Supplemental material
Subset / IVT (106kg/ms) / Zd (m) / Qc (104kg/m2) / Pdrop (hPa)Present / All / 1.47 / 100.06 / 0.92 / 36.62
PT / 1.46 / 96.34 / 1.03 / 30.36
ET / 1.49 / 107.50 / 0.71 / 49.12
Near Future / All / 1.62 / 96.21 / 0.99 / 33.85
PT / 1.62 / 80.98 / 1.24 / 26.90
ET / 1.61 / 107.93 / 0.79 / 39.20
Future / All / 2.14 / 96.13 / 1.27 / 37.02
PT / 2.19 / 89.05 / 1.47 / 28.28
ET / 2.06 / 106.50 / 0.99 / 49.85
Table S1 Average forcings during the last 48 hours before t0 and the associated pressure drop in this period. The IVT is now very similar for the different subsets, with only a slightly higher value for post-tropical systems in the future and a generally increasing trend towards future simulations. Baroclinic forcing is consistently higher for ETCs with no significant trend while a slight decrease is observed for PT systems. The amount of core moisture is considerably larger for PT systems as they carry tropical air into the mid-latitudes. Similar to IVT, the amount of moisture increases in the future for both types of storms. Although storms seem to become more intense in the future, there are no clear trends in their average pressure drop. As they have a higher initial minimum MSLP, extratropical systems generally experience larger pressure drops during their intensification.
EOF / Var. (%) / Pdrop (hPa) / IVT (106kg/ms) / Zd (m) / Qc (104kg/m2)Present / 1 / 71 / 0.54 / 0.53 / 0.48 / 0.44
2 / 15 / 0.30 / -0.16 / 0.55 / -0.76
Near Future / 1 / 53 / 0.22 / 0.60 / 0.59 / 0.50
2 / 31 / -0.82 / 0.08 / -0.23 / 0.52
Future / 1 / 55 / 0.55 / 0.56 / 0.52 / 0.32
2 / 28 / -0.33 / 0.30 / -0.45 / 0.77
All / 1 / 56 / 0.50 / 0.57 / 0.51 / 0.40
2 / 26 / -0.48 / 0.28 / -0.42 / 0.72
All PT / 1 / 54 / 0.47 / 0.59 / 0.48 / 0.44
2 / 26 / -0.50 / 0.27 / -0.47 / 0.68
All ET / 1 / 67 / 0.49 / 0.51 / 0.46 / 0.53
2 / 15 / 0.32 / 0.51 / -0.79 / 0.11
Table S2 Results of a principal component analysis with the obtained correlations between the different forcings and the pressure drop during the intensification period (max 48h). The first 2 EOF’s (out of 4) are given with their explained variance and components. The first EOF contains the typical re-intensification after ETT where all parameters are positively correlated while the second suggests hybrid properties. The latter is negatively correlated with IVT and Qc as systems have more potential to re-intensify when they complete ETT and turn into deep cold core systems. When doing the analysis for all storms, it appears that this second EOF is not present for the extratropical systems as only post-tropical cyclones can have a residual warm core and exhibit hybrid characteristics. The second EOF is most dominant in results for the near future, in agreement with the low or even negative correlations seen in table 3.
Fig. S1 Hart diagram of Amy, calculated at 3-hourly intervals with a 12h running mean applied. Colours indicate intensity in terms of minimum MSLP and marker sizes represent the mean radius of 10m gale force (17.2m/s) wind. Time relative to t0 is given in days. Amy starts as a tropical cyclone with a deep symmetric warm core and starts ETT 3 days before t0. It then goes through a long hybrid phase until it finally becomes a deep cold core cyclone that re-intensifies into a shallow warm core system.
Fig. S2 Total precipitation during the life cycle of Amy. During its tropical and hybrid phase, the small warm core leaves a tight path of intense precipitation. Amounts exceeding 300mm can pose a serious threat to nearby islands such as the Azores. Across the British Isles, the cyclone merges with an extratropical depression and completes its ETT while dropping at least 50mm over a large part of the area and in excess of 150mm in Scotland. Finally, the storm redevelops a warm core resulting in large precipitation amounts in southern Scandinavia.
Fig. S3 Tracks of all storms impacting Western Europe for the present (a,b), near future (c,d) and future (e,f) on the left and those being selected as having a tropical origin on the right. Colours indicate the time relative to the point at which the wind criterion is met and the grey boxes illustrate the regions used to create the according sets of storms.
Fig. S4 Tracks of tropical cyclones reaching Western Europe using the same regions as in (Fig. S3a) from the NOAA IBTrACS database. Coloured tracks show all the selected cyclones during the 1981-2010 period while black lines indicate all of the selected storms during 1842-2011. Colours indicate storm intensity in terms of maximum sustained 10m wind speed. Although this dataset is not appropriate to study a possible re-intensification, the number of storms and their paths agree well with the results of the model (Fig. 4b).
Fig. S5 Frequency (normalised) of at least Beaufort 10 (24.5m/s) 3 hourly 10m wind during Autumn (August-November). Shading represents the difference between future and present simulations while contours indicate the present distribution. The expansion of the Hurricane Basin can be observed along with an eastward shift of the TC genesis region. Especially the area influenced by warm waters of the Gulf Stream sees an increase in severe storms. The region of increased storminess extends all the way to Western Europe, suggesting a link with the tropics.
Fig. S6 Mean 850hPa EPT (a) and PT (c) field for the simulated present climate (shading) and the future change (contours) in K on the left side. The panels on the right side show the zonally averaged 850hPa EPT (b) and PT (d) across 80°W - 40°W, where red represents the future, purple the near future and black the present period. In addition, the zonally averaged meridional gradient is given for the future (orange), near future (pink) and present (blue). Both the EPT and PT show a sharp gradient near (40°N, 60°W) which is the region where many ETCs form and TCs undergo their ETT. There is a large increase in EPT just south of this area due to the Gulf Stream and increased storminess carrying more warm and moist air. There is also a region of high EPT increase related to the eastward expansion of the hurricane genesis region. In general, there is a decrease of the overall temperature gradient between the tropics and the poles. However, the EPT increase is stronger in the tropics due to the moistening of the atmosphere. Unique to the North Atlantic is the predicted weakening of the meridional overturning circulation which will result in a reduced warming in the north-east. This results in a considerable strengthening of the EPT gradient and the absence of a weakening for the PT gradient. The tightening EPT gradient in the near future around 50°N (b) can be observed while a further increase only happens to the south in the future due to further warming of the Tropical Atlantic. Similarly, a small increase in PT gradient is seen for the near future while none is apparent in the future. Together, these changes favour enhanced cyclogenesis for both tropical and extratropical systems during Autumn in the future North Atlantic.
Fig. S7 Mean Hart diagram for all present (a,c) and future (b,d) post-tropical cyclones using the same legends as Fig. S1. In the future, storms are more intense during their tropical phase indicated by a stronger warm core and a lower central pressure. Both periods nicely show the development of a deep cold core during ETT and a consecutive re-intensification with the formation of a shallow warm core. The latter is also stronger and more symmetric for future storms (B is closer to 0). In general, the entire phase space tends to shift towards a warmer stage although the deep cold core does not become weaker. As a result, storms spend more time in the hybrid phase (right top of upper panels) in the future; from 1.5 to 2.5 days. This could lead to an increased number of storms resembling Sandy of 2012, capable of reaching Europe as well.
Fig. S8 Same as Fig. S7 but for extratropical cyclones. In this case, the development of a deep warm core is completely absent, confirming the strongly different life cycle in comparison with post-tropical systems. Also in this case, future storms tend to get more intense, this is supported by the trends in EPT and PT gradients shown in Fig. S6.
Fig. S9 Average 850hPa EPT field of selected storms from the MERRA reanalysis dataset over the 1981-2010 period using minimum MSLP for time selection. The storm is located at the left exit region of a jet streak, exhibits a warm seclusion and a secondary wind maximum south-east of the centre in agreement with Fig. 6a.