资源环境科技发展态势分析平台

Factors governing spatiotemporal variations of the daily outgoing longwave radiation (OLR) are studied using 35-yr (1979-2013) data records by employing multiple linear regression, wavelet transforms, and bandpass filtering methods. From the regression coefficients of nine predictors and the explained variances, we found that the largest contributions to OLR variability are associated with the Madden-Julian oscillation and El Nino-Southern Oscillation (ENSO). The ENSO signatures on OLR show dipole patterns over the Maritime Continent (MC) and Pacific regions with an extension to the Atlantic. Subsequently, the third significant contribution of the Indian Ocean dipole is confined to the Indian Ocean and Africa. Then, the solar cycle and stratospheric aerosols show mainly negative correlations, while a positive linear trend is observed mainly in the Northern Hemisphere. Lastly, factors associated with the stratospheric quasi-biennial oscillation (QBO) are the least significant contributor to OLR. In terms of oscillatory signals, time-longitude variations of the annual cycle (AC) show pairs of contrasting phases that characterize monsoon systems, in which the MC and Pacific regions are found to be in the same phase group. The most consistent AC signals are found to correspond with North and South American monsoons that respectively exhibit weakening and strengthening trends. Wavelet spectra and filtered OLR signals in intraseasonal oscillation, QBO, and ENSO frequency bands show an interdependent relationship that largely varies with time scale and longitudes.