The FLUXNET network with numerous eddy covariance stations distributed worldwide is an important backbone for the study of ecosystem-atmosphere interactions. In order to provide reliable data for a variety of related research fields all parts of the ecosystem-atmosphere interactions need to be fully captured. Energy balance closure can be an indicator that all fluxes are fully recorded. However, in an investigation of the FLUXNET data set over 20 years ago, a systematic imbalance of around 20 % was observed in the surface energy balance. By improving measurement instruments and arrangements as well as data post-processing, the imbalance was reduced to about 15 % within the following ten years. We show that the remaining imbalance has hardly changed to this day. In the meantime, it has become clear that the energy transport through mesoscale secondary circulations, which by definition cannot be captured with single-tower eddy covariance measurements, accounts for a large proportion of the remaining imbalance and leads to an underestimation of atmospheric energy fluxes. Storage changes, which have so far only been partially recorded, were also found to strongly contribute to the imbalance. In addition to recommendations for improving storage change measurements, we therefore present various energy balance closure approaches. These can be used to complement FLUXNET measurements by accounting for those flux contributions that cannot be captured by single-tower measurements or to parameterize the transport by secondary circulations in numerical weather and climate models. Another important finding in energy balance closure research is that secondary circulations contribute not only to energy transport but also to the transport of CO₂ and other substances, but more research is needed in this area. We conclude that research into energy balance closure problem has made great progress in recent years, which is crucial for investigating the role of ecosystems in the Earth system.