A unified nonlinear stochastic time series analysis for climate science

Area of Science:

• Climatology
• Time Series Analysis
• Geophysics

Background:

• Earth's orbit and axial tilt create a strong seasonal cycle in climate data.
• Climate variability is often analyzed as fluctuations in this seasonal cycle due to higher-frequency processes.
• This can be viewed as a competition between orbitally-driven monthly stability and weather-induced noise.

Purpose of the Study:

• Introduce a novel time-series method to quantify seasonal stability and noise from monthly climate data.
• Identify climate fingerprints and understand the underlying mechanisms of seasonal predictability.

Main Methods:

• Developed a new time-series analysis technique.
• Applied the method to monthly-averaged climatological data.
• Analyzed the spatio-temporal distribution of monthly stability and noise magnitude.

Main Results:

• The method successfully quantifies contributions of seasonal stability and noise.
• Spatio-temporal patterns of stability and noise reveal fingerprints of Arctic sea ice, ENSO, Atlantic Niño, and Indian Dipole Mode.
• A destabilizing process operates seasonally across these phenomena, interacting with noise accumulation (memory effect).

Conclusions:

• The memory effect explains phase locking to the seasonal cycle and statistical seasonal predictability.
• This approach provides new insights into the dynamics of major climate phenomena and their seasonal behavior.