The assimilation or interpretation of hydrological or related fields (e.g., vegetation) between models such as MM5 and those derived from remote sensing is complicated by the fact that the scales of the data are generally not commensurate -- this often necessitates some form of aggregation. However, methods for aggregation need to consider the inherent non-linearities that affect the surface energy balance and so, the water balance at the land surface. Aggregation of the Normalized Difference Vegetation Index (NDVI) and the Fractional Vegetation (Fr) cover was investigated using high resolution 10 metre airborne data and coarser resolution satellite data from the Advanced Very High Resolution Radiometer (AVHRR); collected as part of the Southern Great Plains 1997 (SGP97) Hydrology experiment data.

AVHRR data was used to investigate whether the NDVI and Fr fields could be aggregated linearly from 1 km up to 16 km pixel resolutions. Results indicated that linearly averaging the NDVI field results in minimal errors, while averaging the Fr field was unacceptable. However, the calculation of Fr from linearly averaged NDVI was acceptable. Futher, latent heat (LE) fields were derived from a Soil-Vegetation-Atmosphere-Transfer (SVAT) model via an inversion based on surface radiometric temperature and the Fr fields. High resolution airborne estimates of LE compared well with eddy-covariance data collected at the time of overpass while satellite estimates of LE at coarser resolutions compared poorly with the surface measurements. However, after aggregating the high resolution data (Fr and surface radiometric temperature) to the resolution of the satellite, good agreement was observed. Finally, the aggregated remotely sensed vegetation and surface radiometric temperature data were used in conjunction with MM5 to assess any improvement in forcasting ability through the application of a more physically realistic representation of the surface conditions.