The imbalance between the measured available energy and the sum of the turbulent fluxes lead to the energy balance closure problem. In spite of several experimental and modelling studies, the reasons for the lack of closure are not fully understood, particularly, in a heterogeneous terrain. The LITFASS-2003 campaign in Northeastern Germany was designed to develop and to assess different area-averaging strategies of the surface fluxes over a heterogeneous land surface. The micrometeorological measurements of the campaign were targeted at local fluxes over different types of land surface that are essential to study the energy balance closure problem for a complex land surface where the secondary circulations induced by surface heterogeneity are suspected to influence the surface energy budget imbalance. To assess the influence of the secondary circulations we perform large-eddy simulations of a 5.4 × 5.4 km² sub-region of the LITFASS area with a flat topography and composed mainly of agricultural land. The boundary conditions for the simulation domain is derived from the experimental data collected on 30 May 2003. To capture the spatial variation of the fluxes, the surface fluxes of latent and sensible heat flux in the simulated domain are prescribed by composite fluxes derived from multiple surface flux stations operated during the experiment. A grid resolution of 1 m in the vertical and 2 m in the horizontal directions up to 72 m from the ground is achieved by employing a nested large-eddy simulation model. A total of five realizations of the domain is performed to calculate ensemble averages to separate the heterogeneity effect from the turbulence fluctuations and the 30-minute time-averaging ensures more representative statistics. We find the underestimation to be systematic and to increase with height. At a typical eddy covariance tower height of 10 m, we find the dispersive flux represents up to 5% of the prescribed surface fluxes, which partially explains the imbalance in the field measurements.