This section outlines the key findings from this study, first covering the seasonal variation in the jet latitude and speed, highlighting the variations over land masses compared to the ocean basins, before looking at the interannual variability and decadal trends. Where appropriate the results are shown for two periods 1871–2011 and 1940–2011. The latter period is used for comparison, as the spread across the ensemble members is significantly lower after 1940 (Fig. 2) as discussed.
Seasonal jet latitude and speed climatology
The jet latitude seasonal cycle (Fig. 3) shows a poleward shift in summer, but there are regional variations in the amplitude and lag with respect to insolation. For the full analysis period from 1871 to 2011 (Fig. 3a, c, e, g), the jet latitude amplitude is greatest over land; for Eurasia and North America the range is 20° from 34–54° N and 39–59° N, respectively. Over the North Atlantic the seasonal range is lower at 10° from 46–56° N and over the North Pacific the range is about 15° with a narrow peak in July and August. Only considering the 1940–2011 period (Fig. 3b, d, f, h) we find a reduction in the peak latitude in the summer months by around 3° over land and 2° over the ocean basins. Again, the jet latitude amplitude remains greatest over land; with a maximum for Eurasia of 18° whilst the North Atlantic range is only 7°. There is little difference for the interannual variability between the two periods whereas the uncertainty related to the ensemble spread is much reduced for the 1940–2011 period.
The seasonal cycle curves also have different shapes. In all regions there is a response of the jet stream to insolation. Over North America the response of the jet stream broadly follows a sinusoidal curve which lags insolation by about 1–2 months. The peak is broader over Eurasia, particularly for the period 1940–2011. Between February and June, the jet stream moves to its northernmost latitude at around 50° N, it then plateaus at this latitude until October, after which there is a steep decline. The North Atlantic shape is different again and a lag to insolation is evident. There is a broad flat line from January to May with only a 1° increase in latitude. From May onwards there is a 6° increase to September, before a steady fall from September to December. For the 1940–2011 period the seasonal cycle is barely significant and the interannual variability found for any given month largely overlaps with the interannual variability for any of the other months. Over the North Pacific the jet latitude slowly increases from February to June and a narrow peak is reached in July/August which is followed by a steady southward movement from October onwards.
The seasonal cycle of jet speed (Fig. 4) shows a similar pattern across land and the North Atlantic and displays a near-sinusoidal curve with maximum wind speeds seen in January around 59 ms−1 and minimum in July around 40 ms−1. The cycle over the North Pacific resembles the inverse cycle seen for jet latitude, with a steep decrease in speed from May to July and steep increase from August to October. Speeds are also highest over the North Pacific, reaching a maximum in January at 66 ms−1 and minimum in July at 38 ms−1. The average jet speed for each region and season is also over 35 ms−1 for each longitude (not shown). The main change observed between the 1871–2011 and 1940–2011 is that in the latter period the summer speeds have decreased by around 3 ms−1 to 39 ms−1, whilst winter speeds have increased across North America and the North Atlantic.
Looking at the seasonal variations across the northern hemisphere (Fig. 5) indicates that the zonal variability in jet stream latitude is greatest in the winter months (DJF), resulting in a wave like pattern, with two latitude maxima located over the eastern Pacific and Atlantic. The jet stream follows a well-defined path in winter which is tightly confined, particularly on the western boundary of the North Pacific (33° N) near the Kuroshio current and North Atlantic (41° N) aligning with the Gulf Stream/North Atlantic Current. The jet stream troughs are also located in these areas in winter; the main areas for cyclogenesis. In spring (MAM) and autumn (SON) the jet stream troughs are further north, but remain over the western boundaries of the North Pacific (37° N, 41° N) and North Atlantic (43° N, 47° N) respectively. In summer (JJA) the jet stream follows a more zonal path around 55° N with the main ridge west of Hudson Bay (60° N, 100° W).
Of particular note is the narrow range in the mean jet latitude position in winter along the western boundary of the North Atlantic and North Pacific. The jet is located 1° to the north of the maximum gradient of the 2 m air temperature and is aligned with the temperature contours. The relationship between the mean jet position along the western boundary, and the maximum gradient of 2 m air temperature is retained in Spring and Autumn although not as tightly defined with the mean jet position located 3° and 4° north of the maximum gradient, respectively. The 2 m air gradient and land-sea temperature contrast are not as strong in these seasons. No relationship is evident in summer. It would perhaps be expected that the relationship of the maximum gradient of 2 m air temperature and mean jet latitude would also been seen over land. Although there is some relation, it is not as clear.
Multi-decadal trends in jet latitude and speed
This section details decadal trends for the Northern Hemisphere and then on a regional basis. Significant trends are at the 95% confidence level or higher. Jet latitude shows differing trends in each season (Fig. 6). For the northern hemisphere in winter (DJF) there is a significant 1.2° (0.1°/decade) long-term increase from a mean of 37.5 to 38.7° N (Fig. 6). In spring (MAM) there is no significant change in the mean of 44.3° N. In summer (JJA) only the trend after 1940 is considered, due to the range in the standard deviation in the earlier period. There has been a 0.3° N increase in jet latitude. In Autumn (SON) there is a significant 1° (0.1°/decade) decrease in the jet latitude from 49.4° N to 48.4° N over 141 years.
When trends are analysed on a regional basis, a different picture unfolds (Fig. 7). For winter (DJF) significant increasing trends in jet latitude are seen over the North Atlantic of 3.0° (0.2°/decade) from 44° N to 47° N and over Eurasia with an increase of 1.7°(0.1°/decade) from 33.1° N to 34.8° N. Across the North Pacific and North America there is no change in the mean position over the 141 year period. In Spring (MAM) only the North Atlantic shows a significant increasing trend in mean jet latitude of 1.8° (0.1°/decade) from 45.6 to 47.4° N (Supplementary Fig. 1). Over Eurasia there is no change in the mean of 43.6° N. The Pacific and North America show a decreasing trend until the 1940s and increasing thereafter, but the increase is not statistically significant. For the summer (JJA) after 1940, only Eurasia has a significant increase of 1.6° (0.1°/decade) from 51 to 52.6° N (Supplementary Fig. 1). The North Atlantic has a modest 0.4° increase. The North Pacific and North America show a 0.3° and 1.2° decrease, respectively, but neither is statistically significant. In Autumn (SON) the North Atlantic is not in line with the hemisphere trend with an increase of 0.8° N, although not statistically significant (Supplementary Fig. 1). Eurasia has no change, whilst North America and the North Pacific display a significant decreasing trend of 0.15°/decade, however the trend becomes modest after 1940 and is not significant thereafter.
Overall, only the North Atlantic shows an increasing trend in jet latitude across all seasons. Eurasia shows significant increases but only in winter and summer, whilst the North Pacific and North America have either no change or decreases in the seasons.
Jet speed shows a significant increase of 2.0 ms−1 for winter in the Northern Hemisphere (0.1 ms−1/decade) (Fig. 8). In the other seasons, there is a decrease in jet speed from 1871 to 1940 followed by modest (not statistically significant) increase thereafter of between 0.1 and 0.3 ms−1. Again, there are regional differences (Fig. 9 and Supplementary Fig. 2). In winter, we observe significant jet speed increases: North America 4.73 ms−1 (0.3 ms−1/decade), the North Atlantic 4.52 ms−1 (0.3 ms−1/decade) and over Eurasia 1.8 ms−1 (0.1 ms−1/decade). No trend is seen for the North Pacific. In spring, regional trends are in line with winter, with the largest significant increase seen over the North Atlantic (2.4 ms−1, 0.2 ms−1/decade), and speed increases of 1 ms−1 (0.1 ms−1/decade) over North America and Eurasia. In summer, only the period after 1940 is considered. A significant increasing trend is seen over North America of 1.6 ms−1 (0.2 ms−1/decade). For autumn; the North Atlantic has a significant 3 ms−1 (0.2 ms−1/decade) increase, North America a 1.2 ms−1 (0.2 ms−1/decade) increase for the period since 1940, but no change over the full period. Across Eurasia the increasing trend seen is not significant after 1940. Over the Pacific, the decadal jet speed trends are different. In winter, as with jet latitude, there is no change in the mean jet speed of 67.7 ms−1 but there is significant interannual variability from 62 to 72 ms−1. In the other seasons, there are no significant trends since the 1940s.
The extent of the relationship between jet latitude and speed was also evaluated for each region and season (Table 1). Over the North Pacific there is a significant negative correlation in all seasons, which is strongest in winter, explaining 42% of the variance. The highest negative correlation is located in the eastern part of the North Pacific at 170° W, explaining 64% of the variance since 1940 (not shown). Over North America, a significant negative correlation exists in spring and autumn. Over the North Atlantic there is a positive correlation in winter and spring, but negative in summer and autumn. The strongest significant correlation was in winter, for the period 1871–2011, but only explains 9% of the variance. Woollings et al. (2014) also show a low correlation between jet latitude and speed over the North Atlantic. On closer analysis of the North Atlantic in winter, however, there is a negative correlation west of 40° W (maximum value at 60° W r = − 0.3) and positive correlation 40° W to 0° W (maximum value at 10° W, r = 0.5), which is masking the true picture for the region. We note that the eastern Atlantic is the only region where significant positive correlations are found between jet stream latitude and speed.
Table 1 Jet Latitude and Jet Speed Correlation (r) for the periods 1940–2011 (1871–2011) Statistically significant correlations at the 95% confidence level or higher are shown in bold Interannual to decadal variability of jet latitude and speed
The interannual to decadal variability seen in Figs. 3, 4, 6 and 7 is analysed in more detail using wavelet analysis and compared to known atmospheric and ocean indices; Atlantic Multidecadal Oscillation (AMO), North Atlantic Oscillation (NAO), Pacific Decadal Oscillation (PDO) to investigate if any co-variability exists on decadal timescales. The Atlantic Multidecadal Oscillation (AMO) Index, is defined as the area average SST anomaly over the North Atlantic (0° N–65° N, 80° W–0° E). It is usually detrended to show only interannual variability and has a periodicity of around 65–70 years (Schlesinger and Ramankutty 1994). Warm phases have occurred from 1930–1965 and since 1995 with cool phases of the AMO between 1900–1930 and 1965–1995. The linearly detrended AMO index was correlated with the North Atlantic jet latitude and jet speed. For jet latitude and jet speed significant positive correlations were observed in winter (summer) with r = 0.24 (r = 0.32) for jet latitude and r = 0.28 (r = 0.3) for jet speed. The low correlation is similar to the winter findings by Woollings et al. (2014).
The NAO is an atmospheric index based on the surface sea level pressure (SLP) difference between the subpolar (Icelandic) low and the subtropical (Azores) high. During a positive (negative) phase of the NAO there is a greater (smaller) difference between the SLP between the Azores high and the Icelandic low (Hurrell 1995). The PDO index (Mantua and Hare 2002) is the leading empirical orthogonal function (EOF) of monthly SSTA over the North Pacific (poleward of 20° N) after the global SST has been removed, with 2 periodicities; 15–25 years and 50–70 years. We use wavelet cross-coherence to identify the links between jet latitude/speed and the PDO/NAO (Torrence and Compo 1998). Correlations between NAO/PDO and jet stream latitude/speed over the Atlantic and Pacific regions are shown in Table 2.
Table 2 Jet Latitude and Jet Speed Correlation (r) with the NAO/PDO 1940–2011 (1871–2011) Statistically significant correlations at the 95% confidence level or higher are shown in bold When looking at the PDO/NAO we find that the links with the jet stream latitude and speed vary greatly between regions and seasons. Over the North Pacific, jet latitude and PDO are in antiphase (Table 2; Fig. 10) and in winter show continuous significant coherence for periods between 12 and 30 years. Jet Speed and the PDO are in phase (Table 2; Fig. 11) and in winter show continuous significant coherence between 12 and 26 years.
Over the North Atlantic in winter, jet latitude and NAO are in phase and show significant coherence at 20-year timescales for the period 1930–1960 (Fig. 10), and significant coherence at 8–10 year timescales for the period since 1980. Jet speed and the NAO are in phase and in spring show significant coherence at 16–24 year timescales for the period since 1940 (Fig. 11). Over Eurasia, in the transition seasons of spring and autumn, jet latitude and the NAO show significant coherence since the 1930s on timescales of 16–28 years (Fig. 10), whilst jet latitude and the PDO are out of phase over Eurasia and show significant coherence at timescales of 20–28 years (Fig. 10). Winter jet speed and the PDO are in phase and show significant coherence at timescales of 28–40 years (Fig. 11).
The clearest relations occur over the North Pacific during the winter season where 50% of the variance in North Pacific winter jet latitude variability and 28% of the winter jet speed variance is explained through the correlation with the PDO index since 1940 on timescales of about 20 years (Figs. 10, 11). The coherence between the jet latitude/speed and the PDO occurs for periods of about 20 years which are consistent with one of the dominant timescales of the PDO (Mantua and Hare 2002). The correlation between the PDO and the jet latitude over the Pacific domain is substantially weaker during spring and autumn but over Eurasia we find a significant cross coherence between the PDO and jet latitude (Fig. 10), this contrasts with the winter season when the cross-coherence between PDO and the jet latitude is strongest over the Pacific but we find no relationship over Eurasia (Fig. 10).
An interesting aspect emerging from Figs. 10 and 11 is that an apparent link between the PDO and the jet stream is not confined to the Pacific but is also seen over Eurasia. In contrast to the Pacific, the strongest coherence with jet latitude is not found in winter but during spring and autumn. As for the PDO we find significant cross wavelet correlations between the NAO and the jet latitude over Eurasia during spring and autumn for periods of around 20 years but not for winter. Given the large spatial scales of both the PDO and the NAO one could expect these modes of variability to affect the jet stream over Eurasia. So why is this cross correlation not seen during winter when the cross-coherence between jet and PDO (NAO) is strongest over the Pacific (Atlantic)?
For an explanation we look at the seasonal evolution of the Siberian High (SH) and of the related cold air pool. The SH is the strongest centre of action on the Northern Hemisphere during winter. Most pronounced in winter it is also present—albeit weaker—during spring and autumn and only vanishes in summer (Fig. 12). In autumn and spring, the SH expands and wanes. The pool of cold air linked to the SH develops from September onwards in Yakutia and the Baikal region from where it gradually spreads westward reaching its full extent in January. The SH is an extremely persistent winter feature around which the jet stream has to swerve. Even though the strength of the SH varies on interannual timescales, this variability is small compared to the average winter SH strength. This is illustrated with the ratio \({R}_{i}\) in Fig. 12. \({R}_{i}\) is a measure of the average strength of surface level pressure features with respect to their interannual variability:
$$\begin{gathered} R_{i} = \frac{{\left| {\left\langle {SLP_{i} } \right\rangle - \overline{{\left\langle {SLP_{i} } \right\rangle }} } \right|}}{{\sqrt {\frac{1}{n}\sum\nolimits_{j - 1}^{n} {\left( {SLP_{ij} - \left\langle {SLP_{i} } \right\rangle } \right)^{2} } } }},\,i = [1, \ldots ,12] \hfill \\ \overline{{\left\langle {SLP_{i} } \right\rangle }} = \frac{1}{2\pi }\int\limits_{0}^{2\pi } {\left\langle {SLP_{i} } \right\rangle d\vartheta } \hfill \\ \left\langle {SLP_{i} } \right\rangle = \frac{1}{n}\sum\limits_{j = 1}^{n} {SLP_{ij} } \hfill \\ \end{gathered}$$
where SLPij is the sea level pressure for month i in the year j, < > denote the time average for the month i over the 1871–2011 period and the overbar denotes the zonal average (ϑ is the azimuth).
During the winter season the highest ratio \({R}_{i}\) occurs over eastern Siberia over the south-eastern SH. What the high values of \({R}_{i}\) over eastern Siberia in winter suggest is that even in years when the SH is very weak the pressure over eastern Siberia remains higher than at the adjacent regions further south. With \({R}_{i}\) around 5 (Fig. 12, top) the SH is more stable than any of the other centres of action of the Northern Hemisphere in winter (Icelandic Low, Aleutian Low, North American High). Note that the maximum values of the ratio \({R}_{i}\) on the Northern Hemisphere in spring and summer can exceed the winter maxima seen over Siberia. However, these values occur along the south-eastern flanks of the Azores and Pacific Highs as well as over the low pressure area over Southern Asia linked to the Asian Monsoon. All these features are located well south of the jet stream position for these seasons (Fig. 5) and hence do not constrain the jet stream path like the SH in winter.