Utilizing the State of ENSO to Improve Probabilistic Seasonal Forecasts for Temperature and Precipitation
Abstract
This thesis illustrates a novel methodology developed around the ENSO phenomenon -
however, instead of using the ENSO signal itself as a predictor, ENSO is used to bin the
years on record, somewhat analogous to using the seasonal cycle to bin months within
a year. Predictive sea surface temperature anomalies are then elucidated within these
‘phases’ of ENSO, and a marked increase in skill is shown for seasonal precipitation and
temperature forecasts over a wide range of the continental US in nearly every season
as compared to traditional forecasting methods. Skill varies depending on the location,
phase, and season, with temperature forecasts generally showing better skill than precipitation
forecasts. The most interesting results is observed for both temperature and
precipitation during the El Nino phase for most seasons - within El Nino events, the
strength of the El Nino seems to predict temperature/precipitation anomalies in particular
locations with impressive accuracy. Some potential reasons for this are explored.
Subject
temperature
precipitation
statistical models
seasonal forcasting
ENSO