Monthly ENSO Forecast Skill and Lagged Ensemble Size.

Area of Science:

• Climate Science
• Meteorology
• Oceanography

Background:

• The Niño 3.4 index is crucial for monitoring El Niño-Southern Oscillation (ENSO) events.
• Accurate climate forecasting relies on optimizing ensemble prediction systems.
• The Climate Forecast System version 2 (CFSv2) is a key operational climate model.

Purpose of the Study:

• To determine the optimal ensemble size and configuration for CFSv2 monthly forecasts of the Niño 3.4 index.
• To investigate the relationship between ensemble size, initialization frequency, and forecast accuracy (Mean Square Error - MSE).
• To compare the skill of different ensemble types, including lagged, weighted, and burst ensembles.

Main Methods:

• Analysis of Mean Square Error (MSE) for CFSv2 lagged ensembles.
• Parametric modeling of error covariances to extrapolate MSE for various ensemble sizes and initialization frequencies.
• Comparison of forecast skill across different ensemble configurations.

Main Results:

• The MSE for Niño 3.4 index forecasts consistently minimized with a lagged ensemble size of 1-8 days when using four initializations per day.
• The optimal ensemble size identified is consistent with the operational 8-10 day lagged ensemble configuration.
• The skill of both the identified optimal and operational configurations approached the skill of an infinite ensemble.
• Weighted, lagged, and burst ensembles demonstrated comparable skill levels.

Conclusions:

• A lagged ensemble size of 1-8 days offers optimal performance for CFSv2 monthly Niño 3.4 index forecasts.
• Operational ensemble configurations are well-aligned with findings for maximizing forecast skill.
• Further improvements may require addressing issues related to climatology and data discontinuities in error estimation.