Integrating subsurface ocean temperaturesin the statistical prediction of ENSO and Australian rainfall & streamflow

As a global climate phenomenon, the El Niño-Southern Oscillation (ENSO) involvesthe coupling of the ocean and the atmosphere. Most climate prediction studies have, byfar, only investigated the teleconnections between global climatic anomalies and the‘surface’ predictors of ENSO. The prediction models resulting from these studies havegenerally suffered from inadequate, if not the lack of, skill across the so-called boreal‘spring barrier’. This is illustrated in the first part of this thesis where the applicabilityof the SOI phase for long-lead rainfall projections in Australia is discussed. With theincreasing availability of subsurface ocean temperature data, the characteristics of thePacific Ocean’s heat content and its role in ENSO are now better understood. Thesecond part of this thesis investigated the predictability of ENSO using the thermoclineas a predictor. While the persistence and SST-based ENSO hindcasts dropped in skillacross the spring barrier, the thermocline-based hindcasts remained skillful even up to alag of eighteen months. Continuing on the favorable results of ENSO prediction, thethird part of this thesis extended the use of the thermocline in the prediction ofAustralia’s rainfall and streamflow. When compared to models that use ‘surface’predictors, the model that incorporated thermocline information resulted in more skillfulprojections of rainfall and streamflow especially at long lead-times. More importantly,significant increases in skill of autumn and winter projections demonstrate the ability ofthe subsurface ocean to retain some climatic memory across the predictability barrier.This resilience can be attributed to the high persistence of the ocean heat content duringthe first half of the year. Based on weighting, the model averaging exercise alsoaffirmed the superiority of the ‘subsurface’ model over the ‘surface’ models in terms ofstreamflow projections. The encouraging findings of this study could have far-reachingimplications not only to the science of ENSO prediction but also to the more pragmaticrealm of hydrologic forecasting. What this study has demonstrated is an alternativepredictor that is suitable for the long range forecasting of ENSO, rainfall andstreamflow. With better hydrologic forecasting comes significant improvement in themanagement of reservoirs which eventually leads to an increase in the reliability andsufficiency of water supply provision