Taiwanese rainfall variability associated with large-scale climate phenomena

Examining the physical mechanisms through which large scale climate indicators, e.g., El Nirio-Southern Oscillation and Indian Ocean Dipole, affect hydroclimatic variables in the tropics and extratropics is a scientific challenge. In this study, climatic relationships between large-scale climate indices (CIs) and Taiwan rainfall variability were examined. Not only leading patterns of observed monthly total and extreme rainfall were estimated through an empirical orthogonal teleconnection method (EOT), but also correlation and regression analyses were performed for the leading rainfall patterns and various CIs based on atmospheric-ocean circulation dataset. From the spatial structure of the leading EOT patterns for total and extreme rainfall, the north-coast mode in the cold seasons and the middle-inland mode for the warm seasons were detected. The temporal cycle of the leading EOT patterns indicates decreasing trends in the cold seasons and oscillations mainly on decadal timescales in the warm seasons. The leading EOT patterns of total rainfall showed more widespread coherent patterns than those of extreme rainfall with more variance in rainfall variability. The findings from this study illustrate that tropical ENSO forcing has a coherent association with March and October-November rainfall patterns, while the Indian Ocean Dipole is identified as a driver for rainfall variability during spring and fall. The western North Pacific monsoon activity has a negative (positive) correlation with March (October and December) rainfall variabilities and tropical cyclone indices also exhibit significant positive correlation with October rainfalls. The leading patterns of the October and following-year March rainfall time series are predictable at up to nine-month lead time from the tropical Pacific sea surface temperatures (SSTs).