Trends and variability of storminess in the Northeast Atlantic region, 1874–2007

Abstract

This article builds on the previous studies on storminess conditions in the northeast North Atlantic–European region. The period of surface pressure data analyzed is extended from 1881–1998 to 1874–2007. The seasonality and regional differences of storminess conditions in this region are also explored in more detail. The results show that storminess conditions in this region have undergone substantial decadal or longer time scale fluctuations, with considerable seasonal and regional differences. The most notable differences are seen between winter and summer, and between the North Sea area and other parts of the region. In particular, winter storminess shows an unprecedented maximum in the early 1990s in the North Sea area and a steady upward trend in the northeastern part of the region, while it appears to have declined in the western part of the region. In summer, storminess appears to have declined in most parts of this region. In the transition seasons, the storminess trend is characterized by increases in the northern part of the region and decreases in the southeastern part, with increases in the north being larger in spring. In particular, the results also show that the earliest storminess maximum occurred in summer (around 1880), while the latest storminess maximum occurred in winter (in the early 1990s). Looking at the annual metrics alone (as in previous studies), one would conclude that the latest storminess maximum is at about the same level as the earliest storminess maximum, without realizing that this is comparing the highest winter storminess level with the highest summer storminess level in the period of record analyzed, while winter and summer storminess conditions have undergone very different long-term variability and trends. Also, storminess conditions in the NE Atlantic region are found to be significantly correlated with the simultaneous NAO index in all seasons but autumn. The higher the NAO index, the rougher the NE Atlantic storminess conditions, especially in winter and spring.

Appendices

Appendix A: Erroneous/suspicious SLP values

For any sea level pressure (SLP) series {P _i } (i = 1,2,…,N), if |P _i-1−P _i | > 20 hPa and |P _i+1−P _i | > 20 hPa, the ith observation P _i is considered a suspicious value. The 20 hPa threshold was selected considering the typical intervals between two consecutive observations in the early period (typically 8 h on average, i.e., three observations per day) and the limits for pressure tendency check used by Environment Canada [including 14.9 hPa (5 h)⁻¹ and 16.9 hPa (6 h)⁻¹; Environment Canada 2004]. The segment of observations, {P _i −5,P _i −4,…,P _i ,…,P _i  + 4,P _i  + 5}, is checked against the corresponding segment of observations at the available nearest stations (e.g., Aberdeen is checked against Bergen and Torshavn), to determine whether the suspicious value is erroneous or not. Indeed, some of these suspects turn out to be true SLP values (e.g., those shown in Fig. 7) and are thus kept unchanged, while others are apparently erroneous. As a result, a total of 45 erroneous/suspicious values were identified; all of which are shown in Fig. 8, in comparison with the corresponding observations available at two nearest stations (only one station available in some cases). As detailed in Table 5, these values are either set to missing or corrected when possible. Most of these suspicious values are seen in the early period of the record (mostly in the late 19th century).

Fig. 7

Segments of suspicious SLP values (i.e., the lowest values in the solid curves), in comparison with the corresponding segments from two/one nearest stations if available. The x-axis is the hour of observation. These suspicious values are most likely true extremes and hence kept unchanged

Fig. 8

The same as in Fig. 7 but these suspicious SLP values (i.e., the outliers in the solid curves) are most likely erroneous and hence are set to missing or corrected when possible (see Table 5)

Table 5 Erroneous/suspicious SLP values that were either corrected or set to missing

Note that the exclusion/correction of the suspicious SLP values listed in Table 5 makes a few outliers disappear, for example, the few outliers that show very high summer storminess around 1880s in Fig. 9b are not seen in Fig. 4b. However, the general characteristics of the decadal or longer time scale storminess variability are not significantly affected by the exclusion/correction (also see Figs. 4b, 9b). This is not surprising because the exclusion/correction of random errors are not expected to bias the results systematically.

Fig. 9

The same as in Fig. 4 but these are derived from the original SLP data, leaving the suspicious/erroneous SLP values (listed in Table 5) in the series unchanged

Appendix B: Calculation of geostrophic wind speeds

For each pressure triangle, the geostrophic wind speeds (also referred to as geo-winds) are calculated from instantaneous SLP values for the same hour at the three sites that form the triangle (say site 1, 2, and 3). Let P ₁, P ₂, and P ₃ denote the three instantaneous SLP values of the same hour at the three sites (site 1, 2, and 3, respectively). Following Schmith (1995), the geostrophic wind speed is calculated as

$$ w_g = (u_g^2 + v_g^2)^{1/2} $$

(1)

with components

$$ u_g = -{\frac{1}{\rho f}}{\frac{\partial P}{\partial Y}} = -{\frac{b}{\rho f}} \quad \hbox{and} \quad v_g ={\frac{1}{\rho f}}{\frac{\partial P}{\partial X}} = {\frac{a}{\rho f}} $$

(2)

where ρ is the density of air, f the Coriolis parameter, and a and b represent respectively the zonal and meridional pressure gradients and are calculated by fitting a plane to the three SLP values P ₁, P ₂, and P ₃. Specifically, the three SLP values determine unique constants a, b, and c in the following set of equations (Schmith 1995):

$$ \left\{\begin{array}{l} P_1=aX_1 + b Y_1 +c = a R \lambda_1 \cos\phi_1 + b R \phi_1 +c \\ P_2=aX_2 + b Y_2 +c = a R \lambda_2 \cos\phi_2 + b R \phi_2 +c \\ P_3=aX_3 + b Y_3 +c = a R \lambda_3 \cos\phi_3 + b R \phi_3 +c\\ \end{array} \right. $$

(3)

where R is the radius of the earth, and λ _i and ϕ _i are the longitude and latitude (in arch degrees) of site i at the time of observation, respectively (X _i and Y _i are the Cartesian coordinates). In this study, we set ρ = 1.25kg/m³ and R = 63,78,100 m, and used the average of f over the three sites [$f=2 \Omega \sin \phi$ at the latitude ϕ, where Ω = 2π/(24 × 3600) is the rotation rate of the earth (in arch degree per second)].