A numerical model of meteorology, aerosols, and radiative transfer is used to study the impact of a large forest fire smoke plume on atmospheric processes. The simulated smoke optical depths at 0.63 μm. wavelength are in agreement with analyses of satellite data and show values as high as 1.8. The smoke has an albedo of 35%, or more than double the clear-sky value, and cools the surface by as much as 5 K. The best agreement with the analyses of optical depth and surface cooling is obtained when a fuel loading more than 10 times that which has been previously suggested is used to calculate the initial smoke mass load. An imaginary refractive index, nim, of 0.01 yields results which closely match the observed cooling, single scattering albedo, and the Ångström wavelength exponent. An nim of 0.1, typical of smoke from urban fires, produces 9 K cooling. Coagulation causes the geometric mean radius by number to increase from the initial value of 0.08 μm to a final value of 0.15 μm while the specific extinction and absorption increase by 40% and 25%, respectively. After 42 hours, these changes in the smoke optical properties lead to a 32% increase in optical depth and an 11% increase in surface cooling over that found in a simulation where coagulation is not allowed. In the model, 47% of the smoke is removed by scavenging as it is incorporated into the frontal zone over the Great Lakes. Self-lofting of the smoke in a direct, smoke-induced circulation is observed in the baseline simulation and is much stronger in the urban smoke simulation.