The Eddy Covariance (EC) technique is used worldwide to monitor gas and energy exchanges at the interface between ecosystems and the atmosphere. Turbulent flux quality can be assessed through the energy conservation principle stating, in the case of a flux tower, that the sum of latent and sensible heat fluxes measured by EC is equal to the available energy. This is the difference between net radiation flux, ground heat flux and minor heat storage fluxes, which are measured by independent systems. If this equality is verified, the balance is said to be “closed” and, postulating that available energy measurements are unbiased, EC measurements are assumed to be valid. However, many studies indicate a general lack of closure in contrasted ecosystems and climates. This problem has been subjected to intense research for the last 30 years in order to understand the causes of unclosure and to assess the consequences on gas flux measurement quality. This research aims to contribute to the general understanding of EC accuracy at measuring gas and energy fluxes. To do so, we used a multi-year dataset collected over the cropland ICOS station of Lonzée, focusing on the identification of the drivers that lead to improvement or worsening of the energy balance closure following hypothesis found in recent papers. The choice of a cropland site was driven by the opportunity to have a dynamic and contrasted land cover, hopefully facilitating the identification of those drivers. The energy balance closure has been evaluated at different time and spatial scales. Two potential imbalance causes have been investigated: (1) the underestimation and/or the negligence of certain fluxes in the energy balance; (2) the 30 minutes averaging period which may neglect some low-frequency components of latent and sensible heat fluxes. The results showed that, even if energy balance closure could not be achieved for the present studied site, (1) minor heat storage fluxes evaluation into the ground, the biomass, and the air allows improving the ...