Are land management parameters influential to watershed model calibration? Results from a global sensitivity analysis of the HYPE model and detailed land manager survey.

Significant research progress has been made regarding the application of sensitivity analysis techniques to watershed models. There currently exists a battery of sensitivity analysis tests and extant literature on the most commonly used models. While hydrological parameters have been studied quite extensively, far less attention has been paid to land management parameters such as fertilizer application rates, tillage practices, and other things representing the human footprint on the land. Typical practices in modelling include relying on recommendations or regional averages, despite many models being spatially distributed. This study integrates the results of an intensive campaign to survey farmers and monitor streams in a small agricultural watershed in the Laurentian Great Lakes with model sensitivity analysis techniques. We assess the sensitivity of a number of typical watershed model calibration parameters, a number of management parameters, and a pooled set of management and calibration parameters. We also assess this sensitivity across three spatial configurations of the model. We show that some management parameters are more sensitive than the typical calibration parameters for all model configurations. These results will help watershed modellers focus on which management parameters should be prioritized for constraining watershed models.