Characterization of Non-El Niño Induced Dry Conditions across the U.S. Affiliated Pacific Islands.

Abstract

The U.S. A liated Paci c Islands (USAPIs), located in the Western Paci c, have limited water resources making them very susceptible to severe drought conditions. The annual cycle and ENSO response of rainfall di ers between USAPIs north of 7 N and those to the south. Southern stations show a canonical negative correlation between dry season (December to May) rainfall and ENSO. Northern stations, on the other hand, show little correlation with ENSO if the three super El Ni~nos are excluded. Instead, they exhibit two distinct rainfall regimes, the Canonical regime, and a Non-Canonical regime, in which the dry season rainfall is positively correlated with ENSO. Non-canonical years pose an important forecasting challenge. Cool Dry events are of particular interest because they have coincided with several emergency and disaster-level droughts across the Northern USAPIs. Composite analysis of the Canonical and Non-Canonical regimes show stark di erences between dry season atmospheric and SST patterns. Compared to Canonical composites, the Non-Canonical composites show clear and previously undescribed anomaly patterns during the dry season. In Cool Dry events, circulation anomalies over the Western Paci c are anticyclonic, with a band of anomalous dry conditions extending from the central Paci c towards Micronesia that causes unexpected droughts in the Northern USAPIs. Canonical Cool Wet events, on the other hand, show cyclonic West Paci c circulation anomalies and a La Ni~na like horseshoe rainfall pattern over the Paci c Basin. Non-canonical Cool events also show SST anomalies narrowly constrained near the dateline, while Canonical Cool events show their largest anomaly magnitude east of the dateline. Both Non-Canonical and Canonical Cool events show negative rainfall and Western Paci c anticyclonic anomalies before the onset of the Dec-May dry season. In Non-Canonical events, these anomalies persist throughout the dry season, while for Canonical events they shift, rapidly becoming positive rainfall and cyclonic circulation anomalies during the dry season. SST anomalies also evolve di erently, with Non-Canonical Cool events showing anomalies that extend eastward from the central Paci c rather than intensifying in place over the eastern Paci c. The features are similar and opposite for Canonical and Non-CanonicalWarm events. Di erences in the evolution of anomalies suggest that the physical mechanisms governing the Non-Canonical and Canonical ENSO regimes are distinct. These di erences have been leveraged to develop a novel 2-tier forecasting methodology that combines logistic and linear regression to forecast the Dec-May dry season Standardized Precipitation Index in the Northern USAPIs. This 2-tier methodology achieves signi cant improvement in the forecast of Dec-May rainfall anomalies as compared to a benchmark forecast. This type of improved forecasts will help provide local governments and decision makers with guidance for mitigation and relief during Non-Canonical events.