Capturing climatic variation from boreal winter to spring (December to May) is essential for the proper prediction of drought in South Korea. This study uses observed datasets and multi-model ensemble (MME) hindcast experiments (1983/84-2005/06) archived at the APEC Climate Center (APCC) to investigate the variability and predictability of the South Korean climate during the period December-May. It focuses on the leading modes of winter-to-spring variability over South Korea, which are identified based on multivariate empirical orthogonal function (EOF) analysis of the 6-month accumulated precipitation and surface air temperature. The first two leading climate modes, accounting for about 80% of total variance, are characterized by national-scale precipitation and temperature anomalies covering an entire region over South Korea.

A perfect empirical model was developed to reveal atmospheric dynamic linkage (based on the observed datasets), and be used as a standard reference to recognize its physically potential predictability. The potential of using a hybrid dynamical-statistical method for a 6 month-lead (with November as the initial condition) drought prediction was also investigated. Nine one-tier climate models were statistically downscaled for predicting the standardized precipitation evapotranspiration index (SPEI) over 60 stations in South Korea. This bivariate and pattern-based downscaling was employed for both precipitation and temperature, to maintain a physically coherent relation between these factors, and a spatial coherence over the 60 station locations. This study has developed a new downscaling method using a variant of canonical correlation analysis (CCA). Precipitation and temperature predictions, based on dynamically better-predicted sea level pressure (SLP) and 500hPa geopotential height (Z500) were substantially improved in the downscaled MME (DMME). The limitations of the hybrid dynamical-statistical model, and possible causes for these in the current framework of dynamical climate prediction were also discussed for further improvement.

Overall, DMME gives reasonably skillful long-lead forecasts of SPEI for the period winter-spring, compared to raw MME. Our results could lead to more reliable climatic extreme predictions for policymakers and stakeholders in the water management sector, and for the mitigation of problems related to climate change.

1. INTRODUCTION

2. DATA AND METHODOLOGY

2.1 Data

2.2 Methodology

3. RESULTS of RESEARCH

3.1 Dominant modes of winter-to-spring climate variability in South Korea

3.2 Development of a dynamical-statistical prediction model for the winter- to-spring climate

3.3 Prediction and predictability of winter-to-spring drought

4. SUMMARY AND DISCISSION