Local decadal prediction according to statistical/dynamical approaches

Area of Science:

• Climate Science
• Meteorology
• Oceanography

Background:

• Dynamical climate models struggle with initialization due to limited deep ocean data.
• Teleconnection indices, derived from climate variable patterns, can enhance model assimilation.
• Near-term climate predictability is crucial for effective adaptation strategies.

Purpose of the Study:

• To analyze the near-term climate predictability of temperature and precipitation for Barcelona, Bristol, and Lisbon.
• To compare a statistical-dynamical approach using teleconnection indices with drift-corrected dynamical model outputs.
• To assess the effectiveness of teleconnection indices in improving decadal climate projections.

Main Methods:

• Utilized Fourier and wavelet analyses to evaluate the predictability of seven teleconnection indices.
• Employed a cross-validation process with distinct training and validation periods.
• Compared teleconnection-based predictions against a multi-model ensemble from the Coupled Model Intercomparison Project Phase 5.

Main Results:

• Teleconnection-based approaches provide adequate decadal predictions (20-30 years) for both temperature and precipitation.
• Temperature is better predicted within a 5-year horizon using drift-corrected dynamical model outputs.
• The statistical-dynamical method demonstrated superior performance for longer-term predictions.

Conclusions:

• Teleconnection indices are valuable tools for improving the initialization and accuracy of climate models.
• A combination of teleconnection indices and long-term projections enhances local-scale climate predictability.
• The study highlights the potential of teleconnection-based methods for reliable decadal climate forecasting.