The analyses conducted with data from many of the recent large scale field experiments have concentrated on land-atmosphere interaction, and on methodologies of going from local to regional scales. The integration of hydrologic processes into the regional climate models using observations from these experiments is difficult due to the relatively short time scales of the observation periods and logistical constraints on the size of the study areas. Longer term studies have been conducted like ARME and SEBEX over representative biomes, but the spatial variation in precipitation and soil moisture that typically exists over large areas may limit the usefulness of these types of studies to only special regions having relatively weak precipitation gradients.
Remote sensing may help bridge the gap between the logistical constraints on the size of the area and over what time scale ground truth measurements are collected and what is needed for testing climate and large scale hydrologic model simulations. Bolle et al. (1993) emphasize the role of satellite measurements as a potential bridge between the length scales at which atmospheric models operate and that of field observations used for validation. The degree to which validation methods developed for specific areas can be transferred to other regions will dictate the role these experiments will have on advancing large scale hydrologic and atmospheric modeling. Furthermore radar measurements of precipitation from new systems like NEXRAD may provide the spatial rainfall data necessary for large scale hydrologic modeling.
The field experiments conducted thus far indicate that measurement of the surface energy balance using most micrometeorological techniques will typically yield a 20% variation in the fluxes. Aircraft measurements seem to generally underestimate the ground-based estimates, although it has been shown that they do provide very useful information for investigating the spatial variability in fluxes. These results suggest that variations of less than 20% in the fluxes cannot be easily distinguished from measurement errors. Probably most troublesome is the fact that net radiation measurements are only accurate to within 10%. This is the largest term in the energy balance equation and drives both hydrologic and atmospheric processes. Clearly more work is needed to improve the reliability of ground- and aircraft-based flux measurements.
However, the data from these experiments are the most comprehensive and best available to improve parameterization schemes in models coupling hydrologic and atmospheric processes. Within the GEWEX program several initiatives are underway to improve land surface physics and data assimilation in mesoscale models such as the U.S. National Meteorological Center (NMC) mesoscale model (Mitchell et al., 1994). While an integral part of this effort will be to evaluate monthly surface water budgets for 55 catchments in southeastern United States using historical precipitation and runoff data, an important test of the land surface scheme will be performed by comparing a several month simulation for the FIFE site. Comparisons will be made with surface fluxes, soil moisture and soil temperature data over a period encompassing several significant rainfall events. Changes to the parameterizations in the European Centre for Medium-Range Weather Forecasting (ECMWF) model described by Beljaars et al. (1993), show that modifications to the land hydrology component influences the evaporation rate under wet and dry conditions and the runoff component. These modifications were made as a result of comparisons of model simulations with FIFE data.
Improvements in the land hydrology component of weather prediction models give more credibility to studies like Betts et al. (1994) who investigated the coupling between land-surface and atmospheric boundary layer processes using the ECMWF model as a function of scale, namely from local to regional and diurnal to seasonal. Analyses of the monthly precipitation forcasts for July of 1993 using different soil moisture conditions indicate that regional soil moisture may have played a role in the 1993 Mississippi River Flood. The model simulations showed that initial soil moisture had a significant impact on monthly and seasonal precipitation. This result was also supported by Beljaars et al. (1993) where they found the coupling of the atmospheric boundary layer and precipitation has a long time scale. They suggested that this may be useful in predicting monthly and seasonal precipitation fields at continental scales.
The GEWEX Scientific Steering Group has endorsed four regional GEWEX continental-scale projects (GEWEX NEWS, 1994). They include the Large-scale Atmospheric Moisture Balance of Amazonia activity and the Biosphere-Atmosphere Transfer and Ecological Research studies (LAMBADA-BATERISTA), the Mackenzie GEWEX Study (MAGS), the GEWEX Asian Monsoon Experiment (GAME) and the Baltic Regional Experiment (BALTEX). These projects should greatly enhance research related to large scale hydrology and the coupling with climate.
Acknowledgments. The author would like to thank Drs. Jerry Ritchie, Michael Schmidt and Charles Laymon for their helpful comments on an early version of this manuscript.