Spatial Assessment of the Stationarity of Extreme Winds in a Coastal Region Around Jiangsu

Extreme wind analysis commonly assumes stationarity, although climate change may alter the statistical properties of wind extremes. For a coastal region with substantial offshore wind energy development, assessing the spatial applicability of this assumption is important for robust extreme wind characterization. This study examines the spatial variability of extreme wind stationarity using annual maximum 100-m wind speeds derived from hourly ERA5 reanalysis data for 1980–2025. Four generalized extreme value (GEV) models with stationary and nonstationary formulations were compared using the corrected Akaike information criterion (AICc), with the Bayesian information criterion (BIC) and parametric bootstrap likelihood ratio tests (LRTs) as complementary assessments. A parametric bootstrap approach was used to quantify uncertainty in the estimated return levels. Results show that stationary GEV models were selected at 68.8% of locations, with nonstationary models selected at the remaining 31.2%, primarily in the northern offshore area. Among all locations, 12.6% were classified as criterion-consistent nonstationary, based on agreement among AICc, BIC, and nominally significant bootstrap LRT results, whereas 18.6% were classified as AICc-only nonstationary. However, none of the bootstrap LRT results remained significant after Benjamini–Yekutieli correction for multiple testing. A leave-one-year-out analysis further showed a mean stationary/nonstationary classification stability of 95.7% across locations, indicating limited sensitivity of the AICc-based classifications to the exclusion of individual years. The agreement between AICc-based classifications and those based on the BIC and bootstrap LRTs was 83.7% and 81.8%, respectively. Relative differences between AICc-selected and stationary return levels increased from 0.20% at 10 years to 1.41% at 100 years, while the median relative half-width of the 95% bootstrap confidence interval was 15.2% at 50 years and 19.6% at 100 years. These findings support the use of stationary GEV models across much of the study area while underscoring the need to account for spatial differences in the statistical support for nonstationarity in extreme wind risk assessment.

Authors

Institutions

Publication Details

Journal
Atmosphere
Published
2026-11-13
DOI
https://doi.org/10.3390/atmos17111125
Primary Topic
Wind Energy Research and Development
Type
article
Field-Weighted Citation Impact
0.00

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Spatial Assessment of the Stationarity of Extreme Winds in a Coastal Region Around Jiangsu

Yi Yao, Dayang Li, Lu Liu, Yan Zhou et al.
1 links
Atmosphere
Wind Energy Research and Development
article

Spatial Assessment of the Stationarity of Extreme Winds in a Coastal Region Around Jiangsu

Yi Yao, Dayang Li, Lu Liu, Yan Zhou, Yueru Zhao, Zhaonan Wang, Jiayi Zhang
article en

Abstract

Extreme wind analysis commonly assumes stationarity, although climate change may alter the statistical properties of wind extremes. For a coastal region with substantial offshore wind energy development, assessing the spatial applicability of this assumption is important for robust extreme wind characterization. This study examines the spatial variability of extreme wind stationarity using annual maximum 100-m wind speeds derived from hourly ERA5 reanalysis data for 1980–2025. Four generalized extreme value (GEV) models with stationary and nonstationary formulations were compared using the corrected Akaike information criterion (AICc), with the Bayesian information criterion (BIC) and parametric bootstrap likelihood ratio tests (LRTs) as complementary assessments. A parametric bootstrap approach was used to quantify uncertainty in the estimated return levels. Results show that stationary GEV models were selected at 68.8% of locations, with nonstationary models selected at the remaining 31.2%, primarily in the northern offshore area. Among all locations, 12.6% were classified as criterion-consistent nonstationary, based on agreement among AICc, BIC, and nominally significant bootstrap LRT results, whereas 18.6% were classified as AICc-only nonstationary. However, none of the bootstrap LRT results remained significant after Benjamini–Yekutieli correction for multiple testing. A leave-one-year-out analysis further showed a mean stationary/nonstationary classification stability of 95.7% across locations, indicating limited sensitivity of the AICc-based classifications to the exclusion of individual years. The agreement between AICc-based classifications and those based on the BIC and bootstrap LRTs was 83.7% and 81.8%, respectively. Relative differences between AICc-selected and stationary return levels increased from 0.20% at 10 years to 1.41% at 100 years, while the median relative half-width of the 95% bootstrap confidence interval was 15.2% at 50 years and 19.6% at 100 years. These findings support the use of stationary GEV models across much of the study area while underscoring the need to account for spatial differences in the statistical support for nonstationarity in extreme wind risk assessment.

AtmosphereVol. 17(11)
Nanjing Hydraulic Research Institute (CN), Yancheng Teachers University (CN), Yancheng Institute of Technology (CN), Nanjing University (CN)
National Natural Science Foundation of China, National Key Research and Development Program of China
Climate action
Wind Energy Research and Development
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon