Climatological Characteristics of Historical and Future High-Wind Events in Alaska

High winds cause waves, storm surge, erosion and physical damage to infrastructure and ecosystems. However, there have been few evaluations of wind climatologies and future changes, especially change in high-wind events, on a regional basis. This study uses Alaska as a regional case study of climatological wind speed and direction. Eleven first-order stations across different subregions of Alaska provide historical data (1975-2005) for the observational climatology and for the calibration of Coupled Model Inter comparison Project (CMIP5) simulations, which in turn provide projections of changes in winds through 2100. Historically, winds exceeding 25 and 35 knots are most common in the Bering Sea coastal region of Alaska, followed by northern Alaska coastal areas. Autumn and winter are the seasons of most frequent high-wind occurrences in the coastal sites, while there is no distinct seasonal peak at the interior stations where high-wind events are less frequent. An examination of the sea level pressure pattern associated with the highest-wind event at each station reveals the presence of a strong pressure gradient associated with an extratropical cyclone in most cases. Northern coastal regions of Alaska are projected to experience increased frequencies of high-wind events during the cold season, especially late autumn and early winter, when reduced sea ice cover in the late century will leave coastal regions increasingly vulnerable to flooding and erosion.

The high latitudes serve as a key example of a region in which assessments of ongoing and future changes in winds are lacking. While individual cyclones in the Arctic and their impacts on sea ice have been studied in recent years (e.g. [1] [2]), comprehensive data-based evaluations of winds, especially high-wind events, are yet to be performed for the Arctic and sub-Arctic. Moreover, evaluations of historical trends in sub-Arctic storminess and wind events have not provided compelling evidence of trends [3]. There are some indications from models of a northward shift of storm tracks over the North Atlantic Ocean [4] but the northern hemisphere observational data do not show a spatially coherent poleward shift in storm tracks [5].
In the broader context, storms are only one factor in the wind climatology of a region. Episodes of calm can be important for insect harassment of wildlife (and humans). At the other end of the spectrum, a high-wind event can occur without a storm in the immediate vicinity as, for example, a strong high pressure system can be associated with a steep pressure gradient. Topography can also play an important role in the location of strong winds. Given the various factors contributing to winds and high-wind events, this paper addresses the climatology and trends of high-latitude winds regardless of the weather systems or processes responsible for the winds. We address both mean wind and high-wind occurrences, although changes in high-wind events (Section 4) are interpretable in terms of changes in storm tracks.
In view of the heterogeneity of observational data on winds, we focus our analysis on the Pacific sector of the Arctic. This sector includes the Alaska region, which has been on the front line of climate change impacts [6], including coastal flooding and erosion as well as the introduction of wind energy into the mix of power sources. Alaska also has a network of stations for which wind measurements over the past few decades are relatively homogeneous (i.e., the 10-meter anemometer height has been standard since the 1980s at key observing The Arctic region, including Alaska, has warmed in recent decades at a rate that is twice as large as the global mean (e.g., [7] [8]). This polar amplification is due in part to the reduction of sea ice and snow cover, which provides a positive feedback to the warming that drives the loss of snow and ice [9]. Whether or not a large-scale signal of Arctic warming and sea ice loss has yet emerged from the noise of internal variability, climate models project continued Arctic warming and sea ice loss through the 21st century. Sea ice magnifies the impacts of high-wind events in the Arctic through increased wave activity, coastal flooding, and erosion. The combination of sea ice loss and high wind events increases the risks of vessel (and other infrastructure) icing in waters newly accessible for marine transport and industrial activity [10]. However, the effects of a warming climate on high-latitude storms are difficult to anticipate. On the one hand, the increased surface fluxes of heat and moisture from newly ice-free ocean areas might be expected to fuel more and stronger storms. On the other hand, polar amplification decreases the low-level meridional temperature gradients, reducing the potential for storm activity. Nevertheless, because upper-level temperatures show greater increases in the tropics than in the Polar Regions, upper-level meridional temperature gradients actually increase [11]. Hence, the net effect on baroclinicity cannot be simply related to baroclinic disturbances such as extratropical cyclones [12]. Moreover, the Arctic amplification affects the variability of the jet stream, which is directly linked to the vertically integrated meridional temperature gradient via the thermal wind equation. [13] provided a diagnostic assessment of these connections.
The expectation of increased storm activity in the Arctic is supported by several recent modeling studies. [14] showed enhanced extratropical cyclone activity over the Eurasian Arctic in model projections for the end of the century, while [4] found indications of northward shifts of the major storm tracks during autumn in the late 21 st -century model projections. However, analyses of observational data have produced mixed results on trends of high-latitude storminess.
In earlier studies, [15] found an increase of Arctic cyclone activity, while [16] reported northward shifts of storm tracks over the Northern Hemisphere (NH) over the last several decades of the 20th century. [17] detected a northward shift of cyclone activity, primarily during winter, over Canada during 1953-2002, and this meridional shift was confirmed more generally in a more recent study by the same group [18]. There cent US National Climate Assessment [3] points to a poleward shift of storm tracks over the United States during recent decades.
However, [19] found that temporal trends of cyclones in the North Pacific Ocean have generally been weak over the 60-year period ending 2008. The U.S.
Global Change Research Program [20] points to an increase of storminess on the northern Alaskan coast and to associated risks of flooding and coastal erosion along with expected sealevelrise. Since any increases of coastal flooding and erosion are also related to retreating sea ice, high-wind events in coastal areas of the Arctic can pose increasing risks regardless of whether storm activity is changing.
The present paper is a climatological assessment of high-wind events in and around Alaska, a region in which the possibility of future changes in high-wind events has major implications for planning and adaptation. Our study is motivated by the key question: Are high-wind events likely to increase or decrease in the Alaska region as the climate warms? The paper is organized as follows. Section 2 describes the data and methodology. Section 3 then summarizes the historical information on exceedances of wind speed thresholds at a network of observing stations. These results serve as benchmarks for calibrating global climate models, which are then used in Section 4 to evaluate historical and future changes of mean and extreme winds in a set of global climate models. We conclude in Section 5 with a summary of the key results and their implications, together with recommended next steps to address surface winds in the context of a changing climate.

Datasets and Methodology
This study examines the historical and future occurrences of extreme wind at 11 stations over Alaska ( Figure 1). The set of 11 stations is carefully chosen so that it covers all the geographical regions within Alaska, i.e., the Beaufort Sea coast from the list of the top performing CMIP5 models for the Alaska region [22] subject to the availability of the highest temporal resolution of model wind data i.e. 6-hourly wind output. Table 1 contains a detailed description of the six models including their full names, home institutions, and spatial resolutions.   The climate models are known to have biases. In an effort to minimize the effect of the bias a correction factor [23] was then calculated for each model's winds for each station and each month. The correction factor is calculated as follows: No. of time steps exceeding the threshold wind speed value in the station data Correction factor No. of time steps exceeding the threshold wind speed value in CMIP5historical data = This bias adjustment effectively normalizes the models' winds so that the frequencies of their high-wind events are consistent with the historical observational data. By applying the same adjustment to the models' future output, we are making the assumption that the models' biases do not change systematically in the future. Similar assumptions are made in bias-corrections of model projections of other variables, e.g., temperature and precipitation, when the Delta-method is applied in downscaling applications [24] [25] [26]. While there is no proof that this assumption is valid, we are not aware of any evidence to the contrary.     In order to illustrate the types of events that are represented in the seasonal statistics of Figure 3 and Figure 4, we present maps in Figure  The impacts of high-wind events are generally a function of wind direction.

Observational Data Synthesis
Winds with an onshore component are more conducive to elevated sea level and greater impacts on coastlines thana similarly strong winds with an offshore component. Figure 6 shows the distributions of wind directions during the with the examples in Figure 5. Overall, the distributions in Figure 6 are consistent with the seasonal distributions of the hourly events in Figure 3 and Figure   4, with summer minima at the coastal sites and more seasonally varied distributions at the interior sites.

Projections of Future Changes
In order to anticipate changes in the winds over the Alaska region in the future, we utilize the CMIP5 models simulations driven by the RCP 8.5 scenario, which is the RCP scenario that is presently tracking actual emissions most closely. As noted in Section 2, output was utilized from the top-performing global climate models for the Arctic, and a bias-adjustment based on the models' historical simulations was applied to the future output. We present here the differences be- The seasonality of the projected changes in wind speed is shown in Figure 8 as averages across the three models. The band of strongest increases, approach-

Conclusions
We Priorities for future work include the direct use of near-surface winds rather than vertically extrapolated winds from the global climate models. We anticipate that the upcoming CMIP6 simulations will make 6-hourly or hourly surface winds available so that a direct comparison can be made with the hourly station data. In addition, a more comprehensive synoptic-scale analysis of the high-wind events should be made to determine the consistency of the synoptic forcing patterns across different events at a particular location. The relative importance of cyclones and anticyclones, especially their juxtapositions, in producing high winds over the various subregions of Alaska, also deserves further investigation.
The anticipated continuation of the loss of sea ice makes Alaska a natural focus for this type of analysis.