Rainfall regimes in Sri Lanka

ABSTRACT Understanding spatio-temporal variation in rainfall is important to water resources planning and management activities. Empirical orthogonal function (EOF) analysis and harmonic analysis are considered state-of-the-art techniques for analysing rainfall variability. This study considered monthly rainfall data from 18 principal stations and 21 secondary stations in Sri Lanka. K-means clustering was applied for identifying rainfall regimes. Rotated EOF analysis provided four key factors related to climate seasons of which the first two factors were more prominent, explaining more than 80% of the total variance. Harmonic analysis showed that northeastern and southwestern parts of the country possess strong annual and semi-annual variation, respectively explaining 90% of the total variance. Both EOF and harmonic analysis can be considered effective tools for identifying rainfall characteristics. However, harmonic analysis results successfully identified seven well-defined rainfall regime characteristics, which could be explained using tendency in data, amplitude, phase angle, and seasonal contribution.