Skip to main content

AUSPICE logo

Advancing Understanding of the Signal-to-noise Paradox and its Impacts on Climate Ensembles

The challenge

The signal-to-noise paradox in climate forecasting refers to counterintuitive situations where the real world is better predicted than the forecast model suggests.

It arises when predictable mid-latitude atmospheric signals and associated climate impacts are significantly underestimated compared with observations. This discrepancy poses a challenge, as weak signals are statistically problematic to detect, yet underlying dynamical and physical mechanisms, such as remote teleconnections, coupled ocean-atmosphere processes and atmospheric feedbacks, likely play a critical role in the signal-to-noise problem.

Our ambition

The ambition of AUSPICE is to explain the signal-to-noise paradox by integrating robust statistical methods with physical experimentation and reasoning using causal networks. In order to correct for signal-to-noise errors, the causal networks will be used to calibrate climate predictions and projections, leading to self-consistent, physically grounded climate storylines and more plausible bounds for future climates.

Project goals

01

Detect

Apply robust statistical methods to identify where and when predictable signals are underestimated by climate models.

02

Explain

Use physical experiments to probe the mechanisms behind the paradox, including remote teleconnections, coupled ocean-atmosphere processes and atmospheric feedbacks.

03

Connect

Build causal networks that link these processes to the signal-to-noise errors they generate.

04

Calibrate

Use the causal networks to correct climate predictions and projections, producing physically grounded storylines and more plausible bounds for future climates.

Partners

University of Oxford · University of Reading · UK Met Office · European Centre for Medium-Range Weather Forecasts (ECMWF) · NSF National Center for Atmospheric Research (NCAR)

Meet the team →