Abstract
Distinctive annual weather patterns in the Western Kenya Highlands have been attributed to El Niño Southern Oscillation (ENSO) teleconnections. A novel application of a recently developed analytical approach is used to identify statistically significant differences between temperature and rainfall patterns in the Kakamega District during the El Niño and La Niña periods of the ENSO cycle. This approach separates the seasonal trend and 1-day autocorrelation from the statistical noise in an annual data set. The standard deviation of this noise is further analyzed for its own seasonal trend. Thirty-eight years of reanalysis data are analyzed, and statistical comparisons are made on all three aspects of this analysis. El Niño years were characterized by a phase shift in temperature patterns. Larger random variation was detected in El Niño years during the long rains than in La Niña years, leading to a higher probability of anomalously high rainfall. Larger random variation was detected in La Niña years during the short rains, leading to both a higher probability of anomalously high rainfall and a higher probability of no rainfall. The method appears to be a promising tool for analyzing not only the effects of distant teleconnections but also the nature of those effects.