Long-Term Retrospective Forecasts with CFS: Predictability of ENSO and DroughtMark Cane, Dake Chen and Alexey KaplanLamont-Doherty Earth Observatory of Columbia UniversityWanqiu Wang and Yan XueNational Centers for Environmental PredictionThe 4th Annual CTB SAB MeetingSeptember 11, 2008
Outline • Background and Rationale • Objectives and Hypotheses • Approach and Work Plan • Summary and Discussion
Chen et al., Nature, 2004
Cook et al., GRL, 2008
Objectives • Develop coupled data assimilation and model initialization procedure for the CFS; • Generate retrospective forecasts for the past one and a half centuries with the CFS; • Evaluate the predictability of ENSO and drought using the resulting datasets; and • Provide the basis for bias correction of operational seasonal climate forecasts.
Hypothesis I The CFS can be initialized in a coupled manner by assimilating SST and SLP data over the past one and a half centuries; the coupled initialization run and the subsequent retrospective forecasts are realistic enough (at least) for ENSO and drought studies.
CDA vs. ODA atmosphere atmosphere ocean ocean initialization prediction
OBSEREVD AND PREDICTED NINO3.4 SSTA Chen et al., Science, 2004
Hypothesis II The large-scale extra-tropical drought event, especially that over North America, is strongly influenced by the variability of the coupled tropical climate system; certain drought patterns are predictable if the corresponding SST patterns are predictable.
Cook et al., Earth-Science Reviews, 2007
Approaches • develop an appropriate coupled data assimilation procedure by analyzing the existing CFS free runs and retrospective forecasts for the modern era (1981-2004); • test the procedure for the modern era and then use it for a coupled initialization run for the past 150 years; • perform retrospective forecasts for lead times up to 9 months starting twice a year from each summer and winter of the past 150 years; • assess model skills in predicting ENSO and drought using various techniques.
Summary and Discussion • The purpose of this project is to generate and evaluateretrospective forecasts with CFS over the past 150 years to study thepredictability of ENSO and drought. • The key for the success of this study is the development of a coupled data assimilation and model initialization procedure using only SST and SLP observations. • Computer resources are the most severe limitation for carrying out the whole suite of experiments, but even a subset of them will prove to be extremely useful.