High latitude severe storms have been increasingly subjected to scrutiny as they are an important indicator of changes in the global climate and teleconnections with the tropics (Eichler and Higgins, 2006). These intense cyclones are influenced by changes in robust modes of circulation (e.g. the annular modes) (Hall and Visbeck, 2002), changes in sea ice (Stroeve et al., 2007; Comiso and Nishio, 2008; Simmonds et al., 2008), meridional heat and momentum transfers (Uotila et al., 2009), and precipitation anomalies (Hurrell, 1995; Serreze et al., 1997; Renwick, 2002). Under realistic scenarios of climate prediction, the baroclinic zones of both the hemispheres shift poleward, whereas mid-latitude blocking-like ridges become more prominent (Herweijer and Seager, 2008). As a result of this synergy, many countries may receive less rain, whereas others are expected to experience more frequent cyclone development, especially in high latitudes in the Northern Hemisphere (NH), posing a threat to life and property.
Extratropical cyclones have been the subject of studies using the Lorenz energetics in the past with emphasis on understanding the importance of heat and momentum transfers in mesoscale simulations (Smith, 1980; Robertson and Smith, 1983; Michaelides, 1992; Wahab et al., 2002; Marquet, 2003; Decker and Martin, 2005). Recently, the technique was used to study the tropical transition of the first known South Atlantic hurricane (Pezza and Simmonds, 2005) from a large-scale energy-transfer perspective (Veiga et al., 2008). This was the first time that a dynamic transition (from cold to warm core) of a storm was explained entirely based on environmental energetics, opening a new avenue of exploration of extreme weather events and their connection with climate. These recent results demonstrated that the baroclinic and barotropic conversions occurring in the large-scale environment had a fundamental role in establishing the right conditions for the transition (Veiga et al., 2008). The energy transfer is believed to occur indirectly from the environment to the vortex so that once the vortex has been initiated local processes (e.g. thermodynamics) may become more important, but we note that the eddies can also provide energy to the background flow. In the subtropics, the presence of a persistent blocking can facilitate a greater rate of growth when the thermodynamics over the ocean is favorable, but a very little is known of how these interactions come about at high latitudes.
We advance further on this topic here and explore the energetics of a very intense high latitude NH system observed in October 1992. Mesquita et al. (2009) provided a preliminary documentation of this storm, showing that it initially formed in eastern China and exhibited its greatest rate of intensification as it approached the Bering Sea. Although only light rain and snow were recorded in Nome, the storm caused unprecedented flooding as a result of the optimum wind fetch. The storm had minimum pressure of 962 hPa on 6 October with near hurricane-force winds causing over $ 6 million dollars in damage (Mesquita et al., 2009). This system is of significant interest also because of its marked latitudinal displacement of about 30° in 7 days, following a near-constant rate of latitude increase facilitated by a blocking system over the northwest Pacific. As blocking frequency and location are expected to experience significant shifts under global warming scenarios (IPCC, 2007), there is a strong potential for significant changes in the behavior of high-latitude storms. This paper is a first step in elucidating how well the Lorenz energetics can portray fundamental changes in the life cycle of severe storms, opening new avenues for climate change studies of storm behavior. Data and methodology are discussed in Section 2 and the main results are presented in Section 3. A follow-up discussion is presented in Section 4, commenting on new insights and perspectives brought by the results.
2. Data and methodology
A traditional and compact form of presenting energy computations is suggested by Lorenz (1967), where the available potential energy can be partitioned into zonal available potential energy (AZ) and eddy available potential energy (AE). By analogy, the kinetic energy can also be partitioned into zonal (KZ) and eddy (KE) components. In the energy cycle of Lorenz, AZ and AE are produced by generation processes namely, GZ and GE (generation of AZ and AE), whereas energy terms are connected by conversion terms. Following Muench (1965), the energy budget equations for an atmospheric region with open-boundary conditions can be given as:
In an open area, the budget is complicated by the inclusion of nonzero boundary transports. To account for these, we include four new components representing the transport of AZ, AE, KZ, and KE (BAZ, BAE, BKZ, and BKE, respectively). In addition, KZ and KE are associated with the work produced at the boundaries (BϕZ and BϕE, respectively). The last terms in Equations (1a) and (1b) namely, DZ and DE, represent dissipation processes. Those terms are here combined and computed in the form of residuals (Wahab et al., 2002). In a similar manner, the intrinsic errors resulting from the numerical estimations for the remaining terms in Equations (1c) and (1d) are also included as residuals (see Veiga et al., 2008 for further details).
Six-hourly data from the NCEP/NCAR reanalysis (2.5°× 2.5° resolution (Kalnay et al., 1996) were used for the gridded energy calculations, cyclone trajectory, and large-scale atmospheric circulation. Anomalies were calculated over the NCEP 30-year reference period of 1968–1996 (Kalnay et al., 1996) based on weekly and daily climatologies. Horizontal integrations were performed over the areas bounded by 120°E and 150°W, 50°N and 80°N (‘Nome box’ representing the environment), and 120°E/150°W, 30°N/50°N (‘Blocking box’ representing the mid-latitude block). These domains are large enough to portray the high-latitude environment including the Pacific blocking that ‘steered’ the system's trajectory (Figure 1(a)). These domains form one of the most active high-latitude storm genesis regions on the planet, being of immense importance for infrastructure and fisheries development along the countries involved (for instance, coastal infrastructure in Japan and salmon in Alaska). Sensitivity tests showed that our results hold if larger areas are used, and that the separation between blocking and the poleward environment is necessary to accurately quantify the associated energy transfers. For simplicity, key periods for the analysis of this storm were defined: week − 1 as 20–30 September 1992, week 0 as 30 September–10 October 1992, and week + 1 as 10–20 October 1992.
3.1. Blocking and large-scale circulation
Figure 1(a) shows the mean sea level pressure (MSLP) anomaly for week − 1. The MSLP anomaly is indicated in shading for areas with magnitude above 2 hPa. A number of other features are also displayed, viz., the 850–200 hPa wind shear vector anomaly for week − 1 and the 850–500 hPa thickness anomaly (m) averaged for 6 October 1992 (only magnitude greater than 200 m shown). The trajectory as identified by the Melbourne University automatic tracking scheme (Simmonds et al., 2003) has also been superposed to the map, with some key phases indicated, viz., the beginning and ending points and the point of maximum intensity. The location of the maximum intensity (marked with an ‘x’) given by the tracking scheme for the NCEP reanalysis was 968 hPa on 6 October at 1800 UTC (Figure 1(a)), being close to the minimum pressure estimated by the Anchorage analysis chart (962 hPa at 1200 UTC). The Nome and Blocking boxes are also indicated. This figure gives the large-scale panorama from which the storm was formed. The MSLP during week − 1 formed a split blocking pattern of which the polar component resembled very closely the trajectory followed by the storm a week later. The Pacific High over mid-latitudes can also be seen at about 37°N near the dateline. This high is related to the subtropical ridge that is normally found over that area, but it was found to be considerably stronger and further north than normal during week − 1.
The largest positive MSLP anomaly is seen on the poleward side near the Bering Strait. Not only did this area experience intense blocking during week − 1 but also it had the largest wind shear anomaly to the south, which indirectly contributed to steering the storm motion. The thermal wind implication is that the environment was highly baroclinic at mid-levels (the role of environmental baroclinicity is discussed in detail in Section 3.2). It can be seen that over most of the storm's trajectory the wind shear over week − 1 closely follows its path, showing that the storm followed the maximum baroclinicity. The period of maximum intensification (average over 6 October) is given by mid-level thickness anomaly (green contours). This shows a strong response over the storm, corresponding to an environmental warm anomaly of about 5° at 850 hPa.
Figure 1(b) shows the MSLP changes from week − 1 to week 0 plotted for magnitude greater than 4 hPa (shading) and the 500 hPa temperature anomaly (magnitude above 3 °C, contour) for week − 1. The fundamental change in the large-scale pressure is that the blocking loses its poleward-split component and intensifies further at mid-latitudes. This comes about with pressure changes, which have little zonal variability, showing that the changes are not simply a pressure adjustment associated with the vortex. The 500 hPa temperature anomalies show that most of the domain was cold during week − 1, including the poleward blocking split discussed in Figure 1(a). The Pacific High to the south was predominantly barotropic and warm-cored (compare Figure 1(a) and (b)). As discussed later, the energy transfers reveal that the blocking projected a strong barotropic mode about five days before the cyclone was formed, and it also provided the dynamic steering that facilitated the wave propagation and baroclinic growth.
Figure 2 shows the time series of the (a) energy and (b and c) conversion terms integrated between 925 and 100 hPa for (a) week 0 and (b) week − 1 to week + 1 for the Nome box, and (c) week − 1 for the Blocking box. The different phases of the storm are: cyclogenesis (L1), rapid intensification (L2), maximum intensity (L3), and decay (L4). It is evident from Figure 2(a) that the environment changed from a state of lower energy to a state of higher energy about 2 days before the cyclone underwent a rapid intensification (between L1 and L2). This transition started with a peak in AZ followed by the peaks in KE, AE, and KZ. The overall energy then rapidly declined from the time the cyclone started to lose its intensity (L3). Toward the decay phase (L4), the environment reached a state in which AZ was reduced by about 50% of its initial value. Below, we show that the fundamental process that allowed the energy transfer above came about through baroclinic conversion.
Figure 2(b) shows that the conversions over a 30-day period centered in the storm's maximum period of intensification are maximized for the baroclinic terms CE and CA. The maximum conversion took place on 6 October 0000 UTC, about 18 h before the cyclone reached maximum intensity. As discussed earlier, the large-scale domain used for the calculations allows for little direct influence from the vortex in the energy conversion terms, similar to what was noted by Veiga et al. (2008). However, this is not to say that there is no interaction between the environment and the vortex. In fact, our results suggest that the environment played an important role in determining conditions suitable for the vortex to grow. In this sense, there is a suggestion that the vortex grew at the expense of the environment primarily through baroclinic processes. The peak in baroclinic energy transfer was maximized between 350 and 600 hPa (not shown). This is the ideal elevation for baroclinic growth at high latitudes partially because the static stability tends to be low. As discussed earlier, not only the blocking further south created ideal conditions for the steering of the storm, but also it participated in the energy exchange as shown in Figure 2(c). It is evident that a maximum of barotropic conversion occurred on 25 September at 1800 UTC, 5 days before the cyclone was first identified by the tracking scheme [indicated as preliminary trough (PR)]. Notwithstanding the marked barotropic peak in the Blocking box (about 20 W/m2), Figure 2(c) also shows that the other conversion terms were strong during week − 1 (magnitude greater than 9 W/m2), characterizing an environment favorable for storm development.
The vertically integrated time–mean energy cycle is illustrated by the Lorenz diagram in Figure 3. Figure 3(a) shows the schematic of the Lorenz diagram adopted here, where the arrows indicate the direction of the energy flux (see Section 2 for further details). Figure 3(b) and (c) shows the energy interplays over week − 1 and week 0 for the Nome box, and Figure 3(d) shows the energetics for the Blocking box over week − 1. These results account for an integrated view of what was discussed before, showing that the environment presented strong baroclinic conversions from week − 1, intensifying over week 0 (CE reaches 4.1 W/m2), whereas the barotropic conversion CK remains small. However, CK becomes important for the Blocking box (Figure 3(c)), with similar order of magnitude as the other terms. Compared to the results of Veiga et al. (2008), it is apparent that the energetics of the environment can be confidently used to quantify the processes related to extreme synoptic development both at the mid and high latitudes, with the blocking being a crucial component of the meridional (north–south) exchanges.
We also noted that the generation and boundary fluxes terms are also indicated in the flux diagram. The direct calculation of the boundary pressure work terms poses significant difficulties, particularly due to the magnification of initial errors in the data (Wahab et al., 2002 and references therein). Those terms are here combined with the dissipation terms and estimated as residuals (terms RKZ and RKE indicated in Figure 3(a)). In a similar manner, the intrinsic errors resulting from the numerical estimations of the other terms in Equations (1c) and (1d) are also calculated using the same principle (terms RGZ and RGE indicated in Figure 3(a)). Figure 3 also reveals that the generation and boundary fluxes are generally smaller than the greatest energy conversion term in the Lorenz cycle in all the cases except in Figure 3(b), where RGZ dominates. This suggests that the energy transfers are predominantly above the noise.
4. Concluding remarks
In this paper, we analyzed the Lorenz energetics of an extreme high-latitude storm calculated over a large-scale domain representative of the environment. We found that while mid-latitude blocking played a fundamental role in the preliminary phases it was the baroclinic energy transfers in the environment that can explain the intense development over the life cycle. Although such an approach does not allow us to directly associate the energy conversions with processes at the scale of the vortex, it conveys key information as to environmental forcing mechanisms, quantitatively demonstrating the environment's participation, and the synergy between the blocking, which gave the necessary steering, and the large-scale baroclinic growth. Although we do not discuss mesoscale details, eddy momentum and heat flux convergence are a likely process in which energy can be transported from the environment into the vortex (Molinari and Vollaro, 1989).
Our results shed new light on two principal fronts: first, the environmental energy transfers can be used as an informative metric for severe storms if calculated over a sufficiently large area; second, the barotropic transfer is important as a precursor, whereas the baroclinic transfers occurring poleward can explain much of the intensification. These results open new avenues in climate change studies of severe baroclinic storms in that the combined energetics of the blocking and the larger environment have the potential to give an independent measure of storm behavior. This develops further on the recent work of Veiga et al. (2008), suggesting that the Lorenz energetics can be used as a quantitative index of storm activity over high latitudes. In particular, our findings highlight the importance of blocking for downstream cyclone development, suggesting that through the energetics of blocking it is possible to quantify key sensitive information about storm behavior. With high-latitude storm occurrence being one of the crucial areas to be affected by climate change, it is fundamental that the changes in blocking can also be assessed. Although present state-of-the-art climate models do not have enough resolution to resolve all the processes at the vortex level, our results strongly suggest that the Lorenz energetics can be used with very good confidence as a proxy for storm activity. This poses the fundamental advantage that the large-scale approach intrinsic to this methodology allows it to be easily used with climate model output data, with the potential to improve the predictability of storm behavior under climate change scenarios. There is also a promise that the Lorenz energetics could be used in climatological studies of variability of global storm behavior, identifying the areas most prone for storm development based on climatological thresholds of energy conversions.
A. B. P. and I. S. would like to thank the Australian Research Council and the Australian Antarctic Science Advisory Committee for funding parts of this work. We are indebted to Petteri Uotila for thoroughly reading an earlier version of this manuscript. We also thank three anonymous reviewers for the additional comments and suggestions provided.