Recent recovery of surface wind speed after decadal decrease: a focus on South Korea

Abstract

We investigate the multi-decadal variability of observed surface wind speed around South Korea. It is found that surface wind speed exhibits decreasing trend from mid-1950s until 2003, which is similar with the trends reported for other parts of the world. However, the decreasing trend ceases and becomes unclear since then. It is revealed that decreasing wind speed until 2003 is strongly associated with the decreasing trend of the spatial variance in both atmospheric pressure and air temperature across the East Asia for the same period. On the contrary, break of decreasing trend in surface wind speed since 2003 is associated with increasing spatial variance in surface temperature over the East Asia. Ground observation shows that surface wind speed and air temperature exhibit highly negative correlations for both summer and winter prior to 2003. However, since 2003, the correlations differ between seasons. We suggest that mechanisms behind the recent wind speed trend are different between summer and winter. This is on the basis of an interesting finding that air temperature has decreased while surface temperature has increased during winter months since 2003. We hypothesize that such contrasting temperature trends indicate more frequent movement of external cold air mass into the region since 2003. We also hypothesize that increasing summer wind speed is driven by intrusion of warm air mass into the region which is witnessed via increasing spatial variance in surface temperature across East Asia and the fact that both air and surface temperature rise together.

Appendices

Appendix 1: Transition point

To determine the transition point in the time series, yearly difference time series are first obtained from annual mean and annual maximum series, respectively. Then, transition points are calculated from the yearly difference time series. We implemented three methods of two-phase regression technique (Elsner et al. 2000), tabular cumulative summation test (Montgomery 2009), and commercial program based on combination of cumulative summation charts and bootstrapping (Taylor 2000). These methods yield different transition points (Table 2). Here, we follow the result from Elsner et al. (2000)’s method in that this method gives the same year (2003) as the transition point for both annual mean and annual maximum series.

Table 2 Transition points in difference time-series of wind speed

When we implement the same methods with the original annual mean and annual maximum series and transition points are less clearly identified. Using yearly difference time series enhances clarity in determining transition points.

Appendix 2: Statistical significance of trends

All trends reported in this paper, i.e., not only those of wind speed but also of geopotential and temperature, are tested for statistical significance. The significance of linear regression trends is tested within 95 and 90 % confidence levels by student’s t test using Spearman’s rank correlation (1904). Most trends prior to 2003 are found statistically significant at 90 % confidence level (Table 3). Not all trends are found significant and we distinguish statistically significant trend and the rest by using solid or dashed line in all linear regression lines.

Table 3 Statistical significance of trends reported in this study