A filtered and standardized surface heat flux data set, which includes net radiation (Rn), latent heat flux (λE), sensible heat flux (H), and ground heat flux (G), was produced from an analysis of field measurements taken by a network of 16 Bowen ratio and 6 eddy correlation instruments deployed during the First International Satellite Land Surface Climatology Project (ISLSCP) Field Experiment in 1987 (FIFE 1987). The objective was to produce time series of flux data that could be used for studies involving surface-vegetation-atmosphere-transfer (SVAT) models operating in the inverse mode (see Sellers et al. 1992b). To be useful for this work, erroneous flux data had to be identified and systematic instrument-related differences between the net radiation values reported by flux stations had to be accounted for. The following procedures were carried out on the original data sets as contributed by individual FIFE scientists: (1) A flag (IDATA) was assigned to unreliable flux data as defined by the presence of anomalies (''spikes'') in the sensible and latent heat flux time series, discontinuities generated when the temperature and humidity profiles crossed over at dawn and dusk, and data points isolated by the erasure of neighboring unreliable data. In addition, Bowen ratio data were also considered to be unreliable whenever the surface energy budget was reported as nonzero (Rn+λE+H+G≠0).

(2) A simple procedure was used to adjust the net radiation values reported by two of the Bowen ratio stations to appear as if they were using the same type of net radiometer as most of the other Bowen ratio stations deployed in FIFE 1987. Corresponding changes were made to the associated sensible and latent heat fluxes for these two stations; the ground heat fluxes were left unchanged. At one eddy correlation site (site 26), the sensible and latent heat fluxes were suspected to be systematically underestimated; these were proportionately increased to match the site net radiation flux minus a FIFE-area-average ground heat flux. (3) An additional flag (IDF) was assigned to data based on modeling results obtained from the simple biosphere (SiB) model (see Sellers et al., 1992b). Nonzero values of IDF denote periods when the flux data time series diverge significantly from a set of calibrated model results, suggesting a bias in the data or indicating that SiB could not reproduce the observed latent and sensible heat flux time series using a reasonable set of surface parameters. An IDF flag was assigned to each flux station for each intensive field campaign (IFC). (4) Using the IDATA and IDF flags to identify and exclude questionable data points, a site-averaged surface heat flux data set was computed for each IFC from the remaining data. This is compared to a similar ''hand-filtered'' data set produced by Betts and Ball as published by Strebel et al. (1994). ¿ American Geophysical Union 1995<