*3.2. Wind Conditions*

We found large year-to-year variability in the timing of the annual maximum wind speed in the period 1979–2014. Among all 40 stations, the most common month for annual maximum wind speed was December (Figure 3a). However, also months from October to May are rather common for annual maximum wind speeds. Direction wise the annual maximum wind speeds from N, NE, and E were most commonly observed during spring and early summer (April–June), whereas for rest of the directions it was usually observed from October to March (not shown).

It was rather common that the annual maximum wind speed was observed multiple times during a year, partly driven by wind observations having no digits in the first part of the study period. About 25% of all the years among 40 stations had annual maximum wind speeds observed on more than one month of that year. 49% for PP, 59% for CSS, and 62% for SP of these years were ones with similar maximums during frozen and unfrozen seasons (not shown).

**Figure 3.** (**a**) Distribution of annual maximum wind speed observation months over 40 weather stations and period 1979–2014. (**b**) Proportion of annual maximum wind speed (1979–2014) observed either during frozen (black) or unfrozen (grey) soil frost season at weather stations located in southern, central, and northern Finland. CSS (spruce on clay/silt), SP (pine on sand), PP (pine on peat).

Figure 3b presents how observed annual maximum wind speed was split up between frozen and unfrozen seasons on different forest and soil type combinations in southern (14 stations), central (12 stations), and northern (14 stations) Finland. The annual maximum wind speed was observed rather evenly in both seasons in the case of CSS and SP, occurring slightly more often during unfrozen season in southern Finland and frozen season in northern Finland, respectively. For PP the difference was more pronounced, especially in southern and central Finland where annual maximum wind speed was clearly more often observed during unfrozen season.

We also further compared the underlying distributions of seasonal maximum wind speeds, from where return level estimates for 40 stations were calculated, using two-sample Kolmogorov–Smirnov tests to determine if there was statistically significant (*p* < 0.05) difference between frozen and unfrozen soil seasons. For CSS, difference was statistically significant in southern and northern, and non-significant in central Finland. For SP, significant in central and northern, and non-significant in southern Finland. And for PP, significant in southern and central, and non-significant in northern Finland, respectively.
