Characteristics and development of European cyclones with tropical origin in reanalysis data

2.1 Model specifications

For this study the MERRA reanalysis of the Global Modeling and Assimilation Office (GMAO) has been used, which is conducted with version 5.2.0 of the GEOS-5 ADAS. The resolution of the MERRA data set is 0.66\(^\circ\) in longitude and 0.5\(^\circ\) in latitude and it has 72 vertical layers. For the analysis the vertical resolution has been interpolated to an equidistant resolution of 17 layers of 50 hPa from 1000 to 200 hPa, similar to that used in Hart (2003). The main arguments for choosing the MERRA data set are its relatively high horizontal resolution compared to other reanalyses, and the availability of a few fields, such as mean sea level pressure and 850 hPa temperature on an hourly interval (other reanalyses usually have 3-hourly or 6-hourly intervals). High spatial and temporal resolution are important for a reliable analysis and tracking of the cyclones. We realize that reanalysis data is still model dependent and that the horizontal resolution is much coarser than the observation database by design.

Schenkel and Hart (2012) have compared the representation of tropical cyclones in five different reanalysis data sets. In the North Atlantic, which is our focus area, the quality of the ERA-Interim, ERA-40 and MERRA reanalysis data sets appeared to be comparable due to the dense observations. The two reanalysis data sets that are significantly better in terms of tracks and intensity, the Climate Forecast System Reanalysis (CFSR) and the Japanese 25-year Reanalysis (JRA-25) use vortex relocation that can result in a nonphysical life cycle and are therefore not useful for our analysis.

2.2 Cyclone detection and tracking

The analysis period is 1979–2013 for the months of August through November, which is shifted by one month with respect to the main hurricane season (July–October). This is motivated by the enhanced baroclinic instability later in the season, which is important for extratropical transition. The same months were also analyzed by Baatsen et al. (2015). Cyclone tracking is done for the region 100\(^\circ\)W–40\(^\circ\)E and 0\(^\circ\)N–90\(^\circ\)N. Around local 10 m wind maxima (found using the threshold of \(\ge\)14 m s\(^{-1}\)), the lowest sea level pressure within a 10\(^\circ\) range is defined as a cyclone center. The threshold of \(\ge\)14 m s\(^{-1}\) is intentionally chosen relatively low to prevent gaps in the tracking of cyclone. The 10\(^\circ\) range around the wind maximum prevents a single storm’s location to be counted twice.

2.3 Phase space analysis

In addition to these three parameters the cyclone size is computed, which is defined by the mean distance of the gale force wind field edge (16.9–18.5 m s\(^{-1}\)) towards the center of the cyclone.

2.4 Baroclinic instability

3.1 Symmetric warm-core development: tropical cyclone life cycle

Cyclones with tropical characteristics can be characterized by a thermally symmetric or non-frontal structure (B <10 m) and a warm core (\(\mathrm {T}_L\) >0, \(\mathrm {T}_U\) >0). Cyclones of the tropical life cycle start with these characteristics. The lower-troposphere warm core increases upwards due to sustained convection (Hart 2003). Combined with subsidence within the eye, a warm core is established, visible at the start of Fig. 1a by a high \(\mathrm {T}_L\) and a slightly positive \(\mathrm {T}_U\) . They generally lose this warm-core structure when they encounter colder SSTs on their way to Europe. The fact that these cyclones do not develop thermal asymmetry (B remains <10 m, Fig. 2a) can be explained by the fact that they are not subject to strong vertical shear, trough interaction or baroclinicity. In the early phase of this life cycle, the cyclone’s spatial extent in wind field is usually small (200–400 km), and steadily grows in time. The pressure minimum is in some cases found early in the life cycle during tropical intensification, but in other cases it is found near the point where the cyclone, after attaining a deep cold-core structure that has been developed over cold SSTs, re-establishes a shallow warm core. The cold core development in this life cycle (Fig. 2a) is different from other literature (e.g. Hart 2003). We speculate that this caused by the fact that we only investigate cyclones with this life cycle that cross the North Atlantic and reach Europe, whereas in previous literature this category contained mostly cyclones confined to the tropics.

We find that about 15\(\%\) of the observed cyclones follow this life cycle. These cyclones tend to enter Europe in France and Great Britain and never attain latitudes higher than 60\(^\circ\)N. The average wind speed maximum and average pressure minimum of this class are 21.9 ± 1.7 m s\(^{-1}\) and 977 ± 9 hPa, respectively, which makes this the weakest class of low-latitude originating storms entering Europe (see Table 1).

Open image in new window

Open image in new window

3.2 Asymmetric cold-core development: extratropical cyclone life cycle

The cyclones pertaining to the extra-tropical life cycle originate with a cold-core structure (\(\mathrm {T}_L\) < 0, \(\mathrm {T}_U\) < 0) (Fig. 1b) and retain it throughout their development in combination with a strong thermal asymmetry (B > 10) (Fig 2b). Although coming from below 37.5 N, these are the characteristics of an extratropical cyclone. Often visible in these cyclone life cycles is an increase of the cold-core structure in an early phase, which is reflected by \(\mathrm {T}_L\) becoming increasingly negative. The middle- and upper-tropospheric height gradients above the surface cyclone then intensify (isobaric heights decrease) more rapidly than near the surface, leading to an increasing cold-core cyclone signature. With B increasing, thermally direct circulation dictates that cold air is advected in the rear of the cyclone and warm air is advected further north, sustaining the strongly negative \(\mathrm {T}_L.\) This is often followed by an increase of \(\mathrm {T}_L\) as can be seen in Figs. 1b and 2b, where also an increase in pressure is clearly visible. For detailed analyses of this life cycle, see Hart (2003). Occlusion (following the Norwegian cyclone model) of the cyclone fronts can be recognized by re-establishing thermal symmetry in the latter phase of the cyclone life cycle. If the cyclone does not interact with another trough or is not subject to increased surface fluxes, further intensification stops and the cyclone decays. The radius of the system (mean gale force wind radius) is relatively large throughout the extratropical cyclone life cycle (Fig. 1b).

About 13\(\%\) of analyzed cyclones follow an extratropical life cycle. These cyclones tend to penetrate further north than cyclones with a tropical life cycle, reaching Europe at the northernmost point of Great Britain or even near Iceland. The average wind speed maximum and average pressure minimum of this class are 21.5 ± 2.3 m s\(^{-1}\) and 970 ± 13 hPa, respectively, which makes these storms of moderate strength among the low-latitude originating storms entering Europe (see Table 1).

3.3 Extratropical transition: classic ETT cyclone life cycle

The first two classes mentioned above (tropical and extratropical cyclone life cycles) are called conventional or single phase life cycles as they do not undergo major phase transitions in thermal symmetry and thermal wind (Hart 2003). The remaining two classes do undergo major phase transitions and differ in the experienced forcing mechanisms throughout their life cycle.

One of these major phase transitions is extratropical transition (ETT), changing the warm-core symmetric structure of a tropical cyclone into the cold-core asymmetric structure of an extratropical cyclone (Jones et al. 2003). The tropical origin is visible in Fig. 2c, with near-zero or even negative values of B and positive values of \(\mathrm {T}_L.\) This is called the tropical stage. One would expect to find positive values of \(\mathrm {T}_U\), too, which is on average not the case (Fig. 1c). This depends on the maturity of the tropical cyclone. Some do start with a positive \(\mathrm {T}_U\), while others do not as can be seen from the individual life cycles in Fig. 1c. In this tropical stage, the cyclone radius is relatively small but slowly increasing, similar to the early phase of the tropical (conventional) life cycle. We adopt the criterion Hart (2003) proposed for the onset time of the ETT, when B exceeds 10 m, which corresponds to the cyclone entering a baroclinic environment (Hart 2003; Klein et al. 2000).

After the ETT, in the second transformation or hybrid stage, the cyclone interacts with its new environment, which usually consists of merging with an extratropical system or upper-level trough. During this stage, the tropical cyclone generally develops an increased translation speed. In the early stages of ETT, the cyclone tends to weaken first (Hart and Evans 2001), which could be attributed to the interaction between the cyclone and an upper-level trough, as this is associated with high vertical wind shear. The decrease in intensity of the cyclone also depends on the inner-core convection evolution by the environmental changes (Jones et al. 2003). Important changes after an ETT are the loss of organized convection in the inner core, the increase in translation speed, the loss of upper-level outflow circulation, increased frontogenesis, cyclone vertical tilt, asymmetry in the precipitation, moisture and temperature fields and the expansion of the gale force winds area (Hart and Evans 2001). Throughout the ETT, some cyclones directly attain a lower cold core and a thermally asymmetric structure (B >10 m; Fig 2c). However, ETT as it is described in Hart (2003), implies first a transition to the upper-right quadrant in Fig. 2c. The individual cyclone life cycles depicted in grey reveal that indeed many of them undergo this transition. This quadrant refers to the so-called hybrid stage, where a lower warm core (\(\mathrm {T}_L\) > 0) pertains in a baroclinic environment (B >10 m). The latent heat release possible in the warm-core structure of the cyclone during this phase, may contribute to a deepening. The end time of the hybrid stage (\(\mathrm {T}_L\) becomes negative) is sometimes referred to as the end of the ETT (Hart and Evans 2001). The duration of the hybrid stage differs among cyclones. The extratropical stage contains the decay and sometimes a reintensification of the cyclone. Both weak and strong cyclones can reintensify, but weak cyclones need to have a smaller duration of the hybrid stage for them to survive longer. The pressure evolution is therefore determined by weakening in an early phase of the ETT, the intensification during the hybrid stage (of cyclones that have one), and possible reintensification after the ETT (see Fig. 2c). Cyclones of this life cycle fade as cold-core, thermally asymmetric cyclones.

About 19\(\%\) of the observed cyclones follow this life cycle. The tracks of these cyclones start relatively deep in the tropics. They usually arrive in Europe around Great Britain. The average wind speed maximum and average pressure minimum of cyclones of this class are 22.5 ± 1.7 m s\(^{-1}\) and 972 ± 7 hPa, respectively, which is relatively weak (Table 1).

3.4 Transition to warm seclusion: warm seclusion life cycle

Finally, more than 50\(\%\) of the storms undergo another major transition and become what is known as a warm seclusion storm. A warm seclusion occurs, when during the bending of the cold and warm front, part of the cold front gets separated from the cyclone center and propagates eastwards, perpendicular to the warm front (Shapiro and Keyser 1990). This process is called a frontal T-bone fracture. This is associated with trapping of warm air (the so-called bent-back warm front) in the core of the cyclone. A large part of the analyzed cyclones that attain a warm seclusion structure start with a warm \(\mathrm {T}_L\) and a symmetric structure as seen in Figs. 1d and 2d. In their path towards Europe they lose their warm core structure and become asymmetric. During this transition they weaken. What distinguishes this life cycle from the classic ETT life cycle is that after this transition, due to the seclusion, a warm core is re-established again. This is caused by the release of latent heat and advection within the warm conveyor belt. The warm seclusion is often accompanied by a rapid intensification because both baroclinic processes and latent heat release contribute to the development of the storm. The average deepening is about 30 hPa in 2 days. The minimum pressure is attained when the warm seclusion is completed. Thereafter the cyclone tilt disappears and they lose their warm core before they fade away. This life cycle is in agreement with the analysis of Maue (2010). It is interesting to note that sting jets are mainly found in this type of cyclones (Browning, 2004). This structure development is dominant for the storms that reach Europe. Moreover, the warm seclusion life cycle class contains the strongest storms that originate in the tropics. Eight out of the ten strongest storms are all warm seclusion storms. Also, on average they are the strongest storms with an average wind speed maximum of 22.5 ± 2.4 m s\(^{-1}\) and average pressure minimum of 963 ± 14 hPa.

3.5 Summary

We have divided the cyclones that reach Europe and originate from lower latitudes in four different classes based on the characteristic pathways in the Hart diagrams. Each of these classes describes a physically different life cycle. The life cycles of the individual cyclones shown in Figs. 1 and 2, support this classification. The four possible life cycles are schematically represented in Fig. 3, and correspond to the physical processes and transitions that cyclones may experience when they enter the midlatitudes with colder SSTs. This schematic picture aligns with previous studies (Hart 2003; Hart et al. 2006).

The minority of the cyclones experience small structure transitions (tropical and extratropical life cycle, 28\(\%\) in total), but the majority (72\(\%\)) undergoes major transitions in cyclone structure like extratropical transition and warm seclusion. Intensification of these cyclones mainly appears in the form of a reintensification at midlatitudes while warm seclusion intensification generates the lowest pressure values.

Open image in new window

4.1 Cyclone structure

The cyclone structure is analyzed at the pressure minimum point in time because it is both a physical well-founded point in the life cycle to compare different cyclones and because the warm secluding process is usually just prior to the pressure minimum, which will allow for the warm seclusion features to be visible at this point. Figure 4 shows the trajectories of the warm seclusion cyclones together with the location of the minimum pressure. The figure bears large similarity with Fig. 5.6 of Maue (2010). It reveals that the minimum pressure is mostly attained during the later phase of their life cycle, and often within the European domain. The composite tracks of the different life cycles (lower right panel Fig. 4) reveal that the warm seclusion storms have longest tracks, starting in the deep tropics and penetrating far into Northern Europe. An elaboration of the distinction between two different warm seclusion life cycles made in this panel can be found in the next section.

Figure 5 shows the composite structures of the four cyclone life cycles at the time when central sea level pressure is at a minimum. A first observation is the overall lower \(\theta _E\) in the warm seclusion panel with respect to that in the tropical life cycle and ETT panels, which is mainly because of the relatively late stage and more northward location at which these cyclones intensify. Comparing the \(\theta _E\) structure of these cyclones further with that of other cyclone life cycles, the main characteristics are the consistency of the figure, the curled warm conveyor belt (WCB), the eastward progressing cold front and the vertically stacked wind pattern, indicated here by the colocation of the maximum windspeed at 10 m, 850 hPa and, to a lesser extent, 250 hPa. The smoothness of Fig. 5d points to the consistency in the structure of warm seclusions at their highest intensity, which is not the case for the other life cycles. Additionally, the curled WCB displaying a well-defined comma shape is also quite unique to the warm seclusion life cycle. The eastward progressing cold front in Fig. 5d is stronger than in any other cyclone life cycle pressure minimum. In analogy with the curled WCB, the cold front progresses far east, which is characteristic for the T-bone fracture. The intensity of the cyclone at this point in time, the far propagated low-\(\theta _E\) region and vertically homogeneous eastward winds south of the cyclone may point to the presence of a dry intrusion, which favors the occurrence of a sting jet.

Open image in new window

4.2 Development of \(\theta _E\), p and \(\sigma _m\)

The variables \(\theta _E\), sea level pressure and the moist Eady growth rate \(\sigma _{m}\) at the cyclone core provide information about the development of the cyclone and the specific phases the cyclone is in. High values of \(\theta _E\) allow for stronger latent heat driven deepening as occurs in tropical cyclones, while \(\sigma _{m}\) is a measure of baroclinic instability, which is the major energy source for extratropical cyclones.

More than half of the warm seclusion storms, although originating south of 37.5\(^\circ\)N, do not start with a tropical structure that is characterized by a warm core and a non-frontal structure, but already possess extratropical characteristics. Although they are almost indistinguishable in their final warm seclusion structure, the different origin has implications for their ultimate pressure minimum and the development of \(\theta _E\) and \(\sigma _{m}\). We therefore divided the warm seclusion life cycle in two sub-cycles: warm seclusions that originated as tropical cyclones, i.e. following ETT, called tropical warm seclusions, and warm seclusions that originated as extratropical cyclones, i.e. without ETT, called extratropical warm seclusions.

Figure 6a shows the \(\theta _E\) development of the cyclone life cycles. Cyclones that originated as tropical cyclones start with higher values of \(\theta _E\) than their extratropical originating counterparts. The tropical warm seclusions lose rather much of their \(\theta _E\) when they enter Europe. However, together with the extratropical warm seclusions they survive longest in Europe, up to more than 5 days. The large reduction of \(\theta _E\) of the tropical warm seclusions could be explained by the fact that they travel relatively far northwards.

The sea level pressure development for each cyclone life cycle is shown in Fig. 6b. It reveals that both warm seclusion life cycles have their pressure minimum latest of all, about a day after entering Europe, while other cyclone life cycles have this minimum close to the time they enter Europe. Striking is the strong deepening rate of the warm seclusions, both starting a day prior to entering Europe. The tropical warm seclusion storms attain the lowest pressure with an average value of about 960 hPa. Notable are the long time periods of the tropical warm seclusion, extratropical transition and tropical life cycles before entering the midlatitudes (indicated by small vertical colored bars) in contrast to the extratropical and extratropical warm seclusion life cycles. This confirms the rather different origin of air mass type these cyclones have, although they all come from latitudes lower than 37.5\(^\circ\)N. In Fig. 6b, d, for the tropical warm seclusions, a slight minimum in pressure and maximum in wind speed can be distinguished roughly 6 days prior to entering Europe. This may refer to the tropical intensification of these cyclones prior to entering midlatitudes, but the signal is diffused due to the averaging of the individual evolutions. The midlatitude intensification and the loss of \(\theta _E\) in tropical warm seclusions occur roughly at the same time, perhaps both influenced by ETT, which increases cyclone speed and therefore the rate at which its environment becomes colder. This may also enhance the intensification.

Figure 6c shows the moist Eady growth parameter \(\sigma _{m}\). Being a measure of baroclinic instability, an increase of it after the entering of midlatitudes is expected. Remarkable is the increase in \(\sigma _{m}\) of the tropical warm seclusions around their arrival in Europe, reaching the highest values during their lowest core sea level pressure. This indicates that moist baroclinic instability increases during ETT and the forming of the warm seclusion for this cyclone life cycle. A similar behavior is observed for the extratropical warm seclusions. Having these maxima in \(\sigma _m\) near their moments of highest intensity, makes the warm seclusions rather unique with respect to other cyclone life cycles, which do not have such maxima. It points toward the important role of moist baroclinic instability for the warm seclusions during their reintensification and being responsible for their relative strength.