The frequency of storm events in the Barents Sea over the last 35 years
Study of wave conditions in the Arctic seas of Russia is the priority task due to the oil and gas shelf fields development, maritime traffic over the North Sea Route and the related infrastructure development.
The mathematical modeling is the main instrument for wave studing because it is no sufficient observation data. Wave models can provide the data about wave parameters for the retrospective investigations and forecast information. The wind fields from the global reanalysis or forecast models are often use for this purposes [5, 7, 11, 12].
At the moment, the forecast of the wave in the Barents Sea is realized in the Hydrometeorological Centre of Russia by means of WW3 model . The model has the space increment of 9 km and works on the GFS incoming data. The model quality check was carried out on the basis of the satellite data, and the results of the tests were obtained as satisfying.
Among the native works dedicated to the waves in the Barents Sea modeling, the work  can be singled out, where SWAN model was used for the shipping register drawing up, and the assessments showed good coincidence with the observation details.
Computations and forecasts of the wind waving are also carried out with the usage of AANII (PD2-AARI) spectral and parametric model. The model was repeatedly verified by the data of the instrumental observations in the different waters and associated with such world famous models as WAM and WAVEWATCH. The results of the computations and forecasts accuracy statistical estimations are approximately similar for all models (mean root square error is around 0.5 m, the correlation ratio between the observed and calculated indexed is approximately equal to 0.9 [5, 11].
But it is no science investigations about temporal variability of storm events in the Barents Sea.
Data and methods
The study of wave fields in the Barents sea was made by using WAVEWATCH III wave model . This model is very popular instrument for wave studying and it provide a good results . We get wind and ice cover fields data from reanalysis NCEP/CFSR (resolution 0.3°, time step 1 hour). This reanalysis have a good spatial resolution and in work  shown, that wind fields from this reanalysis have a good quality. The data about ice fields is very important for wave modelling in Arctic seas, because ice can seriously limit the wave growing and propagation (fig.1). We use special unstructured computing mesh which include the North Atlantic region from equator to pole with spatial resolution around 1° and the Barents with resolution around 0.2° (Fig.2). The main idea of using so big area in calculations was to minimize the open boundaries effects.
We made wave reanalisys from 1979-2015. The storm events where significant wave height higher then selected criterion (4, 5, 6, 7 m) was calculated. The distribution of storm events frequency is presented for each year.
The implementation of spectral wave model SWAN for the Barents Sea, including the Northern part of the Atlantic Ocean was presented. Computations were performed by using special unstructured mesh, which has spatial resolution in the Atlantic Ocean is 1 °, in the Barents – 0.5 °. We already used this mesh in the previous studies .
Fig.1. Example of the ice concentration in the Barents sea
Fig.2. The nodes of the computational grid for calculation wind wave in the Barents Sea
Our wave data base for the Atlantic ocean and the Barents sea includes the wave parameters (significant wave height, period, wave lengh, wave energy) with time step 3 hours. The example of wave field shows the spatial distributuon of significant wave height in the Atlantic ocean and the Barents sea (fig.3).
Fig.3. The field of significant wave height in the Atlantic ocean and the Barents sea
Than we calculated the storm events where significant wave height higher then 5 m (or enother selected height). It is a serius problem to separate the different events which suffices criteria only by machine algoritm and we used visual correction too. As a result we get the number of storm events for each year till 1979 to 2015. This allows to analyze the frequency of storms in different years.
Preliminary analysis shows that it is no strong trends in the frequency of storm events 1979-2015 but it was several periods when the storm activity was growing.