Sensitivity of circulation, thermal structure, and transient features to surface forcing in a Lake Erie FVCOM model

Lake Erie, the most urbanized and biologically productive of the Great Lakes, currently has multiple operational or semi-operational systems which are used to produce model-generated nowcasts and forecast guidance for water temperatures, currents, water levels, lake ice, harmful algal blooms and hypoxic bottom waters throughout the lake. It is likely that a major source of error in these products is error in the surface wind forcing, given Lake Erie’s relative shallowness and the sparsity of over-lake wind observations. This work investigates the sensitivity of a 2016 version of the FVCOM model used for these systems to the surface forcing used to drive the model. Two different forcing datasets that have been used operationally are used to drive the model for a year, resulting in substantial differences in model circulation and thermocline shape and depth, as well as in transient phenomena such as seiches and near-inertial oscillations. These results highlight some common long-term and short-term forecasting challenges for lakes with wind-driven circulation, and point to the value of using data assimilation (DA) to adjust the forecast initialization to more closely match recent observations, a process not yet used in Great Lakes operational forecast systems. The use of an ensemble-based DA system for Lake Erie will be explored in a forthcoming work.