Vortrag (eingeladen), Flux Measurements in Difficult Conditions, a Specialist Workshop, Boulder, CO, U.S.A.: 26.01.2006 - 28.01.2006
During the late 1980s it became obvious that the energy balance at the earth’s surface could not be closed with experimental data. The available energy, i.e. the sum of the net radiation and the ground heat flux, was found in most cases to be larger than the sum of the turbulent fluxes of sensible and latent heat. This was a main topic of a workshop held in 1994 in Grenoble . In most of the land surface experiments [2, 9, 19] as well as in the carbon dioxide flux networks [1, 20], a closure of the energy balance of approximately 80% was found. An experiment designed to investigate this problem, the EBEX- 2000, took place in the summer of 2000 in near Fresno, California. Its results are now submitted to publication  and discussed in this paper.
The problem cannot be described only as an effect of statistically distributed measuring errors because of the clear underestimation of turbulent fluxes or overestimation of the available energy. In the literature, several reasons for this incongruity have been discussed, most recently in an another overview paper , and the results of EBEX-2000 underlines these results.
The most common point of discussion is measurement errors, especially those of the eddy-covariance technique, which cause a systematic underestimation of the turbulent fluxes. Improvements in the sensors, as well as in the correction methods, and the application of a more stringent determination of the data quality have made this method much more reliable in over the past ten years [12, 13].
The unclosure of the energy balance was also connected with the heterogeneity of the land surface . The authors assumed that the heterogeneities generate eddies at larger time scales than the eddies measured with the eddy-covariance method. This problem is also closely connected with advection and fluxes due to longer wavelengths . And obviously this problem may be the key to the solution. Our investigation underlines the already made assumption  that the averaging length is probably site dependent. In the heterogeneous landscapes of Central Europe or in North American agricultural areas the averaging length is significantly longer than the typical averaging length of eddy covariance measurements of 30 minutes, while over the more uniform African landscape the averaging length is within the 30 minute interval [8, 11].
Oncley  introduced the ogive function into the investigation of turbulent fluxes. The ogive is the cumulative integral of the co-spectrum starting with the highest frequencies. This function was proposed as a test to check if all low frequency range are included in the turbulent flux measured with the eddy-covariance method . It was found  that the ogive converge is not always within 30 minutes. Often the integration time is much longer (see above), or the ogive function has a maximum for shorter integration times than 30 minutes. In this case, the longwave part of the flux has the opposite sign than the shortwave part.
Obviously, the different turbulent energy fluxes do not have a similar ogive function for all conditions. While a similarity of scalars was found for the high frequency spectra , typical differences were found in the low frequency range , probably connected with the sink/source functions of the scalars, which can be different during the daily cycle. Therefore, the results of the ogive analysis for the sensible and latent heat flux or the residual of the energy balance cannot be used for the correction of other turbulent fluxes like the carbon dioxide flux. Each flux must be analysed separately, e.g. by the ogive method.
For further investigations of the energy balance closure problem the main topic of research should be the averaging length of turbulent measurements. Investigations should be made for sites in different landscapes and different meteorological conditions. The energy balance closure problem also significant for all trace gas fluxes like carbon dioxide.
Acknowledgement: The author is grateful to his PhD students C. Liebethal, M. Mauder, C. Thomas and F. Wimmer and all participants of the experiment EBEX-2000, especially S. P. Oncley, R. Vogt, W. Kohsiek and H. Liu.
 M. Aubinet, et al., 2000, Adv. Ecol. Res., 30, 113-175.
 H.-J. Bolle, et al., 2003, Ann. Geophys.,11, 173-189.
 A. D. Culf,, T. Foken, J. H. C. Gash, 2004, in: Vegetation, water, humans and the climate. A new perspective on an interactive system, P. Kabat, et al., Eds., 159-166.
 J. J. Finnigan, et al., 2003, Boundary-Layer Meteoro.l, 107, 1-48.
 T. Foken, S.P. Oncley, 1995, Bull. Am. Meteorol. Soc., 76, 1191-1193.
 T. Foken, et al., 2004, in: Handbook of Micrometeorology: A Guide for Surface Flux Measurement and Analysis, X. Lee, W.J. Massman, B. Law, Eds., 181-208.
 T. Foken, et al., 2005, Atmos. Chem. Phys. Discuss., submitted.
 B. G. Heusinkveld, et al., 2004, Agric Forest Meteorol, 2004. 122: p. 21-37.
 E. T. Kanemasu, et al., 1992, J. Geophys. Res., 97, 18.547-18.555.
 M. Mauder, et al., 2005, Boundary-Layer Meteorol., 2006: revised.
 M. Mauder, et al., 2006, Theor. Appl. Climat., submitted.
 M. Mauder, et al., 2006, Boundary-Layer Meteorol., submitted.
 J. Moncrieff, H., 2004, in: Vegetation, water, humans and the climate. A new perspective on an interactive system, P. Kabat, et al., Eds., 173-182.
 S. P. Oncley, et al., 1990, 9th Symp on Boundary Layer and Turbulence. 237-240.
 S. P. Oncley, et al., 2006, Boundary-Layer Meteorol., submitted.
 G. N. Panin, G. Tetzlaff, A. Raabe, 1998, Theor. Appl. Climat., 60, 163-178.
 R. J. Pearson Jr., S. P. Oncley, A. C. Delany, 1998, J. Geophys. Res., 103 (D15), 18919-18926.
 J. Ruppert, C. Thomas, T. Foken, 2006, Boundary-Layer Meteorol., accepted.
 R. L. Tsvang, et al., 1991, Boundary-Layer Meteorol., 55, 141-160.
 K. B.Wilson, et al., 2002, Agric. Forest Meteorol., 113. 223-234.