AREA-AVERAGED SURFACE FLUXES AND THEIR TIME-SPACE VARIABILITY OVER THE FIFE EXPERIMENTAL DOMAIN

The underlying mean and variance properties of surface net radiation, soil heat flux, and sensible-latent heat fluxes are examined over the densely instrumented grassland region encompassing the First ISLSCP Field Experiment (FIFE). Twenty-two surface flux stations at 20 sites were deployed during the four 1987 intensive field campaigns (IFCs). Flux variability is addressed together with the problem of scaling up to area-averaged fluxes. Successful parameterization of area-averaged fluxes in atmospheric models is based on accounting for internal spatial and temporal scales correctly. Mean and variance properties of fluxes are examined in both daily and diurnally averaged frameworks. Results are compared and contrasted for clear and cloudy situations and checked for the influence of surface-induced biophysical controls (burn and grazing treatments) and topographic controls (slope factors and aspect ratios). Examination of the sensitivity of domain-averaged fluxes to different averaging procedures demonstrates that this may be an important consideration. The results reveal six key features of the 1987 surface fluxes: (1) cloudiness variability and ample rainfall throughout the growing season led to near-consistency in flux magnitudes during the first three IFCs; (2) burn treatment, grazing conditions, and topography have clearly delineated influences on the diurnal cycle flux amplitudes but do not alter the evaporative fraction significantly; (3) cloudiness is the major control on flux variability in terms of both mean and variance properties but has little impact on the Bowen ratio or evaporative fraction; (4) spatial weighting of fluxes based on a biophysical-topographical cross stratification generates a measurable bias with respect to straight arithmetic averaging (up to 20 W m-2 in available heating); (5) structure function analysis demonstrates significant underlying spatial autocorrelation structure in the fluxes, but the observed distance dependence is due to cloudiness controls, not surface controls; (6) Monte Carlo analysis of high resolution vegetation indices obtained from SPOT satellite measurements suggest that the mean domain amplitudes of the diurnal sensible and latent heat flux cycles can be biased up to 30-40 W m-2 by repositioning the 20 site locations within the experimental domain.