Optimization of wind turbine (WT) arrays to maximize system-wide power production (i.e. minimize ‘wind-theft’) requires high-fidelity simulations of array-array interactions at the regional scale. This study systematically compares two parameterizations (Fitch and EWP) developed to describe wind farm impacts on atmospheric flow in the Weather Research and Forecasting (WRF) model. We present new year-long simulations for a nested domain centred on Iowa (the state with highest WT density) in the US Midwest that employ real WT characteristics and locations. Simulations with Fitch and EWP indicate similar seasonality in system-wide gross capacity factors (CF) for WT operating in Iowa, but the gross CF are systematically higher in simulations using EWP. The mean gross CF from the Fitch scheme is 44.1%, while that from EWP is 46.4%. These differences in CF are due to marked differences in the intensity and vertical profile of wakes simulated by the two approaches. Output from EWP also indicates much smaller near-surface climate impacts from WT. For example, when summertime hourly near-surface temperature (T2m) from the 299 WT grid cells are compared (i.e. EWP or Fitch minus noWT) the results show warming of nocturnal temperatures (lowest decile of T2m) but the maximum warming is considerably larger in simulations with the Fitch scheme.