There is now a need to update this data base by incorporating the observational record obtained by the NWS during the decade of the 1990s. However, during this decade, most stations have changed over from a system of hourly or three-hourly human observations to observations made by the Automated Surface Observing System (ASOS). Prior to ASOS, traditional cloud cover observations were made by human observers by dividing the sky vault into 8 regions (4 azimuthal quadrants and 2 zenithal elevations), which then allows the observer to report cloud cover in octas (in the U.S., octas are converted to tenths before reporting). However, at ASOS automated sites, cloud cover measurements are made using a ceilometer that provides a time-continuous stream of data directly overhead of the station. In this case, reported cloud cover octas for a given hour are a function of the fraction of time cloud presence is detected overhead for a 30-minute period prior to the reported observation.
We compared cloud cover observations made at major airports with those derived from ASOS measurements at nearby smaller airports, and found significant biases between the observations in seven regions around the U.S. The biases are particularly prevalent under clear sky conditions, where the major airports indicate a much lower frequency of clear sky observations than the nearby smaller airports. Biases for overcast conditions were much less prevalent, although the trend was for major airports to report a higher frequency of overcast conditions than the nearby smaller airports.
We have determined that a potential cause for these biases may be associated with the different operational procedures between major and smaller airports. ASOS observations at major airports are “augmented” by human observers, which is not the case at the smaller airports. Human observations tend to add clouds to the clear-sky conditions indicated by ASOS. This is possibly due to the access that human observers have to the numerous aircraft-based observations around their station, and also to the fact that they will report any cloud in their field of view that are not directly overhead and not sensed by ceilometers.
These discrepancies have important implications when updating the National Solar Radiation Data Base. This paper will conclude with a discussion on how concurrent satellite-derived irradiance estimates from Geostationary Operational Environmental Satellites (GOES) might be a useful procedure for “normalizing” these cloud cover observation discrepancies across the period of time that ASOS was implemented.