+ Models AGMET-3776; No of Pages 24 Comprehensive comparison of gap-filling techniques for eddy covariance net carbon fluxes Antje M. Moffat a, * , Dario Papale b , Markus Reichstein a , David Y. Hollinger c , Andrew D. Richardson d , Alan G. Barr e , Clemens Beckstein f , Bobby H. Braswell g , Galina Churkina a , Ankur R. Desai h , Eva Falge i , Jeffrey H. Gove c , Martin Heimann a , Dafeng Hui j , Andrew J. Jarvis k , Jens Kattge a , Asko Noormets l , Vanessa J. Stauch m a Max-Planck-Institute for Biogeochemistry, Hans-Kno¨ll-Str. 10, 07745 Jena, Germany b DISAFRI, University of Tuscia, via C. de Lellis, 01100 Viterbo, Italy c USDA Forest Service, Northern Research Station, 271 Mast Rd., Durham, NH 03824, USA d Complex Systems Research Center, University of New Hampshire, Durham, NH 03824, USA e Climate Research Division Atmospheric Sciences and Technology Directorate Environment Canada, 11 Innovation Boulevard, Saskatoon, Sask., Canada f Friedrich-Schiller-Universita¨t Jena, Institut fu¨r Informatik, Ernst-Abbe-Platz 1-4, 07743 Jena, Germany g Institute for the Study of Earth, Ocean, and Space, University of New Hampshire Durham, NH 03824, USA h Department of Atmospheric and Oceanic Sciences, University Wisconsin-Madison, 1225 W Dayton St., Madison, WI 53706, USA i Max-Planck-Institute for Chemistry, Biogeochemistry Department, J.J.v. Becherweg 27, 55128 Mainz, Germany j School of Forestry and Wildlife Sciences, Auburn University, Auburn, AL 36849-5418, USA k Environmental Science Department, Lancaster University, UK l North Carolina State University/USDA Forest Service, 920 Main Campus Drive, Venture Center II, Suite 300, Raleigh, NC 27606, USA m Federal Office for Meteorology and Climatology (MeteoSwiss), Zurich, Switzerland Received 11 March 2007; received in revised form 4 August 2007; accepted 14 August 2007 Abstract We review 15 techniques for estimating missing values of net ecosystem CO 2 exchange (NEE) in eddy covariance time series and evaluate their performance for different artificial gap scenarios based on a set of 10 benchmark datasets from six forested sites in Europe. The goal of gap filling is the reproduction of the NEE time series and hence this present work focuses on estimating missing NEE values, not on editing or the removal of suspect values in these time series due to systematic errors in the measurements (e.g., nighttime flux, advection). The gap filling was examined by generating 50 secondary datasets with artificial gaps (ranging in length from single half-hours to 12 consecutive days) for each benchmark dataset and evaluating the performance with a variety of statistical metrics. The performance of the gap filling varied among sites and depended on the level of aggregation (native half- hourly time step versus daily), long gaps were more difficult to fill than short gaps, and differences among the techniques were more pronounced during the day than at night. The non-linear regression techniques (NLRs), the look-up table (LUT), marginal distribution sampling (MDS), and the semi- parametric model (SPM) generally showed good overall performance. The artificial neural network based techniques (ANNs) were generally, if only slightly, superior to the other techniques. The simple interpolation technique of mean diurnal variation (MDV) www.elsevier.com/locate/agrformet Agricultural and Forest Meteorology xxx (2007) xxx–xxx * Corresponding author. Tel.: +49 3641 576314; fax: +49 3641 577300. E-mail address: amoffat@bgc-jena.mpg.de (A.M. Moffat). 0168-1923/$ – see front matter # 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.agrformet.2007.08.011 Please cite this article in press as: Moffat, A.M., et al., Comprehensive comparison of gap-filling techniques for eddy covariance net carbon fluxes, Agric. Forest Meteorol. (2007), doi:10.1016/j.agrformet.2007.08.011