MODIS Cloud Products and Observed Cloud Attenuation of Solar Flux: An Investigation of a Method for Scaling Local Observations to the Regional Level
Joel
Schafer, NASA/GSFC, joel.schafer@gsfc.nasa.gov
(Presenting)
Brent
Holben, NASA/GSFC, brent@aeronet.gsfc.nasa.gov
Tom
Eck, NASA/GSFC, teck@ltpmail.gsfc.nasa.gov
Paulo
Artaxo, USP, artaxo@if.usp.br
We have observed that the solar flux reductions by clouds are strongly correlated with readily available satellite cloud products such as cloud optical thickness and daytime cloud fraction from MODIS, for instance.
We routinely compute monthly averages of daily integrated flux (insolation) for all our sites. Additionally, we use modeled insolation (assuming clear sky, background aerosol) to determine the fraction of expected clear sky insolation received at the surface.
Despite the large distances between our equatorial and southern Amazonian sites, and great difference in predominant cloud regimes, a simple relationship based on data from 2000-2002 was found that could generate satisfactory predictions of monthly averaged daily insolation at all sites for the 2003 observations using only MODIS cloud parameters. This empirical relationship was found to hold even for intervals with moderate smoke levels (Month Average AOT500nm < 0.5), while it tended to over-predict insolation for the few months with the most intense biomass burning, as might be expected. For the 26 months of data from 5 different sites (low to moderate AOT cases), the rms error in predicted daily insolation was 0.7 MJ m^-2, which amounts to about 4% of the typical daily value. The largest error for any month was 10%.
The effectiveness of this simple function suggested the potential suitability of its application over the entire study region encompassing our sites (20 x 20 deg box) and demonstrates an empirically based technique for networks of flux monitoring sites to scale local cloud effects to regional scale flux fields.