Two-year simulation of the Great Lakes region with a coupled modeling system

In this paper, we report on an experiment aimed at evaluating the feasibility of the application of our coupled regional climate modeling system to long-term climate simulations over the Great Lakes region. The simulation analyzed covers a continuous 24-month period beginning 1 September 1990 and extending to 1 September 1992.

Many aspects of this simulation agreed well with observations. Compared with European Centre for Medium-Range Weather Forecasts (ECMWF) analyses, area-averaged atmospheric temperature and moisture biases were generally small. The largest temperature biases were found in the simulated boundary layer, up to 1–1.5 K colder than observed in most months. Atmospheric moisture biases were of both signs and small in magnitude, almost universally less than 0.5 g kg⁻¹.

Comparison of simulated surface air temperatures with station observations also indicated that model simulated temperatures generally display a cold bias. Simulated precipitation values agreed well with observations during the cold portions of the year while during warm months precipitation was overpredicted by 10%–50%. Spatial patterns of precipitation over the model domain agreed well with observations during the winter months but were not as well simulated during the other seasons.

A one-dimensional lake model was coupled to the atmospheric component of the model to capture the effects of the Great Lakes on regional climate. Lake surface temperatures were generally well simulated by the lake model in the summer and fall seasons, and lake ice extent agreed well with the analysis over at least three of the five lakes. The greatest shortcomings in lake temperature simulation were the earlier-than-observed warm-up in the spring and warmer than observed peak temperatures in the summer over the northern portions of the lakes. Also, lake ice extent was generally overpredicted over Lake Superior and underpredicted over Lake Erie.

In summary, the coupled modeling system described in this paper shows promise for use in climate simulators over regions where lakes are important such as the Great Lakes. It has been shown that many aspects of the simulation are in good agreement with available observations. Areas in which the results point to the need for further work are the model's convective parameterization, the eddy diffusivities in the lake model, and the treatment of clouds in the radiation package.