Estimating the resilience of non-stationary systems.

A regression-based Langevin framework estimates resilience in non-stationary systems, handling seasonal forcing and irregular sampling while remaining a drop-in replacement for autocorrelation-based methods.

• Why it matters: Critical slowing down is widely used to infer stability, but non-stationarity in Earth system components like global vegetation makes resilience estimation difficult. This work targets that gap by enabling resilience inference without heavy preprocessing.
• What they did: The authors formulate a regression approach to the Langevin equation to estimate stability in synthetic and real-world systems, including robustness to data gaps and no requirement for regular time sampling. They also incorporate time-varying data uncertainties, compute uncertainty bounds, and extend the approach to spatial systems.
• The result: The method yields resilience estimates with uncertainty bounds and works directly with existing autocorrelation-based workflows, enabling broader application across Earth system components under non-stationary conditions.

AI summary of the abstract · may contain errors · not medical advice · read the paper.