Abstract For assessing the impacts of wind farms on regional climate, wind farms may be represented in climate models by an increase in aerodynamic roughness length. Studies employing this method have found near-surface temperature changes of 1–2 K over wind farm areas. By contrast, mesoscale and large-eddy simulations (LES), which represent wind farms as elevated sinks of momentum, generally showed temperature changes of less than 0.5 K. This study directly compares the two methods of representing wind farms in simulations of a strong diurnal cycle. Nearly the opposite wake structure is seen between the two methods, both during the day and at night. The sensible heat fluxes are generally exaggerated in the enhanced roughness approach, leading to much greater changes in temperature. Frequently, the two methods display the opposite sign in temperature change. Coarse resolution moderates the sensible heat fluxes but does not significantly improve the near-surface temperatures or low-level wind speed deficit. Since wind farm impacts modeled by the elevated momentum sink approach are similar to those seen in observations and from LES, the authors conclude that the increased surface roughness approach is not an appropriate option to represent wind farms or explore their impacts.