Analysis and characterisation of extreme wind gust hazards in New South Wales, Australia

Abstract

Extreme wind gusts cause major socioeconomic damage, and the rarity and localised nature of those events make their analysis challenging by either modelling or empirical approaches. A 23-year long data record from 29 automatic weather stations located in New South Wales (eastern Australia) is used to study the distribution, frequency and average recurrence intervals (ARIs) of extreme gusts via a peaks-over-threshold approach. We distinguish between gust events generated by synoptic phenomena (e.g. cyclones and frontal systems), hereafter called “synoptic events”, and convective phenomena (i.e. thunderstorms), hereafter called “convective events”, using the wind time series. For synoptic events the frequency of gusts $$>25$$ > 25 m/s decreases systematically inland from the coast, in contrast to convective gusts which are more uniformly distributed geographically and occur more often than synoptic gusts at nearly all inland locations. At inland locations the most extreme wind gusts are likewise dominated by convective events, whereas at coastal stations both gust types have similar intensities at low ARIs but convective events again dominate at the highest ARIs. Extreme gust directions were found to be predominantly westerly at inland locations and southerly at coastal ones, with more variable direction for convective than synoptic events. This study confirms the dominant role of thunderstorms in producing the most extreme gusts in the region, and shows that wind risk varies strongly with distance from the coast.

Biases in Estimating Long‐Term Recurrence Intervals of Extreme Events Due To Regionalized Sampling

“…Previous studies have performed goodness‐of‐fit tests and have demonstrated that the GPD and GEV models are suitable for modeling wind and rainfall extremes in the region of interest (Green et al., 2012; Holmes, 2002, 2019; Johnson & Green, 2018; Wang et al., 2013). The reader is also referred to (El Rafei, Sherwood, Evans, & Dowdy, 2023) in which we demonstrate using quantile plots and return level plots a good fit between the GPD model and the station data from the Sydney region.…”

“…Therefore, the threshold value is specific to each station and not pre‐fixed for the entire region. See (El Rafei, Sherwood, Evans, & Dowdy, 2023) for more details on the algorithm and event classification technique.…”

“…The statistical analysis of extreme-event frequencies and intensities is important to many risk management problems. For example, estimating the design wind speed (the estimated maximum speed of wind that a structure should be designed to withstand) requires the statistical analysis of historical wind data to estimate the strongest wind that might occur over a long time interval (AS/ NZS1170.2:2021NZS1170.2: , 2021El Rafei, Sherwood, Evans, & Dowdy, 2023;Holmes, 2002). The design of offshore and coastal marine structures is governed by statistical estimates of extreme waves (Gulev & Grigorieva, 2004;Meucci et al, 2020).…”