Abstract:
The first crucial step in analysing meteorological variables is the homogenisation process that ensures historical data are of good quality, and detects and eliminates all the artificial effects (i.e. non-climatic changes) and breakpoints. Among all the meteorological variables, wind is one of the most difficult phenomenon to deal with, due to its great variability on all time scales, and in three-dimensional space, and the fact that it is affected by a considerably larger range of factors than other variables, such as changes in instrumentation, observing practices and measurement height, also the effect of terrain conditions, and site relocation. This is particularly true of gust wind speeds that are extremely sensitive to the anemometer response characteristics. Long-term historical mean and gust wind speeds are used in many applications, including estimation of extreme wind speeds for the determination of design wind loads, and the assessment of long-term wind trends and possible effects of climate change. In this study, we evaluated the homogeneity and long-term trends of daily and annual maximum gust wind speeds recorded at Wellington and Auckland international airports in New Zealand, since the 1970s. Due to the importance of meteorological stations at airports for aviation forecasts, the instruments used and data collected at these stations are often of high quality. However, using surface wind-speed observations directly without correction for instrument response characteristics, and also the assumption that airport terrain can be described as open exposure, can introduce significant errors in any subsequent studies. Here we initially determined the breakpoints caused by known factors according to the available metadata, such as changes in instrumentation and observing procedures, and measurement height, then by using a set of correction factors the data were corrected for these sources of error. After that, a statistical method, penalised maximal F test (PMFT) was employed to detect any undocumented shifts in the daily gust time series, and they were corrected by applying the Quantile-Matching (QM) algorithm. The generalised extreme value (GEV) distribution was fitted to both homogenised and non-homogenised daily maximum gusts, to investigate how artificial shifts can affect the data distribution. Lastly, the long-term trends of the homogenised daily and annual gust wind speeds were assessed and it is shown that the daily and annual gust wind speeds at Wellington airport have changed by 0.0004 (p-value<0.01) and −0.202 (p-value ≈ 0.45) m s-1 decade-1 from 1972 to 2017, respectively, though, considering shorter periods and also excluding gigantic storms resulted in different trends. These values for Auckland airport were −0.0004 (p-value<0.01) m s-1 decade-1 for daily gusts and −0.478 (p-value ≈ 0.16) m s-1 decade-1 for annual gusts.