ENSO Influence on Intraseasonal Extreme Rainfall and Temperature Frequencies in the Contiguous United States: Implications for Long-Range Predictability

Abstract

Potential ENSO-related predictability of wintertime daily extreme precipitation and temperature frequencies is investigated. This is done empirically using six decades of daily data at 168 stations distributed over the contiguous United States. ENSO sensitivity in the extreme ranges of intraseasonal precipitation and temperature probability density functions is demonstrated via a compositing technique. Potential predictability of extremes is then investigated with a simple statistical model. Given a perfect forecast of ENSO, the frequency of intraseasonal extremes is specified as the average frequency of occurrence during similar-phased ENSO winters on record. Specification skill is assessed as the cross-validated proportion of local variance explained by this method. The skill depends on varying ENSO sensitivity in different geographic regions and quantile ranges and on consistency or variability from one like-phased ENSO event to another. ENSO sensitivity also varies according to the intensity of the tropical forcing; however, not always in the expected sense. Good predictability is likely for variables and in regions displaying a strong and consistent ENSO signal. This is found in some coherent regions of the United States for various combinations of frequency variable and ENSO phase. ENSO-based predictability of heavy and extreme precipitation frequency is potentially good along the Gulf Coast, central plains, Southwest, and in the Ohio River valley for El Niño winters and in the Southwest and Florida for La Niña winters. Not all large magnitude signals translate into significant specification skill. Extreme precipitation frequency in the Southwest is a good example of this. Extreme warm temperature frequency (EWF) is potentially predictable in the southern and eastern United States during El Niño winters and in the Midwest during the strongest events. La Niña winters exhibit potentially very good EWF predictability in a large area of the southern United States centered on Texas. Despite showing coherent ENSO patterns, extreme cold temperature frequency (ECF) signals are mostly weak and inconsistent, especially during strong ENSO events. Curiously, specification skill improves in the northern United States, along the West Coast and in the southeast during weaker El Niño winters. An improvement in potential ECF predictability is also observed in the Midwest during weaker La Niña winters.