A REVIEW OF THE CURRENT STATE OF PROCESS-BASED AND DATA-DRIVEN MODELLING: GUIDELINES FOR LAKE ERIE MANAGERS AND WATERSHED MODELLERS

We present a comprehensive evaluation of eleven process-based models to characterize the water cycle, nutrient fate and transport within a watershed context, and to find a robust and replicable way to optimize the modelling strategy for the Lake Erie watershed. Our primary objective is to review the conceptual/technical strengths and weaknesses of the individual models to reproduce surface runoff, groundwater, sediment transport, nutrient cycling, channel routing and collectively guide the management in Lake Erie Basin. Our analysis suggests that the available models either opted for simpler approximations of the multifaceted, non-linear dynamics of nutrient fate and transport and instead placed more emphasis on the advanced representation of the water cycle, or introduced a greater degree of biogeochemical complexity but simplified their strategies to recreate the role of critical hydrological processes. Notwithstanding its overparameterization problem, MIKE-SHE provides the most comprehensive 3D representation of the interplay between surface and subsurface hydrological processes with a fully dynamic description, whereby we can recreate the solute transport that infiltrates from the surface to the unsaturated soil layer and subsequently percolates into the saturated layer. Likewise, the physically based submodels designed to represent the sediment detachment and erosion/removal processes (DWSM, HBV-INCA, HSPF, HYPE and MIKE-SHE), offer a distinct alternative to USLE-type empirical strategies. The ability to explicitly simulate the daily plant growth (SWAT and APEX) coupled with a dynamic representation of soil P processes can be critical when evaluating the long-term watershed responses to various agricultural management strategies. While our propositions seem to favor the consideration of complex models that may lack the commensurate knowledge to properly characterize the underlying processes, we contend this issue can be counterbalanced by the joint consideration of simpler empirical models, under an ensemble framework, that can both constrain the plausible values of individual processes and validate macroscale patterns. Finally, our study discusses critical facets of the watershed modelling work in Lake Erie, such as the role of legacy P, the challenges in reproducing spring-freshet or event-flow conditions, and the dynamic characterization of water/nutrient cycles under the non-stationarity of a changing climate.The accepted manuscript in pdf format is listed with the files at the bottom of this page. The presentation of the authors' names and (or) special characters in the title of the manuscript may differ slightly between what is listed on this page and what is listed in the pdf file of the accepted manuscript; that in the pdf file of the accepted manuscript is what was submitted by the author