During the period 1993–2009 the correlation coefficient R of the winter (JFM) NOy deposition with the length of the ice season varied between R = —0.52 over B2 and R = –0.19 over B4. The minimum probability (P-value) was 0.028 and the explanation factor R2 = 26.7% over B2. The anti-correlation Lumacaftor solubility dmso is stronger when December is included in the winter period. When winter is defined either as DJF or DJFM, the correlation is extremely significant (P < 0.0006, R2 > 54%) over B2, the Gulf of Finland, and significant (P < 0.01, R2 > 35%) over B3. For B2 the variation

in the length of the ice season is important, because over this sea area the share of the annual airborne load due to winter and autumn deposition is 55–70%. However, ice conditions depend on the frequency of northerly or easterly continental airstreams, and all other MBL parameters vary with the cold air as well. Figure 15 presents the seasonally averaged correlation of the monthly NAO index with the oxidized

nitrogen deposition to the Baltic Sea subbasins in the years 1993–2009. The correlation was extremely significant over B2, the Gulf of Finland, in winter (JFM), and significant (P < 0.01) over B3 and B4 in winter and over B1 in autumn (OND). The reasons for and the Kinase Inhibitor Library price origin of the episodically-received external load to the northern Baltic Sea sub-basins B1–B3 cannot be explained fully by instantaneous local meteorological factors (wind speed or direction, turbulence, state of other weather elements or the passing of a cyclone), because nitrogen compounds are transported

long distances to the areas of deposition and they remain in the air for several days before the being deposited. Each episode is the result of a chain of events connected to cyclonal and frontal activity. Precipitation and weather extremes are not concurrent with the cyclone centre crossing a given sea area, but tend to occur with a time lead, as the wind field connected with cyclones and fronts is complicated. However, deposition does seem to depend on the frequency of extreme weather events and cyclone activity, which in turn depend on the variation of large-scale weather patterns, such as the NAO, prevailing over the Baltic Sea. The results of the analysis of wind velocity and pressure minimum extremes presented in the previous section can be compared with estimates of storm frequency along the Swedish coast in the southern and northern BS (Eek 2000) or along the western BS5 (Olsson 2002); these show that at the Vingas station there was a distinct minimum in storm frequency between around 1935 and 1968, a maximum in 1920–1930 and in the 1980s. A similar variation, with some differences in the details of years and periods, can be seen from the data of the other stations studied by Olsson and Eek. The total number of severe storms was highest in 1919–1929 and 1940–1949 (Eek 2000).