Cloud, Surface Temperature, and Outgoing Longwave Radiation for the Period from 1979 to 1990

Abstract

Quasi-biennial global, midlatitude, and tropical oscillations were observed using top-of-the-atmosphere outgoing longwave radiation (OLR), surface air temperature (SAT), and cloud amount for the period from 1979 to 1989. The in-phase quasi-biennial variations of OLR and SAT were strongest in the Tropics. Two prominent peaks in these two data fields were observed after the end of the main phases of the 1982?83 and 1986?87 El Niño-Southern Oscillation (ENSO) events, which were also accompanied by a decrease in the mean tropical cloud cover. The quasi-biennial signal was less noticeable in the midlatitudes during the two ENSO events but was strong during two non-ENSO peaks occurring in 1980?81 and 1989?90. In this study, the authors used two SAT datasets comprised of departures estimated from a specific base period, where the record of these two datasets predates the start of this century. The OLR dataset was obtained by concatenating Nimbus-7 (1979?87) and Earth Radiation Budget Satellite (ERBS) (1985?89) measurements. The cloud dataset was generated by concatenating Nimbus-7 (1979?84) estimates with those from the International Satellite Cloud Climatology Program (ISCCP) (1983?90). In the concatenation procedure, adjustments were made for previously identified, long-term nonphysical data trends in the Nimbus-7 datasets; sonar additional experiments were made in which detrending was applied to all the datasets. As a consequence, decade-long trends were not considered. In the detrended datasets, OLR and SAT were strongly positively correlated with explained variances of 74.5% or larger in the Tropics and midlatitudes and of 96.4% on a global scale. OLR and cloud were negatively correlated; however the results were less definitive with explained variances of 36.7% and 79.4%, respectively, for the globe and midlatitudes but only 17.4% in the Tropics. These observations imply a clear identification of a quasi-biennial signal and relationship between OLR and SAT, but the results are less certain when OLR is compared to cloud cover.