The Estimation of Emitted and Reflected Energy over the Central Pacific Using Meteorological Parameters

Abstract

The authors conducted a quantitative investigation of the relationship between meteorological parameters (chiefly cloud cover) and the principal components of the earth's radiation budget. The area of study is the tropical central Pacific, where considerable variation in cloudiness and flux was observed from 1985 to 1989. The observed variations were in response to the El Niñto-Southern Oscillation event of 1987. International Satellite Cloud Climatology Project (ISCCP)C2 parameters were used as independent variables in the development of multiple linear models to predict Earth Radiation Budget Experiment (ERBE) outgoing longwave radiation (OLR) and shortwave (SW) fluxes. Net radiation estimates were obtained from OLR and SW predicted fluxes. The technique of all-subsets regression was used to determine which combination of ISCCP C2 parameters could best predict OLR and SW fluxes. The models were developed for the years 1985 and 1989 (non-ENSO years) and tested by both month and year on the years 1986 through 1988. Predicted fluxes were obtained for three 15° latitude zones, north (7.5°S to 22.5°N), central (7.5°S to 7.5°N), and south (7.5°S to 22.5°S). Over the 60 months of data, explained variances (R2) of over 90% for the development and test periods were typical. An increase in the accuracy of the GLR models was observed when noncloud variables were included. This accuracy improvement was most apparent when cloud amounts were either very low or very high and homogeneous. Biases (predicted-observed) were all less than 4 Wm?2 and rms estimates were within the range of uncertainty for FRBE monthly mean flux estimates. Flux estimates based on linear models could serve as a means of extending long-term radiation balance datasets during intervals of time with limited satellite coverage.