Scholarlink

<p>Detecting non-stationarity in hydroclimatic extremes requires observational records long enough to distinguish structural changes from internally generated multi-decadal variability, particularly in regions characterized by strong low-frequency fluctuations. Here we examine how record length influences the robustness of change detection using a moving-window framework applied to the Genoa, Italy precipitation record (1833–2022; a 189-year temporal span including 190 calendar years of observations). For each window length L, overlapping subseries are analyzed using complementary tools: non-parametric change-point detection (Pettitt test), distributional comparison (moving-window Kolmogorov–Smirnov statistics), and sliding-window Generalized Extreme Value (GEV) modelling. Reliability is assessed through detection rate and stability of the estimated change-point timing, allowing identification of a Maximum Reliability Zone (MRZ) where results become comparatively stable. The analyses indicate that short-to-moderate record lengths can yield variable detection outcomes, whereas longer records provide more consistent diagnostics. For the Genoa case study, stabilization occurs for periods greater than 150 years. Although demonstrated for a single site, the proposed workflow is transferable to other variables and stations, providing practical guidance for assessing the robustness of detected climate changes.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Record-length requirements for detecting non-stationarity in Mediterranean precipitation extremes: evidence for periods greater than 150 years from the Genoa, Italy record (1833–2022)

  • 期刊论文

  • 发表日期：

    2026年7月31日

查看全文
  • Giorgio Russo,
  • Mahmoud M. Elwaheidi

摘要

Detecting non-stationarity in hydroclimatic extremes requires observational records long enough to distinguish structural changes from internally generated multi-decadal variability, particularly in regions characterized by strong low-frequency fluctuations. Here we examine how record length influences the robustness of change detection using a moving-window framework applied to the Genoa, Italy precipitation record (1833–2022; a 189-year temporal span including 190 calendar years of observations). For each window length L, overlapping subseries are analyzed using complementary tools: non-parametric change-point detection (Pettitt test), distributional comparison (moving-window Kolmogorov–Smirnov statistics), and sliding-window Generalized Extreme Value (GEV) modelling. Reliability is assessed through detection rate and stability of the estimated change-point timing, allowing identification of a Maximum Reliability Zone (MRZ) where results become comparatively stable. The analyses indicate that short-to-moderate record lengths can yield variable detection outcomes, whereas longer records provide more consistent diagnostics. For the Genoa case study, stabilization occurs for periods greater than 150 years. Although demonstrated for a single site, the proposed workflow is transferable to other variables and stations, providing practical guidance for assessing the robustness of detected climate changes.