The effect of clouds on the computation of longwave radiation budget parameters is studied using two sets of cloud fields that are obtained from two different satellite systems using two different retrieval schemes: the High Resolution Infrared Sounder 2 (HIRS2)/Microwave Sounding Unit (MSU) retrievals at Goddard Laboratory for Atmospheres (GLA) and the International Satellite Cloud Climatology Project (ISCCP) products at Goddard Institute for Space Studies. The 60% global mean coverage from the ISCCP-produced cloud amount is much larger than the 43% global mean coverage from the HIRS2-retrieved cloud amount. Despite this fact, the ISCCP-derived outgoing longwave radiation (OLR) is very close to the HIRS2-derived OLR; the global mean OLR difference between the two data sources is -2.6 W m-2, with a standard deviation of 7 W m-2. The small difference is due to the higher cloud-top pressure obtained in the ISCCP data, which tends to counterbalance the differences in coverage between the ISCCP and HIRS2 data sets. However, the produced surface downward flux using the ISCCP-derived cloud fields is much larger than that produced from HIRS2 (the global mean difference is around 11.7 W m-2, with a standard deviation of 13 W m-2), reflecting the differences in cloud coverage of the two data sets.

The effect of temperature-humidity fields on the computation of longwave radiation budget parameters is studied using five sets of temperature-humidity fields produced at different agencies which use different retrieval systems/assimilation systems (HIRS2 retrievals at GLA, TIROS Operational Vertical Sounder retrievals at the National Environmental Satellite Data Information Service, European Centre for Medium-range Weather Forecasts assimilation, National Meteorological Center assimilation, and GLA assimilation). Despite all differences found in the geophysical parameters, the derived OLR fields are close to each other. The global mean differences among these five OLRs vary from 0.28 to 4.4 W m-2 with a standard deviation ranging from 4.2 to 7 W m-2. However, large differences are found in the computation of surface downward flux. The global mean differences vary from 3.5 to 19.4 W m-2, with standard deviations ranging from 12.3 to 20.6 W m-2. Most discrepancies in the OLR fields using different cloud fields and geophysical parameters are less than 10 W m-2, which is within the accuracy of current satellite techniques. The discrepancies in the surface downward flux are much larger than 10 W m-2. ¿1991 American Geophysical Union