Principal oscillation pattern analysis of the 30- to 60-day oscillation in the tropical troposphere

Abstract

The “Principal Oscillation Pattern” technique is used to derive an index of the 30- to 60-day oscillation in the tropical troposphere. In the 200-mb equatorial velocity potential field, one dominant pair of POPS is found. Its properties compare very well with the properties of the oscillation identified in previous studies. In particular, a good correlation between the time evolution of the POP coefficients and area-averaged outgoing long-wave radiation (ORL) is found. The POPS are derived from a 2-year subinterval of the whole 5-year data set. This leaves independent data for subsequent verification. The patterns and their characteristic numbers are almost unchanged if the whole data set is analysed. Also, the analysis is insensitive to changes of the analysis area: if the analysis is limited to 90°-longitude equatorial sectors, the signal is also identified and its patterns are consistent with the patterns derived from the full data set. Interestingly, the signal is best defined in the eastern hemisphere. The POPS may be used to derive “associated correlation patterns” of other quantities in winter and summer separately. The path of the oscillation has a marked annual cycle: in northern winter it migrates from the Indian Ocean across northern Australia into the region of the South Pacific Convergence Zone (SPCZ) and in northern summer it moves from the Indian Ocean across South Asia along the intertropical Convergence Zone (ITCZ) to South America. The POP coefficient may be seen as a bivariate index of the state (phase and strength) of the 30- to 60-day oscillation. Since the POP technique incorporates a prediction equation for the phase of the POP coefficients, the POP model allows for the prediction of the complex amplitude of the oscillation. In a sequence of forecast experiments, of which about two-thirds used independent data, the POP forecasts were found to be useful in about half of all cases for lead times of several days. The correlation and RMS skills were calculated for the POP forecast and for persistence. The POP forecast appears to be considerably better with respect to both measures. The correlation skill scores 60% after 7 days. The POP forecast is most skillful in northern winter and if strong signals are present with minima of velocity potential in the eastern hemisphere.