University of Southampton 29 th August 2014 FACULTY OF NATURAL AND ENVIRONMENTAL SCIENCES OCEAN AND EARTH SCIENCE

Transcription

1 University of Southampton 29 th August 2014 FACULTY OF NATURAL AND ENVIRONMENTAL SCIENCES OCEAN AND EARTH SCIENCE OCEANIC INFLUENCES ON THE SEASONAL TO DECADAL VARIABILITY OF THE NORTHERN HEMISPHERE JET STREAM. Samantha Hallam A dissertation submitted in partial fulfilment of the requirements for the degree of M.Res. Ocean Science

2 As the nominated University supervisor of this M.Res. project by Samantha Hallam, I confirm that I have had the opportunity to comment on earlier drafts of the report prior to submission of the dissertation for consideration of the award of M.Res. Ocean Science Signed Supervisor s name: Joel Hirschi 54 55

3 Acknowledgements Support for the Twentieth Century Reanalysis Project dataset is provided by the U.S. Department of Energy, Office of Science Innovative and Novel Computational Impact on Theory and Experiment (DOE INCITE) program, and Office of Biological and Environmental Research (BER), and by the National Oceanic and Atmospheric Administration Climate Program Office. The Ferret program has provided the analysis and graphics in this paper. Ferret is a product of NOAA's Pacific Marine Environmental Laboratory. (Information is available at The MATLAB and Statistics Toolbox Release 2012b has provided the statistical analysis in this paper. MATLAB is a product of The MathWorks, Inc., Natick, Massachusetts, United States. Wavelet software was provided by C. Torrence and G. Compo, and is available at URL: I would like to thank Joel Hirschi for his guidance and support throughout this project, particularly in the use of the Ferret program. Finally, appreciation goes to my family, Toby, Jake and Gemma and my mum and dad, without whose support the project would not have been completed on time.

4 Abstract Seasonal to decadal regional variations in Northern Hemisphere jet stream latitude and speed are presented for the period from the Twentieth Century Reanalysis dataset. Jet stream variations have a significant impact on storm activity and temperature patterns across the northern hemisphere, and accordingly impact the environment and society. The ocean influence on jet variations is evident throughout the results; on a seasonal basis, there is variation in the annual range of the jet latitude across the northern hemisphere, specifically comparing ocean basins and land; with the largest range of 20 over Europe and Asia and minimum range of 9 seen over the North Atlantic. A lag is also evident in the jet latitude maximum over the ocean basins of one month or more compared land, primarily related to the influence of the larger heat capacity of the ocean. The mean jet latitude range is at a minimum in winter particularly along the western boundary of the North Atlantic and North Pacific with only a 2 and 1 range respectively across the 141 year period. The small range suggests the importance of these oceanic regions in influencing jet location. The 141 year trend in jet latitude shows significant differences on a regional basis; an increase of 3 and 1.5 in latitude is seen across the North Atlantic and Eurasia respectively, although no increase is seen across the North Pacific or North America. In terms of jet speed an increase is evident in all seasons over the North Atlantic, Eurasia and North America but there is no change in the jet speed over the North Pacific. The results highlight the influence of the ocean on jet variability and that in a changing climate the regional response across the 4 regions of the northern hemisphere in terms of jet latitude and speed are different. This is an important factor from a climate modelling perspective and for making future climate predictions in the near and longer term. Target Journal: Climate Dynamics

5 Introduction Jet streams are fast narrow bands of air, which flow around the globe in both hemispheres near the tropopause at around 10,000m (Archer and Caldeira, 2008). The flow is predominantly zonal from west to east and results from the temperature gradient between the poles and the equator and the coriolis force. Jet speeds reach between 45-70ms -1 and possibly higher in winter (Barry and Chorley, 1982). Globally there are two main jets; the polar front jet (PJ), which forms along the polar front in the region where there is a sharp temperature contrasts between polar air and subtropical air (Holton,1992) and the subtropical jet (STJ), which forms on the poleward side of the Hadley cell due to the sharp temperature gradients between the Hadley and Ferrel cells and also angular momentum (Pena-Ortiz et al., 2013). Jet streams and their seasonal to decadal variability form an important part of natural climate variability due to its influence on the mid latitude storm tracks (Hurrell,1995), which form in the region ahead of an upper level trough where reduced cyclonic vorticity causes divergence, favouring surface convergence, cyclonic circulation and storm track formation (Barry and Chorley, 1982). In winter the storm tracks are responsible for bringing heat and moisture to regions that would otherwise be much cooler and drier but they can also cause extreme weather events, both of which have a significant impact on society (Trenberth and Hurrell, 1994). Jet stream variability is therefore an important component of climate 'noise' and understanding the seasonal to decadal variability can help inform the study of what climate change will look like on a regional basis in the near term. Long term jet stream changes are being viewed as potential indicators of a changing climate (Pena-Ortiz et al., 2013). Hartmann et al., (2013) highlighted there is evidence for a poleward shift in the jet stream and storm tracks since the 1970 s but there are significant differences between studies on the magnitude of the migration and changes in jet velocities. Archer and Caldeira, (2008) found a poleward migration in the Northern Hemisphere Jet (NHJ) of / decade using NCEP-NCAR reanalysis dataset from , whilst Fu and Lin (2011) suggest the NHJ has shifted polewards by 1 ± 0.3 and the Southern Hemisphere Jets have shifted poleward by 0.6 ±1 over the 30 year period from 1979 to 2009, utilising the Microwave Sounding Unit

6 lower stratospheric channel monthly brightness temperature. Pena-Ortiz et al. (2013), however, found the winter NHJ had moved poleward by between to 0.13 /decade and the SHJ had moved poleward by 0.1 /decade using NCEP/NCAR( ) and the 20 th Century Reanalysis ( ) datasets. Woollings et al. (2014) found a poleward shift of the North Atlantic Jet (0-60 W) of 0.2 / decade using the 20 th Century Reanalysis dataset covering the period from In terms of jet speed Strong and Davis (2007) found a 15% increase in the NHJ mean speed whilst Archer and Caldeira, (2008) found a decrease of -0.2ms -1 /decade. All of the above studies were of the recent past (1958 onwards) only Woollings et al. (2014) looked at data from 1871 and then only for the North Atlantic, hence a key motivation for the work here is to study the whole northern hemisphere for the period from The summary also highlights the variety of results obtained. Some of these differences are likely to be caused by differing methodologies used to define the jet streams, dataset used, geographical area studied, and the differing time periods. The global studies outlined do not specifically identify any jet stream trends over oceanic areas compared to land masses despite research which suggests that western boundary currents (WBC), through deep atmospheric convection, can influence the entire troposphere on interannual and decadal time-scales (Sheldon and Czaja, 2013 and Czaja and Blunt, 2011). The significant sea surface temperature (SST) gradients found along WBC provides an environment for differential sensible heating, which enhances baroclinicity, and leads to surface cyclonic wind convergence and effectively anchors the storm track (Nakamura et al., 2004 and Minobe et al., 2008). The sensible heating occurs mainly where cold air from the continents flows over the warm waters (Hoskins and Valdes, 1990). For example, O Reilly and Czaja (2014) highlight that the changes in the jet stream and storm track over the western Pacific are linked to the variations in the Kuroshio Extension Front. When the surface SST gradient was strong the storm track was zonally localised, but there was less atmospheric influence when the SST gradient was weaker, in the 19 year period analysed. In addition Gan and Wu, (2013) found in the North Pacific that cold mid latitude

7 SST anomalies led to an increase in baroclinicity and poleward intensification of the storm track. As oceanic influences on the jet stream are now considered more important than previously, a key motivation for this study was to look at seasonal to decadal northern hemisphere jet stream variability and the differences over ocean basins compared to land masses in terms of patterns of variability and long term trends. Accordingly the northern hemisphere jet stream is analysed over 4 regions; North Atlantic (60 W -0 W), Eurasia (0-120 E), North Pacific (120 E-120 W) and North America (120 W-60 W). Only the northern hemisphere is included in view of the more significant land mass to provide a comparison to the ocean basins and to manage the scope of the project. Pena-Ortiz et al. (2013) highlighted that establishing longitudinal jet stream trends was an important next step. Accordingly an additional motivation is to provide a regional jet stream analysis, on a longitudinal basis, across the longest available dataset, using the same data and methodology for all regions. This does not appear to have been accomplished yet and will provide a broader understanding of the natural variability and be comparable regionally. Ensuring the full range of natural jet variability is captured in climate models, and how it varies regionally, is important for regional climate predictions in the near term. Alongside this, the regional longer-term trends in jet latitude and speed will either confirm or challenge studies, which are based on shorter timescales Data To understand the jet stream variability the longest reanalysis dataset currently available; the Twentieth Century Reanalysis (V2) (20CR) covering the whole available period from 1871 to 2011 was used. 20CR contained all the required data; air temperature, wind velocities and geopotential height. 20CR is a global atmospheric circulation reanalysis dataset spanning 1871 to 2011, which assimilates only surface pressure reports and uses an ensemble Kalman Filter data assimilation method and one global numerical weather prediction model (Compo et al. 2011).

8 Compo et al. (2006), Whitaker et al. (2004) and Anderson et al. (2005) have all shown that reliable reanalyses can be obtained of earlier periods, where only sparse data is available, using only surface pressure observations, where standard corrections are known, when combined with more advanced data assimilation methods. Compo et al. (2006), highlighted that using an ensemble Kalman filter provides results which not only cover large scale features but also many synoptic features and had a smaller analysis error when observations are sparse, when compared to other data assimilation methods. For the northern hemisphere extratropics (20-90 ) in the upper troposphere the zonal and meridional wind components have an error and anomaly correlation skill at a level of 0.8 or above from 1895 onwards for both summer and winter. The summer analysis errors are, however, larger than winter. Utilising only surface pressure reports to compile the dataset can help overcome issues from differing conventional observations (Pawson and Fiorino, 1999). For example Archer and Caldeira, (2008) utilised the ERA-40 and NCEP/NCAR datasets but only for the period from 1979 when satellite observations were available due to the differences seen in the dataset once satellite observations were introduced. Woollings et al. (2014) has, however, compared North Atlantic Jet latitude and speed from the NCEP-NCAR reanalysis and found extremely good agreement for the period from 1948 with the 20CR data. 3. Methodology Using appropriate indices to define the jet stream and validating the robustness of the 20CR data was an important part of the study and discussed in this section. Jet streams are diverse in nature and can vary spatially and temporarily which has led to different definitions being used and may possibly lead to the inconsistencies in the results outlined in the introduction. The main approaches adopted to define the jet stream are either the maximum wind speed over one or more isobaric levels or average wind speed over 30ms -1 across one or more isobaric surfaces. Woollings et al., (2014) used the maximum zonal wind at the 850mb level after identifying the results

9 were almost identical to averaging over mb level. Pena-Ortiz et al., (2013) used the maximum zonal wind above 30ms -1 and frequency at each longitude across the mb level.. Koch et al. (2006), and Strong and Davis (2007) used maximum wind speed above 30ms -1 and 27ms -1 respectively. Archer and Calderia, (2008) used the mass weighted monthly wind speed averages of the zonal and meridional components between 400 and 100mb whilst Gan and Wu, (2013) used the 300mb meridional wind velocity to define the jet stream. However, even with the more complex approaches adopted, for example by Archer and Calderia (2008), only a single jet structure was identified for the northern hemisphere; starting over the Canary Islands and ending, after circumnavigating the globe, over England. In this study a combination of the above approaches has been used. In line with Archer and Calderia (2008) the absolute wind speed has been based on the zonal and meridional (u and v) wind velocity components, to help ensure all the features of the jet stream are incorporated. An aspect of 20CR is it is based on an ensemble method, with 56 ensemble members, which provides uncertainty estimates on the results obtained. As absolute wind speeds were being computed from the zonal and meridional components it was important to ensure the absolute wind speeds were calculated for each of the ensemble members; UU ii = [uu 2 + vv 2 ] 0.5 Where [ii = 1 56] ensemble members. as opposed to the ensemble mean; UU eeee = [ uu 2 + vv 2 ] 0.5 Where indicates ensemble mean. Using the average of the ensemble means to calculate the absolute wind speeds produces speeds that are too low and can suggest different long term trends, as outlined in figure 1. Wang et al. (2013), also found that using ensemble mean data to track cyclone activity produced different results, compared to tracking each of the 56 analysis results separately and Woollings

10 et al. (2014) also identified that misleading results could be obtained using ensemble mean data Figure 1. Winter (DJF) North Atlantic Jet Speed for the period Black line indicates the jet speed based on the 56 ensemble members (UU ii ). The Blue line indicates the jet speed based on the ensemble mean (UU eeee ). The most accessible 20CR u and v component wind velocity data for each of the ensemble members was at the 250mb level. As this study was primarily concerned with longer term trends, using only one isobaric level was considered acceptable in line with the methodology adopted by Woollings et al., (2014). Accordingly, the wind components used to define the jet stream are the 6-hourly 250mb meridional and zonal wind velocity for each of the 56 20CR ensemble members spanning the 141 year period from 1871 to 2011 at 2 longitude-latitude horizontal resolution (Compo et al. 2011) The maximum windspeed was used to define jet latitude and jet speed, in line with the methodology adopted by Woollings et al. (2014, 2010). The algorithm used to calculate the jet stream proceeds as follows. First the 250mb 6-hourly zonal and meridional wind velocity for each of the 56 ensemble members for each year were obtained. From this the absolute wind velocity for each ensemble member for each year was calculated: Ui = [uu 2 + vv 2 ] 0.5 Where [ i = 1,.56] ensemble members. Then the average absolute wind velocity was calculated:

11 302 UU = 1 nn UU nn ii=1 ii for n= The jet speed was defined as the maximum value of UU at each longitude. The jet latitude was defined as the latitude of the maximum average absolute wind velocity (UU maximum), for each longitude. On occasions when the jet is split into two branches only the strongest is considered To establish the robustness of the dataset, and understand the variance of the ensemble members over the 141 year timeframe, 3 separate jet latitude standard deviations were calculated; based on the 6 hourly data, based on the 6 hourly data smoothed over 31 days using a low pass parzen filter, and based on the 6 hourly data smoothed over 91 days using a low pass parzen filter. The low-pass filter preserved low frequency jet features whilst eliminating short time-scale noise. The 3 separate jet speed standard deviations were also calculated on the same basis. The jet latitude range, across the ensemble members, based on 3 different standard deviations is illustrated in Figure 2. A seasonal cycle is evident and the range reduces significantly over the period from The largest variability in the standard deviation is where no smoothing has been applied. There is a marked reduction in the standard deviation in jet latitude during the 1930 s to around 2 to 1 for the standard deviations smoothed over 31 days and 91 days respectively. However, the spread in the data before the 1940 s, due to insufficient data coverage to constrain the model, does need to be incorporated into the interpretation of results and its potential effect, as also highlighted by Woollings et al. (2014). (Incorporate into the appendix standard deviation for jet speed and comment) Importantly, the reduction in the standard deviation spread, for the standard deviations where a 31 day or 91 day smoothing is applied, is broadly similar across all regions indicating the data is similarly robust across all regions which is very important for this study. To assist the analysis and understanding of the jet stream variability this study also uses temperature data which is obtained from the 2m air temperature monthly ensemble mean, and tropopause monthly ensemble mean as well as

12 334 geopotential height data obtained from the 300mb geopotential height monthly ensemble mean. Again all data is from the 20CR dataset for the period 1871-

13 Results 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 and The latter period is used for comparison as the spread across the ensemble members is significantly lower after Seasonal Jet Latitude Climatology. The jet latitude seasonal cycle shows a poleward shift in the summer months but the are regional variations in the amplitude and lag with respect to insolation which are highlighted in Figure 3. For the full analysis period from (Figure 3a), the jet stream amplitude is greatest over land; for Europe and Asia the range is 20 from N and for North America the range is 19 from N. Over the ocean basins the range is lower; the North Pacific range is 17 from N whilst the North Atlantic range is 9 from N. Reviewing the period from (Figure 3b) there is a general reduction in the peak latitude in the summer months by around 3 over land and 2 over the ocean basins. Again the jet stream amplitude remains greatest over land; for Europe and Asia the range is 17 from N and for North America the range is 16 from N. Over the ocean basins the range is lower; the North Pacific range is 16 from N whilst the North Atlantic range is only 7 from N. A large part of the change seen, between the 2 periods, is most likely to be related to the changes in the variance of the ensemble spread, which was greatest in summer. In terms of jet stream latitude maximum and minimum, for the full period from , the regional pattern over for Europe and Asia most closely fits with insolation with a minimum in February and maximum in July. For North

14 370 (a) Seasonal Jet Latitude Climatology (b) Seasonal Jet Latitude Climatology Figure 3. Regional Variations in the Seasonal Cycle of the Jet Latitude in the Northern Hemisphere. (a) The seasonal climatology for the period (B) The seasonal climatology for the period Green line indicates Europe and Asia (0-120 E), Blue line indicates North Pacific (120 E-120 W), Red Line indicates North America (120 W-60 W), Black Line indicates the North Atlantic (60 W-1 W). The lower plot provides a regional view of the areas considered overlaying the annual average 2 m air temperature.

15 Figure 4. Seasonal Cycle of the Jet Stream Latitude in the Northern Hemisphere by region for periods (a), (c), (e), (g) and (b), (d), (f), (h). Black line is mean jet latitude. Grey area is ± 2 standard deviations smoothed over 31 days using a parzen filter based on the 56 ensemble members. Green line is ±2 standard deviations based on the interannual variability for the period. 388

16 America the peak is also in July but a minimum in March. Over the oceans there is a lag with the maximum latitude reached in August and September for the North Pacific and North Atlantic, respectively. The minimum latitude over the North Atlantic is seen in March. Also of interest is the very different shapes seen in the seasonal cycle curves, which is further illustrated in Figure 4. Over land the shape broadly follows a sinusoidal curve with a smooth more linear oscillation. Over North America there is a narrow peak. The peak is broader over Europe and Asia, particularly for the period , with a steep decline seen in October. The shape is more non-linear over the ocean basins. Over the North Pacific there is a very steep rise from May to July of 9 with a narrow peak then a steep fall of 5 in August. In the North Atlantic the shape is different again. There is a broad flat line from January to May where there is virtually no increase in latitude. From May there is a steady increase to September before a steady fall is seen from September to December. Looking at the seasonal variations as a global picture (Figure 5) indicates the regional variations in jet stream latitude are greatest in the winter months (DJF), resulting in 2 waves around the hemisphere with larger amplitude particularly over North America and the North 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. The Jet stream ridges in winter are located on the eastern side of the North Atlantic and North Pacific, where it travels over the Rocky Mountains. In spring (MAM) 3 waves are evident, although with lower amplitude than in winter. The additional loop arises as the jet stream passes to the north of the Himalayas in all but the winter months. The jet stream troughs have moved northwards but remain over the western boundaries of the North Pacific (37 N) and North Atlantic (43 N). 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). In Autumn (SON) the path

17 Figure 5. Mean Seasonal Jet Stream Position overlaying the 2m air temperature for the period The dark blue line indicates the mean jet stream position and the green line ± 2 standard deviations of the 6 hourly jet latitude smoothed over the 91 day period shown based on the 56 ensemble members. The jet stream overlays the seasonal average 2 m air temperature.

18 has moved equatorwards and a jet stream with 2 waves is evident. Of particular note is the narrow range in the mean jet latitude position in winter over the 141 year period along the western boundary of the North Atlantic (41 N) and North Pacific (33 N). The range is only 2 over the North Atlantic and 1 over the North Pacific and the jet is located about 1 to the north of the maximum gradient of the 2m air temperature and is aligned with temperature contours. The relationship between the mean jet position along the western boundary, and maximum gradient of 2m 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 2m air gradient and land/sea temperature contrast 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 2mair temperature and mean jet latitude would also been seen over land. Although there is some relation it is not as tightly defined. To examine the relationship further the correlation between the jet latitude and the 2m air temperature was undertaken and the results for the period are highlighted in Figure 6. The correlation is generally strongest just south of the mean jet stream position, as might be expected but it is not uniform in strength across all latitudes and seasons. The correlation is strongest in winter over the oceans; specifically the Eastern North Pacific (40 N 160 W,) at over 0.8, and the western boundary of the North Atlantic (30 N 80 W) at 0.7. Perhaps surprisingly the correlation is not as strong on the western boundary of the North Pacific as in the central and East North Pacific. In spring (MAM) the correlation is strongest across the North Pacific at 0.6.

19 Figure 6. Jet Latitude correlation with 2m air temperature for the period on a seasonal basis. (a) DJF, (b) MAM, (c) JJA, (d) SON. Correlations over 0.4 are significant at the 95% confidence level. Correlations over 0.6 are significant at 99% confidence level.

20 The correlation is strongest in summer over land specifically the east coast of America (55 N 80 W) at 0.6 and Eurasia (45 N 100 E) at 0.7. There is no significant correlation over the North Atlantic.The most significant correlation exists in the Autumn (SON) across the Pacific reaching a maximum at 0.6. For completeness the correlations for the full period from 1871 are included as supplementary Figure 1 but the correlation is not as strong as the last 71 years related to the uncertainty in the ensemble members data for the earlier period. In view of the tightly defined mean jet position seen in winter over the WBC of the North Atlantic where the 2m air temperature gradient was strongest a correlation between the maximum gradient at the western boundary and jet latitude was undertaken. Significant correlations (97% or higher) were seen on all timeframes explaining 25% of the variance since the 1940 s, 45% of the variance since 1970 s and 80% of the variance between Seasonal Jet speed Climatology The seasonal cycle of jet speed shows a similar pattern across land and the North Atlantic and displays a smooth linear curve with maximum wind speeds seen in January around 59ms -1 and minimum in July around 40ms -1. The pattern over the North Pacific is not linear, and resembles an inverse pattern 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 66ms -1 and minimum in July at 38ms -1. The main change observed between the 1871 and 2011 and is that in the latter period the summer speeds have decreased by around 3ms -1 to 39ms -1 whilst winter speeds have increased across North America and the North Atlantic. Decadal trends are outlined in section 4.4.

21 Figure 7. Seasonal Jet Speed Climatology. (a) (b) Green line indicates Europe and Asia (0-120 E), Blue line indicates North Pacific (121 E- 118 W), Red Line indicates USA (119 W-61 W), Black Line indicates the North Atlantic (60 W-1 W). The lower plot provides a regional view of the areas considered overlaying the annual average 2 m air temperature.

22 Decadal Jet Latitude Variability Jet latitude shows differing trends in each season for the northern hemisphere (Figure 8). Significant trends are at the 95% confidence level or higher. 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. In spring (MAM) there is no change in the mean of 44.3 N. In summer (JJA) there appears to be a significant drop in the jet latitude during the 1930 s resulting in an overall decreasing trend of 6 from 58.5 N to 52.1 N. Care is needed interpreting this change because summer is the season of greatest range in the standard deviation of the ensemble members before 1940, and since 1940 there has been a 0.3 N increasing trend 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. However, when trends are analysed seasonally on a regional basis a different picture unfolds (Figure 9). In winter (DJF) a 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 long term trend with the jet latitude mean remaining around 38.8 N and 39.5 N, respectively. In Spring (MAM) only the North Atlantic shows a significant increasing trend in mean jet latitude of 1.8 N (0.1 /decade) from 45.6 to 47.4 N (supplementary Figure 2.). Over Eurasia there is no change in the mean of 43.6 N. The Pacific and North America show a decreasing trend until the 1940 s and increasing thereafter, but the increase is not statistically significant. For the summer (JJA) period only the trend after 1940 is considered, due to the range in the standard deviation in the earlier period. Only Eurasia shows a significant increase of 1.6 N (0.1 /decade) from 51 to 52.6 N. The North Atlantic has a 0.4 N increase to 52.6 N. The Pacific and North America show a 0.3 and 1.2 decrease, respectively but neither is statistically significant. 531

23 Figure 8. Jet Latitude Variability for the Northern Hemisphere from (a)winter jet latitude(djf). (b) Spring jet latitude (MAM). (c) Summer jet latitude (JJA). (d) Autumn jet latitude (SON). The thick black line indicates the seasonal mean. The red line indicates the seasonal mean with a Parzen filter smoothing over 11 years. The blue line indicates the 5-year running mean. The thin black lines indicate ±2 standard deviations based on the 6 hourly data for the 56 ensemble members smoothed over 91 days.

24 Figure 9. Winter decadal trend in jet Latitude by region from The thick black line indicates the seasonal mean. The red line indicates the seasonal mean with a Parzen filter smoothing over 11 years. The blue line indicates the 5-year running mean. The thin black lines indicate ±2 standard deviations based on the 6 hourly data for the 56 ensemble members smoothed over 91 days.

25 In Autumn (SON) the North Atlantic is not in line with the hemisphere trend and with an increase of 0.8 N, although not statistically significant. The North Pacific and Eurasia show no change whilst North America shows a significant decrease of 0.5 (0.03 /decade). 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 small decreases in the seasons. 4.4 Decadal Jet Speed Variability When looking at the jet speed a similar trend to jet latitude might be expected and from a whole northern hemisphere perspective the trends are similar. In winter a jet speed significant increase of 2.0ms -1 (0.1ms -1 /decade) is seen. In the other seasons the trend is for a decrease in jet speed from but then a modest increase thereafter of between ms -1, but the increases are not statistically significant. Again however, when the regional picture is analysed for each season differences emerge; increases are seen in jet speed in all regions, in all seasons, except the North Pacific. In winter significant jet speed increases are seen in North America of 4.73ms -1 (0.3 ms -1 /decade), North Atlantic of 4.52ms -1 (0.3 ms -1 /decade) and Eurasia of 1.8 ms -1 (0.1 ms -1 /decade). In Spring the regional trend is in line with winter with the largest significant increase seen over the North Atlantic of 2.4ms -1 (0.2 ms -1 /decade), North America has a 1.2ms -1 (0.1 ms -1 /decade) increase, and Eurasia 1.0ms -1 (0.1 ms -1 /decade). In summer, only the position after 1940 is considered, a significant increasing trend is seen in North America of 1.6ms -1 (0.2 ms -1 /decade). In Autumn the North Atlantic show a 3 ms -1 (0.2 ms -1 / decade) significant increase and the USA a 1.2ms -1 (0.2ms -1 /decade) significant increase for the period since Over Eurasia the increasing trend seen is not significant after 1940.

26 Figure 10. Jet Speed for the Northern Hemisphere from (a)winter jet speed(djf). (b) Spring jet speed (MAM). (c) Summer jet speed (JJA). (d) Autumn jet speed(son). The thick black line indicates the seasonal mean. The red line indicates the seasonal mean with a Parzen filter smoothing over 11 years. The blue line indicates the 5-year running mean. The thin black lines indicate ±2 standard deviations based on the 6 hourly data for the 56 ensemble members smoothed over 91 days.

27 Figure 11. Winter decadal trend in jet speed by region from The thick black line indicates the seasonal mean. The red line indicates the seasonal mean with a Parzen filter smoothing over 11 years. The blue line indicates the 5-year running mean. The thin black lines indicate ±2 standard deviations based on the 6 hourly data for the 56 ensemble members smoothed over 91 days.

28 Over the Pacific in winter, as with jet latitude, there is no change in the mean jet speed of 67.7ms -1 but there is significant interannual variability from 62 to 72ms -1. In the other seasons a significant decreasing trend is seen until the 1940 s thereafter the mean jet speed displays seasonal decreases between ms -1, which are not significant. The extent of the relationship between jet latitude and speed was also tested. Whilst a significant relationship exists in all season across both the North Atlantic and North Pacific no significant relationship between jet latitude and speed is present over land. Over the North Pacific there is a negative correlation in all seasons, which is strongest in winter explaining 41% of the variance. Over the North Atlantic there is a positive correlation in winter and spring but negative in summer and autumn. The strongest correlation was in winter but only explains 9% of the variance. 4.5 Interannual Variability of Jet Latitude and Speed The plots showing decadal trends have highlighted some of the interannual variability in jet latitude and speed. Now the interannual variability is analysed in more detail using wavelet analysis and comparison to known indices; North Atlantic Oscillation(NAO), Pacific Decadal Oscillation (PDO) and Atlantic Meridional Oscillation (AMO). Wavelet analysis, using a continuous wavelet transformation, analyses the variations in power in a timeseries and determines the dominant modes of variability and how those modes vary over time (Torrence and Compo, 1998). A Chi squared test is then applied to find the 95% confidence level contour. Once again differing pictures emerge in the interannual variability in jet latitude and speed on a regional basis. Over land significant trends are seen in spring and autumn (supplementary Figures 8 and 10). Over the oceans significant trends are seen in winter and summer (Figure 12 and supplementary Figure 9).

29 Figure 12. Winter Interannual Variability of Jet Latitude and Speed by Region. Colour bar indicates signal strength. Black circles indicate statistically significant features.

30 Significant interannual variability is seen in defined areas. Over the North Pacific in winter a 20 year variability in jet latitude and speed exists. Over the North Atlantic in summer a year band of variability exists in jet speed. Over Eurasia a 24 year variability exists in jet latitude in spring and autumn. Over North America a 32-year variability in jet speed exists in spring. Shorter timescale events between 2 and 6 years are also seen but not consistently across the timeseries. Tests have been undertaken to establish the extent to which the interannual trends identified in jet latitude and jet speed correlated with well known indices. The North Atlantic trends were correlated with the North Atlantic Oscillation (NAO) index (Hurrell, 1995) which is based on fluctuations in the difference of atmospheric pressure at sea level between the Icelandic low and the Azores high. In winter and spring a significant positive correlation was found with jet latitude explaining 32% of the winter variance. In summer and autumn a significant negative correlation was found but only 6% of the variance was explained. A significant positive correlation with jet speed and NAO was found in all seasons with the maximum variance of 21% explained in spring. The Atlantic Meridional Oscillation (AMO) Index (reference), which highlights the pattern of sea surface Temperature changes in the North Atlantic was correlated with the North Atlantic jet latitude. The correlation across all seasons is low at 19% versus the AMO Warm index (where no warming has been removed) and only explains 10% of the variance, in line with the findings by Woollings et al. (2013). The Pacific Decadal Oscillation index (reference), highlights the pattern of SST anomalies in the North Pacific and has been correlated with the North Pacific jet latitude and jet speed. With respect to jet latitude there is a significant negative correlation in all seasons with the highest correlation in winter explaining 50% of the variance. With respect to jet speed there is a positive correlation but only significant in winter and spring again with the highest correlation in winter explaining 28% of the variance.

31 Discussion of Results Utilising the 20CR dataset has enabled the jet latitude and speed to be analysed over a 141 year period across the 4 regions of the Northern Hemisphere, whilst also applying an uncertainty to the results obtained which has assisted the interpretation. This study has highlighted that the jet variability, on seasonal to decadal timescales, is different in each region; North Atlantic, North Pacific, Eurasia and North America. Although some similarities exist there are significant differences suggesting separate mechanisms are moderating the jet behaviour, which are now discussed in more detail. These finding have important implications for climate prediction on a seasonal to decadal basis. 5.1 Seasonal Jet Climatology Over the oceans, particularly the North Atlantic, the annual range in jet latitude is significantly lower. This is linked to the lower variability in the seasonal 2 m air temperature over the oceans, which is due to the greater heat capacity of the oceans of 6 x J K -1 compared to the atmosphere of 5 x10 21 J K -1. The larger oceanic heat capacity results in an accumulation of heat in the mid latitudes oceans during the summer, which is released during the winter months and reduces the overlying temperature range. The slow release of heat also seems to act as a lag to the jet stream reaching its maximum and minimum latitudes. Over the North Atlantic and North Pacific the maximum is in September and August respectively, suggesting how the ocean is moderating the jet latitude on a seasonal basis. These findings are similar to Woollings et al. (2014) with a peak maximum latitude in September over the North Atlantic, although Woollings et al. (2014) show a seasonal range in latitude of 5 compared to the 9 found in this study, perhaps related to the different pressure heights analysed, 850mb versus 250mb here. There is, however, also a significant variation in the amplitude of the jet latitude between the North Atlantic and the North Pacific. A key difference between these basins is the poleward meridional heat transport (MHT) by the oceans. In the North Atlantic the oceanic MHT is from 20 S and reaches a maximum of 1.2PW at 25 N, transported by the Atlantic Meridional

32 Overturning Circulation (AMOC) (Johns et al.., 2011). The oceanic MHT means the North Atlantic region is 4 C warmer than expected at similar latitudes (Levitus, 1982) whilst also reducing the seasonal range in the 2m air temperature. The oceanic MHT poleward in the Pacific is lower at 0.76PW at 25 N (Bryden et al., 1991) which could be the reason for the larger seasonal range in temperature and greater amplitude in jet latitude and may suggests changes in poleward heat transport by the oceans could impact the seasonal variations in the jet latitude but further research on this hypothesis would be required. The findings seem in line with the Bjerknes compensation, however, where an increased oceanic MHT leads to a decrease in poleward temperature gradient and results in a reduced atmospheric MHT (Van der Swaluw et al., 2007). The tightly defined winter mean jet stream position (2 and 1 respectively) over the extended period from over the western boundary of the North Atlantic and North Pacific in the region where the 2m air temperature gradients are strongest is a key finding in this study and in line with the findings by O Reilly and Czaja (2014) for the western Pacific who found that where the SST gradient was strong the jet stream and storm track where zonally localised but there is less atmospheric influence when the SST gradient is weaker. Nakamura et al. (2004) also found that strong SST gradients along western boundary currents enhance baroclinicity and effectively anchors the storm track. Perhaps most importantly the findings support the study by Minobe et al. (2008) that the Gulf Stream, on the warmer flank of the Gulf Stream front, leads to deep convection impacting from the surface to the tropopause. The significant correlation between location of maximum gradient of 2m air temperature on the western boundary of the ocean and jet latitude is also key accounting for 45% of the variance since 1970 and 80% of the variance from 2001 to 2011 in the North Atlantic. These findings are important for climate modelling and climate prediction and suggest changes in the maximum gradient in the temperature on the WBC of the North Atlantic could be used to predict winter jet latitude, although significant further work would be required to validate this. Ensuring the ocean has a sufficiently high resolution to accurately capture SST gradients over the western boundary currents (WBC) would be important

33 to this prediction. Scaife et al. (2011, 2014) found increasing the ocean resolution from 1 to 1/4 improved the modelled position of the North Atlantic Current and forecasting skill in winter over the North Atlantic. In an evaluation of seasonal jet speed, it would appear there is less of an oceanic influence, compared to jet latitude, with the North Atlantic following a similar amplitude and curve to Eurasia and North America suggesting different mechanisms are effecting jet latitude and speed. This is in line with the findings by Woollings et al. (2014) for the North Atlantic. The North Pacific has the highest significant anti correlation between jet latitude and speed explaining 41% of the variance in winter and the seasonal cycle curves broadly resemble a mirror image, which suggests different mechanisms are affecting jet latitude in the North Atlantic compared to the North Pacific Decadal Jet Latitude Variability For the Northern hemisphere as a whole there is a significant 0.1 /decade increase in winter, no significant trend in spring or summer and a significant 0.1 /decade decrease in autumn which is in line with the seasonal findings by Pena-Ortiz et al. (2013) using the same data but the winter and autumn trends were not significant due to the shorter period analysed. However, the long term trend from in jet latitude is not uniform across the northern hemisphere. Only the North Atlantic shows an increasing trend across all seasons, with a maximum increase seen in winter of 3.0 (0.2 /decade) which is in agreement with Woollings et at (2014). Eurasia shows significant increases but only in winter (1.5, 0.1 /decade), in line with Strong and Davies (2007) and summer (1.6, 0.11 /decade) where the increase is greater than the North Atlantic. The North Pacific and North America have either no change or small decreases over the seasons which does not agree with the work of Strong and Davies (2007) who found an increasing trend over the East Pacific between It is possible the difference in regional trends are due to the difference in the uncertainty in the dataset on a regional basis with the best available data in

34 the North Atlantic and most sparse dataset in the early years in the North Pacific. However, it is unlikely as it has been accounted for in the interpretation of the results by using data post 1940 for the summer period, and also where the jet latitude trend changes materially after Two broad patterns of jet latitude have emerged covering the North Atlantic/Eurasia where increasing trends are seen and in the North Pacific/ North America where flat or decreasing trends are observed. The increasing jet latitude trends seen over the North Atlantic and Eurasia are in line with other northern hemisphere research (Woollings et al. (2014), Pena-Ortiz (2013) and consistent with decreasing temperature gradient between the poles and the equator in the northern hemisphere as highlighted in Figure Figure 13. Temperature Difference between the Equator and Poles at the Tropopause for the period Pink line indicates winter (DJF), Green line spring (MAM), Blue Line summer (JJA), Black line autumn (SON) Decadal Jet Speed Variability For the whole northern hemisphere there has been a significant 0.1ms -1 /decade increase in jet speed in winter, which is lower than the significant findings by Pena-Ortiz (2013) using the NCEP/NCAR data for the shorter period. For the other seasons a decrease was seen from and very modest increases thereafter which are not significant. The findings are in agreement with PenaOrtiz (2013) using the CR2 data set but lower that the increasing trends in speeds seen using NCEP/NCAR data for the shorter period.

35 Again, however this masks the regional trend where increasing trends are seen in all seasons in the North Atlantic, Eurasia and North America, many of which are significant and up to 0.3ms -1 /decade in winter. The findings are in agreement with PenaOrtiz (2013) significant increasing trends and magnitudes in speeds seen using NCEP/NCAR data for the shorter period. The North Pacific region shows no change in winter wind speed and decreasing trends in the other seasons this is not in agreement with Strong and Archer (2007) who find a 1.75ms -1 /decade increasing trend over the west and central Pacific between The northern hemisphere and regional trends in jet speed identified in this study are consistent with changes in the 300mb geopotential height gradient between the poles and equator where there is an increasing trend of up to 50mb in each season is seen (Figure 14) and would be expected to lead to an increase in jet speed. Over the North Pacific there is a decreasing trend in geopotential height gradient, consistent with the decreasing wind speed trend seen. 5.4 Interannual Variability The importance of the oceanic influence is again seen on interannual timescales with 50% of the variance in North Pacific winter jet latitude variability explained through the correlation with the PDO index which is similar to research findings by Gan and Wu (2013) who found variability of wintertime storm tracks closely related to the PDO. A 20 year variability is seen over the North Pacific in winter which is in line with the broad PDO timescales (reference) with a decrease in jet latitude and stronger wind speeds seen when the PDO moves into a positive phase. The 20 year timescale is also evident with respect to jet latitude in spring and autumn over Eurasia suggesting a wider impact of the PDO. Shorter timescale events between 2-6 years, most prominent over the North Pacific but also seen in other regions appear to be in line with El Nino Southern Oscillation (ENSO) events, and the findings by Torrence and Compo (1998) but are not consistently seen throughout the timescale.

36 Figure mb Geopotential Height Difference between the poles (70-90 N) and equator (0-20 N) for the period from (autumn plot to be added) 825

37 In the North Atlantic there is no evidence of significant interannual variability in jet latitude in any season. In this study the NAO index explains 32% of the variance in jet latitude similar to findings by (reference) and it is difficult to understand why no significant interannual variability is seen. Perhaps this is related to other research, which suggests the phasing of the NAO index is adjusting with a changing climate (reference). The interannual variability has shown significant timescales up to 32 years and other variability up to 40 years in all regions across both jet latitude and speed. Care is therefore needed with research based on timescales of 4 decades or less, as they may not be capturing a decadal trend only observing interannual variability. 6. Conclusions. For jet latitude and speed this study has shown the CR2 dataset to be a robust in all regions from 1871, with the exception of summer which is only robust from 1940, in view of the general good agreement with other results on shorter timescales. It is also evident that jet latitude and speed studies need to be carried out on a regional basis other wise trends and magnitudes in the subset of the data may be masked. Importantly, the methodology for defining the jet stream appears robust providing, when using the CR2 dataset, wind speeds are calculated based on the 56 individual ensemble members not using the ensemble mean to ensure the accuracy of the speeds and trends. Oceanic influences on jet latitude and speed are seen from seasonal to decadal timescales. Seasonally the ocean acts to reduce the amplitude of the jet latitude, particularly over the North Atlantic, where the oceanic MHT is greatest, where a 10 reduction is seen. Oceans also introduce a 1-2 month lag in the seasonal maximum and minimum latitude, compared to land, related the greater oceanic heat capacity.

38 Winter jet variability is very tightly confined over the western boundary of the North Atlantic (41 N) and North Pacific (33 N) with a range of only 2 and 1 over the 141 year period, located just north of the maximum gradient in the temperature gradient. The correlation of jet latitude and maximum gradient on the western boundary accounts for 45% of the variance since 1970 and 80% of the variance from 2001 to 2011 in the North Atlantic and perhaps it may prove better method of jet latitude prediction than the NAO. Interannually PDO and ENSO are seen to influence both jet latitude and speed at timescales of 20 years and 2-6 years, respectively. Interannual variability has been identified up to 40 years suggesting studies on shorter timescales require caution when interpreting the trends observed. On a decadal basis increases in jet latitude are seen in all seasons in the North Atlantic (0.2 /decade in winter) and over Eurasia increases are seen in winter and summer (0.1 /decade) but no increasing trends are seen over the North Pacific and North America. Increases in jet speed have been seen in all regions, in all seasons over the North Atlantic, Eurasia and North America. Over the North Pacific no significant change in jet speed has been found. These trends are consistent with the changes in the geopotential height gradients over the period and region. Overall, these findings have significant implications for climate prediction in both the near and longer term to ensure; seasonal cycles are accurately reflected on a regional basis, sufficient ocean resolution over the western boundary currents to ensure the gradients are accurately captured, variations in decadal trends on a regional basis are observed.

39 References Anderson JL, Wyman B, Zhang S, Hoar T (2005) Assimilation of surface pressure observations using an ensemble filter in an idealized global atmospheric prediction system. J Atmos Sci 62: Archer CL, Caldeira K (2008) Historical trends in the jet streams. Geophys Res Lett 35: Ll /2008GL Barry BG, Chorley RJ (1982) Atmosphere, Weather and Climate. 4 th Ed. Methuen, New York, 407 pp. Bryden HL, Roemmich DH, Church JA (1991) Ocean heat transport across 24 N in the Pacific. Deep Sea Research Part A. Oceanographic Research Papers 38: Compo GP, Whitaker JS, Sardeshmukh PD (2006) Feasibility of a 100- year reanalysis using only surface pressure data. Bull Amer Meteorol Soc 87: Compo GP, Whitaker JS, Sardeshmukh PD, Matsui N, Allan RJ, Yin X, Gleason BE, Vose RS, Rutledge G, Bessemoulin P, Brönnimann S, Brunet M, Crouthamel RI, Grant AN, Groisman PY, Jones PD, Kruk MC, Kruger AC, Marshall GJ, Maugeri M, Mok HY, Nordli Ø, Ross TF, Trigo RM, Wang XL, Woodruff SD, Worley SJ (2011) The Twentieth Century Reanalysis Project. Q J R Meteorol Soc 137: DOI: /qj.776 Czaja A, Blunt N (2011) A new mechanism for ocean atmosphere coupling in midlatitudes. Q J R Meteorol Soc 137: doi: /qj.814 Fu Q, Lin P (2011) Poleward shift of subtropical jets inferred from satelliteobserved lower stratospheric temperatures. J Clim 24: doi: / JCLI-D

40 Gan B, Wu L (2013) Seasonal and Long Term Coupling between Storm Tracks and Sea Surface Temperature in the North Pacific. J Clim 26: Hartmann DL, Klein-Tank AMG, Rusticucci M, Alexander LV, Bronnimann S, Charabi Y, Dentener FJ, Dlugokencky EJ, Easterling DR, Kaplan A, Soden BJ, Thorne PW, Wild M, Zhai PM (2013) Observations: Atmosphere and Surface. In: Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. [Stocker TF, Qin D, Plattner GK, Tignor M, Allen SK, Boshung J, Nauels A, Xia Y, Bex V, Midgley PM (eds.)] Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA. Holton JR (1992) An Introduction to Dynamic Meteorology. Elsevier, New York, 511 pp. Hoskins BJ, Valdes PJ (1990) On the existence of storm-tracks. J Atmos Sci 47: Hurrell JW (1995) Decadal trends in the North Atlantic Oscillation: Regional temperatures and precipitation. Science 269: Johns WE, Baringer MO, Beal LM, Cunningham SA, Kanzow T, Bryden HL, J. Hirschi JJM, Marotzke J, Meinen CS, Shaw B, Curry R (2011) Continuous, Array-Based Estimates of Atlantic Ocean Heat Transport at 26.5 N. J. Clim 24: doi: Koch P, Wernli H, Davies HC (2006) An event-based jet-stream climatology and typology. Int J Climatol 26: DOI: /joc.1255 Levitus, S (1982) Climatological atlas of the world ocean. Noaa Professional Paper. 13 Minobe S, Kuwano-Yoshida A, Komori N, Xie SP, Small RJ (2008) Influence of the Gulf Stream on the troposphere. Nature 452: doi: / nature06690

41 Nakamura H, Sampe T, Tanimoto Y, Shimpo A (2004) Observed associations among storm tracks, jet streams and midlatitude oceanic fronts. In: Wang C, Xie SP, Carton JA (eds) Earth s climate: the ocean-atmosphere interaction, vol 147. AGU Geophysical Monograph Series O'Reilly CH, Czaja A (2014) The response of the Pacific storm track and atmospheric circulation to Kuroshio Extension variability. Q J R Meteorol Soc doi: /qj.2334 Pawson S, Fiorino M (1999) A comparison of reanalyses in the tropical stratosphere. Part 3: inclusion of the pre-satellite data era. Clim Dyn 15: Pena-Ortiz C, Gallego D, Ribera P, Ordonez P, Alvarez-Castro MDC (2013) Observed trends in the global jet stream characteristics during the second half of the 20th century. J Geophys Res Atmos 118: doi: /jgrd Scaife AA, Copsey D, Gordon C, Harris C, Hinton T, Keeley SJ, O'Neill A, Roberts M, Williams K (2011) Improved atmospheric blocking in a climate model. Geophys Res Lett 38: L23703, doi: /2011gl Scaife, AA, Athanassiadou M, Andrews M, Arribas A, Baldwin M, Dunstone N, Knight J, MacLachlan C, Manzini E, Müller WA, Pohlmann H, Smith D, Stockdale T, Williams A (2014) Predictability of the quasi-biennial oscillation and its northern winter teleconnection on seasonal to decadal timescales. Geophys Res Lett 41: doi: /2013gl Sheldon L, Czaja A (2014), Seasonal and interannual variability of an index of deep atmospheric convection over western boundary currents. Q J R Meteorol Soc 140: doi: /qj.2103 Strong C, Davis RE (2007) Winter jet stream trends over the Northern Hemisphere. Q J R Meteorol Soc 133:

42 Torrence C, Compo GP (1998) A practical guide to wavelet analysis Bull Am Meteorol Soc 79: Trenberth KE Hurrell JW (1994) Decadal atmosphere-ocean variations in the Pacific. Clim Dyn 9: Van der Swaluw E, Drijfhout SS, Hazeleger W (2007) Bjerknes Compensation at High Northern Latitudes: The Ocean Forcing the Atmosphere. J. Climate, 20: doi: Whitaker JS, Compo GP, Wei X, Hamill TM (2004) Reanalysis without radiosondes using ensemble data assimilation. Mon Weather Rev 132: Wang XL, Feng Y, Compo G, Swail V, Zwiers F, Allan R, Sardeshmukh P (2012) Trends and low frequency variability of extra-tropical cyclone activity in the ensemble of twentieth century reanalysis. Clim Dyn 40: doi: / s Woollings T, Hannachi A, Hoskins B (2010) Variability of the North Atlantic eddy-driven jet stream. Q J R Meteorol Soc 649: Woollings T, Czuchnicki C, Franzke C (2014) Twentieth century North Atlantic jet variability. Q J R Meteorol Soc 140: doi: /qj.2197

43 1019 Appendix Supplementary Figure 1. Jet Latitude Correlation with 2m air temperature for the period on a seasonal basis. (a) DJF, (b) MAM, (c) JJA, (d) SON.

44 Supplementary Figure 2. Spring (MAM) Jet Latitude Decadal Trend by region. The thick black line indicates the seasonal mean. The red line indicates the seasonal mean with a Parzen filter smoothing over 11 years. The blue line indicates the 5- year running mean. The thin black lines indicate ±2 standard deviations based on the 6 hourly data for the 56 ensemble members smoothed over 91 days.

45 Supplementary Figure 3. Summer (JJA) Jet Latitude Decadal Trend by region. The thick black line indicates the seasonal mean. The red line indicates the seasonal mean with a Parzen filter smoothing over 11 years. The blue line indicates the 5- year running mean. The thin black lines indicate ±2 standard deviations based on the 6 hourly data for the 56 ensemble members smoothed over 91 days.

46 Supplementary Figure 4. Autumn (SON) Jet Latitude Decadal Trend by region. The thick black line indicates the seasonal mean. The red line indicates the seasonal mean with a Parzen filter smoothing over 11 years. The blue line indicates the 5- year running mean. The thin black lines indicate ±2 standard deviations based on the 6 hourly data for the 56 ensemble members smoothed over 91 days.

47 Supplementary Figure 5. Spring (MAM) Jet Speed Decadal Trend by region. The thick black line indicates the seasonal mean. The red line indicates the seasonal mean with a Parzen filter smoothing over 11 years. The blue line indicates the 5- year running mean. The thin black lines indicate ±2 standard deviations based on the 6 hourly data for the 56 ensemble members smoothed over 91 days.

48 Supplementary Figure 6. Summer (JJA) Jet Speed Decadal Trend by region. The thick black line indicates the seasonal mean. The red line indicates the seasonal mean with a Parzen filter smoothing over 11 years. The blue line indicates the 5- year running mean. The thin black lines indicate ±2 standard deviations based on the 6 hourly data for the 56 ensemble members smoothed over 91 days.

49 Supplementary Figure 7. Autumn(SON) Jet Speed Decadal Trend by region. The thick black line indicates the seasonal mean. The red line indicates the seasonal mean with a Parzen filter smoothing over 11 years. The blue line indicates the 5- year running mean. The thin black lines indicate ±2 standard deviations based on the 6 hourly data for the 56 ensemble members smoothed over 91 days.

50 Supplementary Figure 8. Spring Interannual Variability of jet latitude and jet speed by region. Colour bar indicates signal strength. Black circles indicate statistical significance.

51 Supplementary Figure 9. Summer Interannual Variability of jet latitude and jet speed by region. Colour bar indicates signal strength. Black circles indicate statistical significance.

52 Supplementary Figure 10. Autumn Interannual Variability of jet latitude and jet speed by region. Colour bar indicates signal strength. Black circles indicate statistical significance.