Gap filling strategies for long term energy flux data sets

At present a network of over 100 field sites are measuring carbon dioxide, water vapor and sensible heat fluxes between the biosphere and atmosphere, on a nearly continuous basis. Gaps in the long term measurements of evaporation and sensible heat flux must be filled before these data can be used for hydrological and meteorological applications. We adapted methods of gap filling for NEE (net ecosystem exchange of carbon) to energy fluxes and applied them to data sets available from the EUROFLUX and AmeriFlux eddy covariance databases. The average data coverage for the sites selected was 69% and 75% for latent heat (lambdaE) and sensible heat (H). The methods were based on mean diurnal variations (half-hourly binned means of fluxes based on previous and subsequent days, MDV) and look-up tables for fluxes during assorted meteorological conditions (LookUp), and the impact of different gap filling methods on the annual sum of lambdaE and H is investigated. The difference between annual lambdaE filled by MDV and lambdaE filled by LookUp ranged from -120 to 210 MJ m(-2) per year, i.e. -48 to +86 mm per year, or -13 to +39% of the annual sum. For annual sums of H differences between - 140 and +140 MJ m(-2) per year or -12 to +19% of the annual sum were found. (C) 2001 Elsevier Science B.V. All rights reserved.