrelated. From modeling causality perspective, uncertainty, sensitivity, performance measures all relate how information from different sources “flows” through model produce target, output. We address structure, broadly defined as mapping inputs an output, combines with source dependencies range of flow pathways target. apply decomposition, partitions reductions in uncertainty into synergistic, redundant, unique types, cases. Toy models show that structure both restrict types interactions can arise between targets. Regressions based weather data illustrate structures their sensitivity dependencies, thus affecting predictive functional performance. Finally, we compare Surface Flux Equibrium theory, land-surface model, neural networks estimating Bowen ratio, find trade off particularly when have highest lowest dependencies. Overall, this study extends theory-based evaluation framework incorporate influence dependency pathways. This could be applied explore behavioral ranges for machine learning process-based models, guide development by highlighting deficiencies would not apparent existing measures.