The influence of El-Niño Southern Oscillation and Pacific Decadal Oscillation on secular rainfall variations in Hawai'i

Large-scale teleconnections, particularly the El-Niño Southern Oscillation (ENSO) and the Pacific Decadal Oscillation (PDO), have a strong influence on rainfall patterns in Hawai'i. Over the last century, we have observed statistically significant declines in rainfall across the state, and it is unknown whether these declines are due to changes in these natural large-scale variations in climate, or whether these downward trends can be explained by anthropogenic effects. To better aid managers and decision-makers, it is important to understand what is driving current trends. Here we use an empirical approach to study long-term trends in a geographically complex region and diverse climate. Using a time series of month-year rainfall maps for Hawai'i starting in January 1920 at 250 m resolution, an empirical orthogonal function (EOF) analysis was performed to study the spatiotemporal variations and trend patterns. We further correlate the leading spatial and temporal components with ENSO and PDO indices, linear trends, and secular trends. More of the variability is contained in the first component in the winter (December-January-February) than in the summer (June-July-August), especially in the northern islands (Kaua'i and O'ahu) suggesting that natural climate variability has a stronger effect on the spatiotemporal rainfall patterns during the winter season than the summer season. Currently, independent efforts to downscale future climate projections for Hawai'i have produced different future outlooks for rainfall. In the absence of adequately designed control experiments with regional climate models, we propose evaluating differences between observed and projected trend patterns as an alternative criterion for measuring the significance and plausibility of future climate change projections. Our results show the difficulties of separating anthropogenic and natural rainfall trends, e.g., identifying spatial (and seasonal) patterns of the trends that are different from the rainfall modes controlled by ENSO and PDO variability. In future work, we will apply formal detection and attribution methods with the aid of regional climate model simulations.