Most land surfaces are quasi-randomly nonhomogeneous, yet most boundary-layer studies assume homogeneous or simply varying surface conditions. In this study, nonhomogeneous surface fluxes of realistic scale and amplitude are applied to a large-eddy-simulation (LES) model. The boundary conditions are representative of conditions observed near Chickasha, Oklahoma during the Boundary Layer Experiment 1983 (BLX83), while the model is a modified version of Moeng's LES model. The afternoon of 28 May 1983, which had light winds and cloud-free skies, is simulated by using data collected during BLX83 to provide initial and boundary conditions; the simulation verifies fairly well against observations. A second simulation of the case-study afternoon uses horizontally homogeneous surface fluxes. The results of the two runs are compared to see what effect the quasi-random nonhomogeneous conditions have on mixed-layer development. The inclusion of realistic size (450–900 m) and amplitude (∼2 (°C)2 variance in ... Abstract Most land surfaces are quasi-randomly nonhomogeneous, yet most boundary-layer studies assume homogeneous or simply varying surface conditions. In this study, nonhomogeneous surface fluxes of realistic scale and amplitude are applied to a large-eddy-simulation (LES) model. The boundary conditions are representative of conditions observed near Chickasha, Oklahoma during the Boundary Layer Experiment 1983 (BLX83), while the model is a modified version of Moeng's LES model. The afternoon of 28 May 1983, which had light winds and cloud-free skies, is simulated by using data collected during BLX83 to provide initial and boundary conditions; the simulation verifies fairly well against observations. A second simulation of the case-study afternoon uses horizontally homogeneous surface fluxes. The results of the two runs are compared to see what effect the quasi-random nonhomogeneous conditions have on mixed-layer development. The inclusion of realistic size (450–900 m) and amplitude (∼2 (°C)2 variance in ...