Wednesday, 12 January 2005
Development of the second-generation hydrosphere-atmosphere research model (CHARM) for the Laurentian Great Lakes region
The Coupled Hydrosphere-Atmosphere Research Model (CHARM) was developed to investigate the fully coupled hydrologic system of the Laurentian Great Lakes basin, using a dynamical regional climate model (RAMS) and surface models representing the land surrounding the Great Lakes as well as the lakes themselves. The final output of CHARM is the net basin supply in the basin of each of the Great Lakes, from which lake levels can be calculated. CHARM has been applied to scenarios of increased greenhouse gases, representing future times. This has generated much interest among policymakers and stakeholders, particularly as it gives results that contrast with a number of previous studies that used one-way coupling between the results of general circulation models for climate and a regional hydrologic model. Namely, CHARM indicates an increase in net basin supply and rise in lake levels as a result of increased greenhouse gas concentration, while the offline hydrologic models have indicated decreased net basin supply and a drop in lake levels. However, some distinct shortcomings remain in the simulation by CHARM. Most notable among these is a warm bias during the wintertime, which is believed to be linked to a heavy and persistent stratus cloud deck in the simulation. Additionally, there is concern over mismatch between the free atmosphere, whose character is largely determined by the boundary conditions imposed by a global-scale dataset or model, and the boundary layer, whose simulation is much more strongly influenced by the regional climate calculations within CHARM. The second generation of CHARM is under development, based on RAMS 4.4 (previous CHARM was based on RAMS 3a), which imparts such advantages as a fully interactive soil column and a dynamic model of snow. One particular remedy that is being tried for the cloud problem is to use interactive, rather than spatially interpolated, surface temperature for lakes, including the Great Lakes plus smaller lakes within the model domain. Another is to use land cover datasets that provide greater heterogeneity, thus varying the character of the boundary layer in space and possibly increasing vertical mixing. Yet another is to simply impose a minimum value of vertical diffusion coefficient in order to increase vertical mixing.