Skip to content

Asymptotic Distribution of Mann-Kendall

Home / Open Questions / Asymptotic Distribution of Mann-Kendall

Background: The Mann-Kendall test statistic’s asymptotic distribution is well-defined under independence but behaves unpredictably in the presence of the complex autocorrelation structures commonly generated by early warning signal estimation methods. Developing a robust theoretical characterization is necessary for accurate critical transition forecasting.

Question / Future Work: Deriving a rigorous characterization of the asymptotic distribution of Mann-Kendall’s statistic specifically for autocorrelated time series produced by common EWS indicators, as an alternative to expensive resampling-based inference.

Why It Matters: Valid theoretical distributions are required to eliminate inflated false-positive rates that currently plague critical transition forecasting when using standard trend tests.

Evidence: We finally note that a better characterization of the asymptotic distribution of Mann-Kendall’s tau could still serve as the basis for a dedicated statistical trend test for EWS detection.

Metadata & Links