Detrended partial-cross-correlation analysis: a new method for analyzing correlations in complex system

Area of Science:

• Complex Systems Analysis
• Time Series Analysis
• Climate Science

Background:

• Non-stationary signals often exhibit complex interdependencies.
• Existing methods like DCCA may not fully isolate intrinsic relationships due to confounding factors.

Purpose of the Study:

• Introduce detrended partial-cross-correlation analysis (DPCCA) to quantify relationships between two non-stationary signals.
• Remove the influence of other signals to reveal intrinsic correlations.
• Apply DPCCA to climate data to demonstrate its utility in natural complex systems.

Main Methods:

• Developed detrended partial-cross-correlation analysis (DPCCA) by integrating partial-correlation technique into DCCA.
• Conducted numerical tests to validate DPCCA's ability to handle non-stationary signals and isolate intrinsic relationships.
• Applied DPCCA to analyze the influence of Pacific Decadal Oscillation (PDO) and Nino3 Sea Surface Temperature Anomaly (Nino3-SSTA) on Summer Rainfall over the middle-lower reaches of the Yangtze River (SRYR).

Main Results:

• Numerical tests confirmed DPCCA's effectiveness in handling non-stationary signals and identifying intrinsic correlations.
• DPCCA identified significant correlations between SRYR and Nino3-SSTA on 6-8 year time scales (1951-2012).
• DPCCA revealed significant correlations between SRYR and PDO on 35-year time scales.

Conclusions:

• DPCCA is a valuable method for analyzing relationships within complex systems, particularly for non-stationary time series.
• The findings provide new evidence for the influence of PDO and Nino3-SSTA on Yangtze River summer rainfall.
• The method's ability to isolate physically explainable correlations enhances its applicability in environmental and climate research.