Understanding the interactions between the land surface and the atmosphere is key to modelling boundary-layer meteorology and cloud formation, as well as carbon cycling and crop yield. In this study we explore these interactions in the exchange of water, heat and CO 2 in a cropland–atmosphere system at the diurnal and local scale. To that end, we couple an atmospheric mixed-layer model (MXL) to two land-surface schemes developed from two different perspectives: while one land-surface scheme (A-g s ) simulates vegetation from an atmospheric point of view, the other (GECROS) simulates vegetation from a carbon-storage point of view. We calculate surface fluxes of heat, moisture and carbon, as well as the resulting atmospheric state and boundary-layer dynamics, over a maize field in the Netherlands, on a day for which we have a rich set of observations available. Particular emphasis is placed on understanding the role of upper-atmosphere conditions like subsidence in comparison to the role of surface forcings like soil moisture. We show that the atmospheric-oriented model (MXL-A-g s ) outperforms the carbon storage-oriented model (MXL-GECROS) on this diurnal scale. We find this performance is partly due to the difference of scales at which the models were made to run. Most importantly, this performance strongly depends on the sensitivity of the modelled stomatal conductance to water stress, which is implemented differently in each model. This sensitivity also influences the magnitude of the surface fluxes of CO 2 , water and heat (surface control) and subsequently impacts the boundary-layer growth and entrainment fluxes (upper atmosphere control), which alter the atmospheric state. These findings suggest that observed CO 2 mole fractions in the boundary layer can reflect strong influences of both the surface and upper-atmosphere conditions, and the interpretation of CO 2 mole fraction variations depends on the assumed land-surface coupling. We illustrate this with a sensitivity analysis where high subsidence and soil moisture depletion, typical for periods of drought, have competing and opposite effects on the boundary-layer height h . The resulting net decrease in h induces a change of 12 ppm in the late-afternoon CO 2 mole fraction. Also, the effect of such high subsidence and soil moisture depletion on the surface Bowen ratio are of the same magnitude. Thus, correctly including such two-way land-surface interactions on the diurnal scale can potentially improve our understanding and interpretation of observed variations in atmospheric CO 2 , as well as improve crop yield forecasts by better describing the water loss and carbon gain.