Iterated Application of a Linear Inverse Temperature Model to Lake Michigan SSTs Generated by CH3D

A methodology is presented describing the adjustment of a Lake Michigan hydrodynamic simulation through iterated application of a linear inverse temperature model. CH3D, a numerical model developed by the U.S. Army Corps of Engineers, is used to generate Lake Michigan surface temperatures for Mar 14, 2000. The simulation is initiated from Jan 1, 2000 under hourly windstress observations using a spatially uniform eddy diffusivity field. An inverse model is derived from the 2-d linear temperature diffusion equation which solves for eddy diffusivity given a surface temperature field. Observational SSTs are blended with model predicted temperatures and the inverse model is applied resulting in an adjustment to a spatially non-uniform eddy diffusivity field. The lake simulation is then re-initialized for the same time period with the refined diffusivity field. The process is repeated until the diffusivity field converges. The technique is tested using Great Lakes Forecasting System SST model output as a substitute for observational data, but the method is suitable for NOAA satellite SST data. This work was supported by the National Science Foundation.