We studied the effects of shading by shallow cumulus (shallow Cu) and the subsequent effect of inducing heterogeneous conditions at the surface on boundary-layer characteristics. We placed special emphasis on quantifying the changes in the characteristic length and time scales associated with thermals, shallow Cu and induced thermal circulation structures. A series of systematic numerical experiments, inspired by Amazonian thermodynamic conditions, was performed using a large-eddy simulation model coupled to a land-surface model. We used four different experiments to disentangle the effects of shallow Cu on the surface and the response of clouds to these surface changes. The experiments include a 'clear case', 'transparent clouds', 'shading clouds' and a case with a prescribed uniform domain and reduced surface heat flux. We also performed a sensitivity study on the effect of introducing a weak background flow. Length and time scales were calculated using autocorrelation and two-dimensional spectral analysis, and we found that shading controlled by shallow Cu locally lowers surface temperatures and consequently reduces the sensible and latent heat fluxes, thus inducing spatial and temporal variability in these fluxes. The length scale of this surface heterogeneity is not sufficiently large to generate circulations that are superimposed on the boundary-layer scale, but the heterogeneity does disturb boundary-layer dynamics and generates a flow opposite to the normal thermal circulation. Besides this effect, shallow Cu shading reduces turbulent kinetic energy and lowers the convective velocity scale, thus reducing the mass flux. This hampers the thermal lifetime, resulting in a decrease in the shallow Cu residence time (from 11 to 7 min). This reduction in lifetime, combined with a decrease in mass flux, leads to smaller clouds. This is partially compensated for by a decrease in thermal cell size due to a reduction in turbulent kinetic energy. As a result, inter-cloud distance is reduced, leading to a larger population of smaller clouds, while maintaining cloud cover similar to the non-shading clouds experiment. Introducing a 1 m s −1 background wind speed increases the thermal size in the sub-cloud layer, but the diagnosed surface-cloud coupling, quantified by characteristic time and length scales, remains.