A cellular automaton based hydrologic model is reproduced involving soil–atmosphere interactions, with the purpose to studying the hydrological balance over a hypothetical region. The resulting time series of relative soil moisture, evapotranspiration, average rainfall and runoff, as well as the spatial fields of soil moisture and rainfall are quantified in terms of the emergence of temporal and spatial persistence. Our results indicate that strong soil–atmosphere interactions produce scale–invariant rainfall fields, both in space and in time, as is the case for observed rainfall records. Soil moisture time series shows scale invariance regardless of the strength of the land surface–atmosphere interactions. Long–term simulations involving the presence of annual and interannual cycles controlling the dynamics of rainfall, produces series of average precipitation which exhibit similar emergent properties as those observed in nature.