Spatial distribution of extreme wind speeds over Sakhalin Island based on observations and high-resolution modelling data

An analysis of extreme wind speeds over Sakhalin region shows that a set of wind speed extremes obtained from observations is a mixture of two different subsets, each of them neatly described by a Weibull distribution. The empirical tail of the pdf diverges from a linearized Weibull model, indicating that another model could fit the most extreme wind speed data better. Each of these subsets is characterized by parameters k and A, regarded as coefficient and free terms of the linear Weibull model, these samples are labelled BSs and Ds. The Ds are responsible for the strongest extremes. Mesoscale modelling is used to investigate a possibility of the models to reproduce these statistical features. A detailed hydrodynamic simulation of major meteorological parameters (1985 - 2014) has been performed for the Sea of Okhotsk and Sakhalin Island with horizontal resolutions of ~13.2, ~6.6, and ~2.2 km by using a regional climate model, COSMO-CLM. This dataset is utilized for an investigation of the statistical structure of extreme wind speeds. First, it is shown that a model with a detailed spatial resolution is able to reproduce the statistical structure in general, and a special mechanism is responsible for the generation of the largest of wind extremes. However, a model with a resolution of ~13,2 km could not reproduce some essential parts of the wind speed maximum’ statistical properties, underestimating the parameters k and A of Ds Weibull distribution significantly. This gap could be covered by using a higher resolution, as well as by areal estimation techniques and many others.