Premium
A new analysis of variability and predictability of seasonal rainfall of central southern Africa for 1950–94
Author(s) -
Mwale Davison,
Yew Gan Thian,
Shen Samuel S. P.
Publication year - 2004
Publication title -
international journal of climatology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.58
H-Index - 166
eISSN - 1097-0088
pISSN - 0899-8418
DOI - 10.1002/joc.1062
Subject(s) - teleconnection , empirical orthogonal functions , climatology , predictability , sea surface temperature , mode (computer interface) , indian ocean , environmental science , spatial ecology , oceanography , geology , el niño southern oscillation , biology , ecology , physics , quantum mechanics , computer science , operating system
Using wavelet analysis and wavelet‐based empirical orthogonal function analysis on scale‐averaged‐wavelet power and individual scale power, we identified the non‐stationary sea‐surface temperature (SST) fields of the South Atlantic and Indian Oceans that are associated with coherent regions of rainfall variability in central southern Africa (CSA). The dominant mode of CSA rainfall is out of phase between the coastal areas and the centre of CSA and has been decreasing consistently since 1970. The frequencies associated with this mode are between 2–2.4 and 5.6–8 years. The Benguela ocean current SSTs form the dominant spatial pattern of the South Atlantic Ocean, and the Brazil and Guinea ocean current SSTs form the second leading mode. The Benguela spatial patterns were found to migrate seasonally between Africa's west coast and South America's east coast. The northern Indian Ocean SST forms the leading mode of variability, followed by the south Indian Ocean SST. Using predictor fields identified from both oceans, we achieved encouraging results of predicted CSA rainfall using a non‐linear statistical teleconnection artificial neural network–genetic algorithm model. At 3 month lead time, correlations of between 0.8 and 0.9, root‐mean‐square errors of between 0.4 and 0.9 and Hansen Kuipers skill scores of between 0.4 and 0.8 were obtained between observed and predicted CSA rainfall. Copyright © 2004 Royal Meteorological Society