Net radiation is usually the largest component of the surface energy balance and is typically measured with a single instrument sensitive to the electromagnetic radiation at all wavelengths. Comparisons of measured net radiation using instruments from different manufacturers by Goutorbe (1991) with Hydrological Atmospheric Pilot Experiment-Modelisation du Bilan Hydrique (HAPEX-MOBILHY; André et al., 1986) data, by Field et al. (1992) and Nie et al.(1992) using First ISLSCP Field Experiment (FIFE; Sellers et al, 1988) data and by Stannard et al. (1994) with Monsoon '90 data (Kustas et al., 1991) showed variations of around 10%. This degree of discrepancy is rather serious because of the relative contribution of net radiation to the surface energy balance and the fact that logistically the measurement of net radiation is the easiest to make of the four components (i.e., net radiation, soil heat flux, sensible heat flux and latent heat flux).
How net radiation is partitioned into the turbulent fluxes of sensible and latent heat may have major implications to regional hydrology and climate (e.g., Segal and Arritt, 1992). Measurement of fluxes by direct (i.e., eddy correlation technique) or indirect (e.g., Bowen ratio, variance and aerodynamic techniques) methods have been used (Dabberbt et al., 1993). From intercomparisons of various methods it appears that one can expect about a 20% variation in flux estimates for short term measurements (say half-hourly). These results come from the same field experiments mentioned above for net radiation ( André et al., 1990; Goutorbe, 1991; Nie et al., 1992; Stannard et al., 1994). For longer time scales, the variation probably reduces somewhat. For example, André et al. (1988) found better than 10% agreement between water balance measurements collocated with the aerodynamic technique used in HAPEX-MOBILHY on a monthly time scale. This result is promising since for many hydrological applications, longer term water balance data is most appropriate for evaluating hydrologic models.
Turbulent flux measurements from aircraft-based sensors have been made in several of these experiments. The rationale for these measurements is that it may be the only technique for assessing experimentally the impact that spatial variation in surface fluxes have on regional scale atmospheric flow and potential feedbacks to land-surface processes. This includes the investigation of aggregation schemes for land-surface parameters and fluxes, which has also involved the integration of remote sensing information from studies like HAPEX-MOBILHY (e.g., Noilhan et al., 1991a). Indeed, patterns of surface wetness, vegetation cover, surface temperature and albedo have been related to remotely sensed data (Choudhury, 1991), and efforts to use this type of information collected from large scale field experiments such as FIFE and Monsoon '90 for quantifying the surface energy balance have been explored (Hall et al., 1992; Kustas and Goodrich, 1994). However, the role these surface properties have in affecting local and regional weather and hydrology have not been adequately addressed.
Results from HAPEX-MOBILHY (André et al., 1990), FIFE (Kelly, 1992) and LOTREX (LOngitudinal land-surface TRansverse EXperiment) (Jochum et al., 1990) indicate systematic underestimates of the surface fluxes by 10% to 30%. Although the aircraft fluxes may not yield reliable actual surface fluxes, they have been valuable in evaluating effective parameters for mesoscale modeling (Mahrt and Ek, 1993) and in evaluating the relationship between land-use, scaling of fluxes and implications to atmospheric boundary layer processes (André et al., 1990). Consequently, there appears to be strong evidence that these measurements play an important role in large scale field experiments carried out over heterogeneous landscapes.
From both ground and aircraft observations of the
turbulent fluxes, some important conclusions about the
scaling of land-atmosphere interaction have been documented,
including the development of a conceptual framework for
treating surface heterogeneity (Shuttleworth, 1988).
Shuttleworth (1988) defined two types of non-homogeneous
land surfaces, which were inspired by observations from
HAPEX-MOBILHY. One surface type would have horizontal
inhomogeneities with length scales significantly smaller
than 10
km, resulting in the planetary boundary layer
turbulence mixing all discontinuities. The other type would
contain landscape features having large discontinuities at
length scales greater than 10
km. The result would be the
development of coherent atmospheric structures at the
mesoscale. Atmospheric modeling studies using satellite
observations seem to support the concept of coherent
atmospheric structures developing from large discontinuities
in vegetation and surface moisture conditions (e.g., Rabin et
al., 1990). However, analysis of atmospheric boundary
layer (ABL) data collected during the Monsoon '90 experiment
indicated that surface moisture variations having length
scales on the order of 10 km were not sufficient to effect
regional scale processes (Hipps et al., 1994).
Preliminary analysis of aircraft and ground data from the ECHIVAL Field Experiment in a Desertification-threatened Area, EFEDA, (Jochum, 1993) suggest simple weighting of evaporative fluxes to derive area-averaged values is not generally applicable. A similar conclusion came from an analysis of ground-based flux data from FIFE-89 where nonuniform surface conditions existed, which was caused by sparse and infrequent precipitation (Brutsaert and Sugita, 1992). They compared regional fluxes estimated by aggregating surface flux measurements from a network of surface flux stations with those derived from a procedure using ABL data and remotely sensed surface temperature. Under these nonuniform conditions, it was difficult to aggregate the array of surface flux stations so that regional value was in consistent agreement with the ABL method. Hence more complicated techniques of scaling hydrologic and atmospheric fluxes and processes in space will be required since many land surfaces are not homogeneous at the regional scale.