Characterizing and forecasting climate indices using time series models

The objective of the current study is to present a comparison techniques for forecasting low-frequency climate oscillation indices with focus on Great Lakes system. A number time series models have been tested including traditional autoregressive moving average (ARMA) model, dynamic linear model (DLM), generalized conditional heteroskedasticity (GARCH) as well nonstationary resampling (NSOR) technique. These were used forecast monthly El Niño-Southern Oscillation (ENSO) and Pacific Decadal (PDO) which show most significant teleconnection net basin supply (NBS) system from preliminary study. overall predict future water levels, ice extent, temperature, planning decision making purposes. results showed that DLM GARCH are superior ENSO index, while forecasted values ARMA presented good agreement observed within short lead ahead PDO index.