TRENDS AND DIRECTIONS IN CLIMATE RESEARCH

Identification and Climatology of Cut-off Lows near the Tropopause R. Nieto,a,b M. Sprenger,c H. Wernli,d R. M. Trigo,b,e and L. Gimenoa aUniversidad de Vigo, Facultad de Ciencias, Ourense, Spain bUniversity of Lisbon, CGUL-IDL, Lisbon, Portugal cInstitute for Atmospheric and Climate Science, ETH Zurich,¨ Zurich,¨ Switzerland dInstitute for Atmospheric Physics, University of Mainz, Mainz, Germany eUniversidade Lusofona,´ Departamento de Engenharias, Lisbon, Portugal

Cut-off low pressure systems (COLs) are defined as closed lows in the upper tropo- sphere that have become completely detached from the main westerly current. These slow-moving systems often affect the weather conditions at the earth’s surface and also work as a mechanism of mass transfer between the stratosphere and the troposphere, playing a significant role in the net flow of tropospheric ozone. In the first part of this work we provide a comprehensive summary of results obtained in previous studies of COLs. Following this, we present three long-term climatologies of COLs. The first two climatologies are based on the conceptual model of a COL, using European Centre for Medium-range Weather Forecasts (ECMWF) analyses (1958–2002) and National Centers for Environmental Prediction–National Center for Atmospheric Research (1948–2006) reanalysis data sets. The third climatology uses a different method of detection, which is based on using potential vorticity as the physical parameter of diagnosis. This ap- proach was applied only to the ECMWF reanalysis data. The final part of the paper is devoted to comparing results obtained by these different climatologies in terms of areas of preferential occurrence, life span, and seasonal cycle. Despite some key dif- ferences, the three climatologies agree in terms of the main areas of COL occurrence, namely (1) southwestern Europe, (2) the eastern north Pacific coast, and (3) the north China–Siberian region. However, it is also shown that the detection of these areas of main COL occurrence, as obtained using the potential vorticity approach, depends on the level of isentropic analysis used.

Key words: cut-off low pressure systems (COLs); conceptual model; potential vorticity

Introduction ally starts its life cycle as a trough in the mid- dle and upper troposphere (Fig. 1). Tradition- A cut-off low pressure system (COL) ally, COLs have been recognized as depres- corresponds to a closed low in the up- sions mostly located in mid latitudes, which are per troposphere that has become com- characterized by closed geopotential contours pletely detached (“cut off ”) from the basic in isobaric maps (with a cold core, Fig. 2) that westerly current in the jet stream1,2 and which have more or less concentric isotherms around is usually advected toward the equatorial side the central core.1 Because the jet stream corre- of the mid-latitude westerlies.3 A COL usu- sponds to the boundary between two very dif- ferent air masses, the air mass trapped within a COL maintains its polar characteristics (i.e., Address for correspondence: Raquel Nieto, Edificio de F´ısica, Campus Universitario de Ourense–Universidade de Vigo, As Lagoas s/n, 32004 it is colder than the surrounding air). Its inten- Ourense, Spain. [email protected] sity is highest in the upper troposphere and the Trends and Directions in Climate Research: Ann. N.Y. Acad. Sci. 1146: 256–290 (2008). doi: 10.1196/annals.1446.016 C 2008 New York Academy of Sciences. 256 Nieto et al.: Identification and Climatology of Cut-off Lows near the Tropopause 257

Figure 1. Process of isolation and evolution of a COL. (1) Undulation of the meridional circulation in the jet stream (arrows). Ridges are denoted by R. (2) Southerly north–south elongation. L indicates the COL during its initial stages. (3) Initial isolation or strangulation. (4) Total isolation. The COL has become completely detached from the originating circulation and has developed its own circulation. A ridge or region of high pressures often develops on its northerly part, H in the figure. (5) Start of absorption. (6) South–north elongation and complete absorption magnitude of the associated cyclonic circula- the detachment of a discrete cut-off area. PV tion decreases toward the earth’s surface, being may, therefore, be regarded as another useful even possible to find anticyclonic circulation at quantity for the identification of cut-off lows the surface.4 (Fig. 3). COLs can also be recognized as maxima Irrespective of the method of identi- of potential vorticity (PV) on isentropic sur- fication, COLs can be associated with faces.5,6 It is well known that bands of strong PV intense stratosphere–troposphere exchange gradients coincide with the jet streams and act (STE) (Refs. 8–11 and references therein). In as wave guides for Rossby wave propagation.7 fact, STE associated with COLs has been rec- Rossby wave breaking in the upper troposphere ognized as being responsible, at least to a cer- (near the tropopause) can lead to pronounced tain extent, for the appearance of anomalously excursions of stratospheric air southward and, high values of tropospheric ozone in northern eventually, to the formation of COLs.8 The mid latitudes12,13 and subtropical areas.14,15 In genesis of these systems in mid latitudes can COLs, the tropopause is anomalously low and be seen as the displacement of a region of might be characterized by steep structures and high PV from its polar reservoir, leading to folds. It is proposed that STE in these systems 258 Annals of the New York Academy of Sciences

Figure 2. Image of a COL at 18 UTC on 10 October 2001 in an isobaric map at the 500 hPa level. Solid lines show the geopotential (m) and dashed lines show the temperature (◦C). The isolated COL is over the SW Iberian peninsula. The background image is from the Meteosat infrared (IR) channel. (Figure from the INM, National Meteorology Institute of Spain.) occurs mainly from condensational or radiative ment of a COL is usually slow, and it remains erosion of the tropopause (e.g., Refs. 16, 17). In over a region until it reconnects with the polar addition, turbulent three-dimensional mixing reservoir or until it weakens and eventually dis- near the jet stream associated with the COL sipates. These systems are, therefore, capable and tropopause folding along the system5,18 can of considerably affecting weather conditions at be considered as another possible mechanism the earth’s surface, for periods of several days leading to STE. Separation of air from the gen- at a time (Ref. 2 and references therein). eral flow into a COL is typically associated with As a general rule, the troposphere below the an intense STE,19 and, during the later phase COL is unstable and severe convective events of a COL life cycle, filamentation of the outer can occur, depending on surface conditions. layers20 can be responsible for additional very However, COLs produce significant precipita- effective STE. tion only when the air mass below the COL is Once COLs are disconnected from the envi- significantly moist and becomes potentially un- ronmental zonal flow, the jet stream no longer stable.21 This instability occurs quite frequently controls the motion of the isolated system and below COLs in the southern mid latitudes (i.e., its motion becomes more erratic and difficult near 30–40◦N), and convective events can oc- to predict. The COL may remain stationary cur, sometimes severely, in the affected areas. and spin for days or, on other occasions, it may Because of their erratic movement, their move westward in the opposite direction to the self development and occasionally their dissi- prevailing flow (i.e., retrogression). The move- pative characteristics, and the thermodynamic Nieto et al.: Identification and Climatology of Cut-off Lows near the Tropopause 259

Figure 3. COL detected over the North Atlantic Ocean according to the potential vorticity field (PV; 1 PVU = 1 × 10−6m2s−1Kkg−1) at 300 hPa, based on a HIRLAM model on 30 March 2003 at 12 UTC. Image of the water vapor background for the same time from the Meteosat satellite. A long thin strand links the COL with the PV-rich source or production region from which it became detached. Image from the INM (National Meteorology Institute of Spain).

instability frequently associated with these fea- trophic weather events in terms of precipitation tures, weather forecasting for systems that are rate.22–25 affected by COLs is particularly difficult. COLs The main objectives of this paper are are associated with many substantial forecast- twofold: (1) to provide a review of the COL ing problems, mainly from the different char- concept and its relevance for weather and cli- acteristics of the underlying terrain and to the mate in several regions and (2) to present an presence/absence of a warm ocean that per- update of two, more recent, comprehensive, mits/inhibits convection.21 Thus, prediction of hemispheric climatologies of COLs.6,26 The the distribution of precipitation associated with remainder of the paper is organized as fol- COLs presents a considerable challenge, espe- lows. Section 2 defines the synoptic conceptual cially when the precipitation is from enhanced model of COLs and their physical parameters. convection over a warm sea. When moisture is In Section 3, a review of previous climatologies available, COLs can trigger multiple bands of using subjective and objective methods is pre- clouds, and these can bring moderate to heavy sented. In Section 4, we present updated COL rainfall over large areas. In particular, they are data sets for National Centers for Environ- among the most severe weather systems that mental Prediction–National Center for Atmo- affect southern Europe and northern Africa spheric Research (NCAR–NCEP) and ERA-40 and are responsible for some of the most catas- reanalyses, using the objective method of Nieto 260 Annals of the New York Academy of Sciences et al.,26 and for ERA-40 reanalyses, using the search Action on the Development of Nowcast- method of Wernli and Sprenger.6 In Section 5, ing Techniques.”2 we present a comparison of these climatologies The usual conceptual model of a COL2 de- and our conclusions. scribes this system as a nonfrontal synoptic- scale phenomenon with a typical life cycle that is separated into four stages: upper-level Conceptual Model of a Cut-off Low trough, tear off, cut off, and final stage, as de- scribed below. This section also provides infor- A conceptual model is a visual represen- mation about cloud structure during the differ- tation of the essential features of a meteoro- ent phases of the COL life cycle, the associated logical phenomenon, including the principal horizontal and vertical meteorological charac- processes that are associated with it. A com- teristics, and, finally, weather events associated plete conceptual model provides: (a) a defini- with COLs. tion of the phenomenon in terms of features recognizable by observations, analysis, or val- Meteorological Background idated simulations; (b) a description of its life cycle in terms of its appearance, size, inten- The schematic COL life cycle in terms of its sity, and accompanying weather patterns; (c) a meteorological structure is shown in Figure 4, statement of the controlling physical processes, with Figure 5 showing the evolution of a real which enables an understanding of the factors COL during January 1998 over central Europe. that determine the mode and rate of evolu- These two figures provide an insight into the tion of the phenomenon; (d) a specification four stages of the COL life cycle: of the key meteorological fields that demon- strate the main processes; and (e) guidance for (1) The upper-level trough: An essential con- predicted meteorological conditions or situa- dition for COL development is the exis- tions using the diagnostic and prognostic fields tence of an amplifying synoptic-scale wave that best discriminate between development or in the upper layers of the troposphere. nondevelopment, in addition to guidance that The temperature wave situated behind may be used for predicting displacement and the geopotential wave advects substantial evolution.2 amounts of cold air into the upper-level Conceptual models are useful because they trough and warm air to the ridge. It is provide meteorologists with a tool for both un- also a requirement that the vertical axis of derstanding and diagnosing observed phenom- the trough tilts backward with elevation ena. For both forecasters and modellers, they (Fig. 6). If these conditions occur simul- provide tools for diagnosing the output of nu- taneously, the geopotential and temper- merical models and for identifying errors in ature waves are amplified. Typically, this numerical forecast models.2 amplification is also accompanied by a de- Some details of the meteorological proper- crease of the zonal wavelength. The iso- ties of COLs can be found in the meteorolog- hypses and isotherms show a shift toward ical literature that describes case studies (e.g., the equator, which leads to a deepening of Refs. 27, 28) and in more general studies about the trough. upper-level flow structures (e.g., Refs. 1, 29), but (2) The tear-off stage: In this stage, the shape a complete description of the conceptual model of the geopotential contour assumes an of COLs was developed under the COST ini- inverted omega form. The increase in tiative (European Cooperation in the Field of the amplitude of the waves continues, the Scientific and Technical Research) Action 78 trough deepens, and it starts to detach entitled “Implementation of an European Re- from the meridional stream. The cold air Nieto et al.: Identification and Climatology of Cut-off Lows near the Tropopause 261

Figure 4. Diagram of the typical synoptic situation of a cut-off low showing the different stages of its life cycle using the geopotential field at 200 hPa.

streaming toward the equator is cut off to the mid troposphere (500 hPa) and, from the general polar flow,and the warm in some cases, even to the earth’s sur- air flowing toward the pole is cut off from face. The maintenance of the COL, and the general subtropical flow. At the end therefore its life span, depends on various of this process, a cold-core, upper-level, factors, in particular on thermodynamic low-pressure system is formed within the stability below the COL. southern part of the trough. (4) The final stage: The main factor leading (3) The cut-off stage: This stage begins to the dissolution of a COL is convec- when the tear-off process is finished and tion. If the thermodynamic stability be- the upper-level low becomes more pro- low the COL is sufficiently reduced, con- nounced. The wind field at the up- vection and associated cloud diabatic ef- permost tropospheric levels acquires a fects can lead to a decay of the COL (and well-developed closed circulation, often strong associated STE as described in Sec- completely cut off from the general zonal tion 1) within a few days. This effect is flow. This circulation is strongest near more efficient over warm surfaces (and 200 hPa but typically penetrates down in the presence of a significant vertical 262 Annals of the New York Academy of Sciences

Figure 6. Temperature wave (dashed lines) and potential wave (solid lines). The vertical axis of the trough is denoted by the solid black line. The temper- ature field shows cold advection (CA) in the area of the geopotential trough and warm advection (WA) at the ridge of the wave. (Figures adapted from ZAMG.)

the main zonal flow or it drifts over a warm surface. About 15% of COLs are caught by the main flow.2 A COL typi- cally lasts for about 2–5 days before being destroyed by diabatic heating.5,16

Cloudiness Structure COLs have a very well-defined complex cloud structure and are easily detected in satel- lite imagery on different channels: infrared (IR), Figure 5. Cut-off low system detected over Eu- visible (VIS), and water vapor (WV). The de- rope (indicated with a cross) showing the different gree of cloud cover in COLs depends greatly on stages of its life cycle. Top panel: upper-level trough the stage of its life cycle, but a higher percentage detected on 23 January 1998 at 12 UTC; Middle: of cloud cover is always present in the frontal tear off detected on 24 January 1998 at 00 UTC and zone because of higher concentrations of high Bottom: cut-off stage detected on 24 January 1998 at 12 UTC. Images of the infrared (IR) background clouds and deep convective clouds. The frontal for the same time from the Meteosat satellite showing zone is also characterized as being more active, the cloud cover. Height contours at 500 hPa are con- with cloud-top temperatures of high- and deep- tinuous lines and temperature field (in K) at 500 hPa convective cloud lower than in the zone behind is denoted by dotted lines. (Figures adapted from it.30 Zentrale Anstalt fur¨ Meteorologie und Geodynamik A schematic of each cloudiness stage of the of Austria (ZAMG).) COL life cycle is shown in Figure 7 for the IR temperature gradient), while over cold and WV channels.2 In Figure 8, images from surfaces the COL will not dissolve until the WV Meteosat channel show another real it is either caught by the main stream and COL event. They are described for the four merges with a large upper-level trough in stages of the life cycle2,30: Nieto et al.: Identification and Climatology of Cut-off Lows near the Tropopause 263

Figure 7. Diagram of the typical synoptic situation of a cut-off low showing the different stages of its life cycle using the geopotential field and the satellite images (overlapped dashed areas for IR channel and shaded areas for WV channel). (Figures adapted from ZAMG.)

(1) The upper-level trough stage is charac- east part of the trough). The southern part terized by a well-developed frontal cloud of this cloud band starts to curve cycloni- band. This band appears as white in VIS cally.Insomecases,thiscyclonicprocess and IR images, indicating the high thick- also occurs in the rearward fibrous cloud ness and the multiple layers of clouds. band. The differences in cloudiness be- Some fibrous and rough high cloudiness tween the two zones increase as a result appears behind the upper-level trough,30 of deep convection in the frontal zone. IR which is normally associated with a warm images show white tones from light gray front. This cloudiness is in the upper tro- to white and VIS images show the cloud posphere, so it is possible to detect it only bands in dark gray to gray, restricted to in IR and WV images. These two bands the area of the tear off. WV images show appear as light gray bands of water va- a notable cyclonic band of water vapor por (wet areas) in WV images, and the extended over the entire trough, indicat- core of the COL appears in black because ing the moisture sources. At this stage, the of the dry air that characterizes the area center of the COL becomes dominated by around the trough axis. In the center of dark gray zones, indicating the existence the trough there is also cold air cloudiness of dry (at upper-level stratospheric) air. that can only be detected in IR and VIS (3) In the cut-off stage, a cyclonic spiral of images. Usually about 50% of the COL moisture appears in WV images starting area is covered by some type of cloud, with from the edge of the COL core. This stage a larger percentage in the frontal region. is characterized by a more uniform dis- (2) In the tear-off stage, the cloudiness is tribution of cloudiness, with IR imagery caused predominantly by convergence. showing two white cloud bands ahead of The frontal cloud band becomes broken and to the rear of the trough. At the same where the isolines fold (in the far north- time, convection within the core of the 264 Annals of the New York Academy of Sciences

Figure 8. Cut-off low detected on water vapor images (WV) from Meteosat during the upper-level trough and cut-off stages on 24 January 1998 at 06 UTC and 12 UTC, respectively (Images from ZAMG.)

COL is possible because of reduced strati- (4) In the final stage when the COL life cycle fication beneath the COL. High and deep ends, it is possible to detect in IR images convection clouds appear in the center of that the cloudiness of the COL merges the COL, although with differing inten- with the developed and structured cloud sity and structure depending on whether band (white color) that is associated with the COL is continental or oceanic. Over the main zonal flow. land the convective cells are detected as being embedded in a continuous mantle A general characteristic throughout the en- of lower cloud, while over relatively warm tire life span of a COL is that in the frontal ocean waters the core convection is very zone the percentage of very bright IR pixels is intense and the cells appear isolated. higher than in the rearward zone (70% versus Nieto et al.: Identification and Climatology of Cut-off Lows near the Tropopause 265

Figure 9. Diagram of the typical synoptic situation of a cut-off low showing the upper-level trough and cut-off stages of its life cycle using the geopotential fields at 200 hPa and 1000 hPa (left), the equivalent thickness and TFP scheme (middle), and the PV scheme (right). The cloud cover is overlapped by shaded areas. (Figures adapted from ZAMG.)

30%), because of the contribution of cloudiness velopment, these contours tend to form an associated with convective systems. This feature inverse omega shape that leads to a closed is particularly evident for the first stage over the cyclonic circulation (Fig. 9). However, at northern part of the frontal cloud band. More- 1000 hPa, sometimes no distinct low-level over, the difference between the rearward and features can be observed at all. The gradi- frontal zones seems to be accentuated for COLs ent of the geopotential height contours is during the cold season. generally weak at this level. Some weak cy- clonic circulation may appear, initiated by Horizontal Physical Parameters the circulation from aloft, during the later phases of a COL. The meteorological characteristics discussed (ii) Equivalent thickness is the thickness of so far are also associated with some of the the atmospheric layer between two pres- principal parameters in horizontal charts, for sure surfaces and is often used to locate example, the geopotential field at multiple fronts.31,32 In a low, this feature is char- levels (from the earth surface to the upper tro- acterized by a thickness ridge in front of posphere), the equivalent thickness, the ther- the low and a trough or a distinct min- mal front parameter (TFP), and the potential imum behind or in the center of the vorticity (PV). These are each considered in low. The equivalent thickness is given by: ∗ g lev2 Pu turn. Z =− Te ln( ), where g is the ac- T R lev1 Pl celeration from gravity, R is the universal (i) It is useful to consider three geopotential gas constant, Te is the layer mean equiva- heights in relation to the structure of COLs, lent temperature, Pu is the pressure at the namely 200, 500, and 1000 hPa. As pre- top of the layer, and P l is the pressure at the viously mentioned, during the initial stage bottom of the layer. the absolute topography at 200 hPa shows (iii) COLs are characterized by two baroclinic an upper-level trough. Geopotential height zones, one in front of the low, which is con- contours at the 500 hPa level exhibit the nected with a front-like cloud band, and an- same behavior.During the later stages of de- other one behind the low, connected with a 266 Annals of the New York Academy of Sciences

Figure 10. Left image shows the vertical cross-section line (A to B) chosen for a cut-off low over the Mediterranean Sea. Right image shows the typical parameters. (Figures adapted from ZAMG.) baroclinic boundary.33 There is a clear rela- Vertical Structure tionship between the well-known basic defi- In order to have a complete description of the nition of a front and the TFP.34 The TFP is a conceptual model of a COL, it is also necessary scalar variable. At its maximum, it fixes the to describe its vertical structure. Vertical cross position of the cold front where the tem- sections of (equivalent) potential temperature perature begins to fall and the position of showcoldairinornearthecenteroftheCOL the warm front where the temperature stops and a ridging of the isentropes below the COL. rising.35 The mathematical definition of the In some cases, equivalent potential temperature TFP is: TFP =−∇|∇T |·(∇T/|∇T |), even decreases with height in the lowest part of where T is the temperature of the analyzed the troposphere, indicating thermodynamic in- level. The temperature used in the TFP can stability. Relative vorticity shows a maximum be taken at any level. The first term of the that is coincident with the ridge structure of equation (−∇|∇T|) indicates the change of the isentropes. This vorticity maximum reflects the temperature gradient and the second the strong circulation around the center of the one (∇T/|∇T|) the projection in the direc- COL. Given that the air within a COL is colder tion of the temperature gradient. In other than its surroundings, translations and rotation words, TPF is the change of temperature produce a distinct cold advection (CA) ahead gradient in the direction of the tempera- of the COL and warm advection behind it. As ture gradient. The typical fields of equiva- noted above, the center of the COL is char- lent thickness and TFP in a cut-off low are acterized by a PV maximum. Therefore, in a shown in Figure 9. vertical cross section, higher values of PV are (iv) It is instructive to view COLs in terms of found within a COL than in its surroundings. PV on isentropic surfaces.5 The cut-off pro- Some of these parameters are shown in the cess of cold polar air is manifested as a cut vertical cross section in Figure 10. Typically, off of high-PV stratospheric air masses, and the region with high positive PV values in the thus COLs can easily be identified as iso- upper part of the COL is also characterized by lated features with stratospheric PV (PV> − − − (very) low water vapor content. 2PVU;1PVU=10 6 Kkg 1 m2 s 1)on isentropic surfaces.6 COLs are associated with low tropopause height compared to the Weather Events surrounding regions (i.e., the tropopause is generally lowest where the isentropic PV is The occurrence of COLs is often connected maximum). with severe weather conditions. Convection is Nieto et al.: Identification and Climatology of Cut-off Lows near the Tropopause 267 a key factor that determines the distribution probability of precipitation in these cases being of precipitation that is associated with a COL relatively low. Moderate precipitation is more event. Generally, the frontal cloud band at common in cases of deep COLs and when the the leading edge of a COL is thick enough COLs have a large vertical extension in terms of to produce precipitation and, in some cases, PV. The interaction of the medium- and high- frontal cloudiness at the rear edge.2 In gen- level flows of a COL with the orography and eral, COLs produce significant precipitation surrounding areas can be another important only when they receive a significant transport of factor. The orography (changing from the sea moisture at low and mid levels. In these cases, surface to the coast and littoral mountains) al- they become potentially unstable air masses.21 ters the flow in lower layers, creating the condi- Different rainfall patterns may be expected, de- tions for the development and organization of pending on the type of terrain below the low convective systems. system. When a COL occurs over land, its cen- ter is usually covered by a low cloud layer, Previous Analysis and Climatology suppressing the condition for convective pre- of COLs cipitation. Over warm ocean regions, COLs can be associated with deep convection and in- In recent years, COLs have been studied tense precipitation. However, the generalized in some detail, the focus of attention being idea that most COLs produce intense convec- either the numerical modeling of case stud- tive rainfall has been shown to be mislead- ies,39–43 the impact of COLs on tropospheric 25,36 ing in some studies. The severity of the ozone,44–48 mesoscale analysis of the decay of a storm depends on surface conditions, particu- COL,20,49 or observations of the associated syn- larly those related to moisture input. If the con- optic weather.45,50 Fewer climatological studies ditions are favorable, COLs can produce mod- of COLs also exist that use a variety of methods erate to heavy rainfall over large areas. Studies to identify individual COLs, usually employ- over the Mediterranean region (where the oc- ing different time periods, and that have been 26 currence of COLs is common ), show that developed to characterize limited geographi- the peak intensity of cyclonic-related precipi- cal domains. The approaches also differ in the tation is located at 300–400 km from the COL procedure used to identify the systems, with 25,37 center. Convective rainfall occurs within some of them using subjective analysis (e.g., a radius of about 300 km of the COL cen- Refs. 51,52) while others rely on the use of ter, with a pronounced peak close to the COL an objective (automated) analysis. These objec- center that is displaced marginally eastward, tive approaches can be based on the analysis of in front of the COL. The region of convec- geopotential height on a pressure surface (e.g., tive precipitation coincides with an area of re- Refs. 53–55), of PV data (e.g., Refs. 6, 14, 56), duced stability and with the leading baroclinic or on the physical parameters of the conceptual 2,4 area. At the same time, large-scale precip- model of COL.26 A review of previous clima- itation is distributed along the east–west axis tologies using different methods is presented in that passes through the COL. In about 50% the following section and in Table 1. of cases, it is this precipitation that dominates. It has also been shown that nearly 25–30% Subjective Analysis of COLs (mainly during autumn) do not in- duce any type of rainfall36 and only 25% are Price and Vaughan51 and Kentarchos and dominated by convection. The intensity of the Davies52 (hereafter PV92 and KD98, respec- COL can also influence the precipitation type. tively) developed similar subjective climatolog- Shallow COLs tend to produce convective pre- ical analyses of COLs over the entire North- cipitation very close to the low core, with the ern Hemisphere. These subjective studies are 268 Annals of the New York Academy of Sciences N) ◦ N) Geopotential charts at 200 hPa E) and 500 hPa S) W) ◦ ◦ ◦ ◦ W and at 500 hPa ◦ E–20 –60 –50 ◦ –75 ◦ ◦ –150 ◦ ◦ –75 ◦ N/30 E/20 S, 80 ◦ ◦ W/15 ◦ ◦ N) N) S/110 S, 80 N) S) tracks at 500 hPa ◦ ◦ ◦ ◦ ◦ ◦ N) N) ◦ ◦ –40 –20 ◦ ◦ –82 –70 –40 –35 W–80 –70 W–45 –60 ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ –70 –70 35 ◦ ◦ (24 (20 (25 (20 (100 (0 (30 (60 1.10.1982–30.9.1983 (0 May–September SubjectiveSubjective (30 (10 ObjectiveObjective 1950–1985Subjective 1963–1985 1985–1989ObjectiveSubjective Northern Hemisphere Northern 1989–1999 Hemisphere 1983–1996 Northern HemisphereObjectiveObjective Geopotential minimum at Northern 1953–1999 500 Atlantic hPa Southern Australia 1953–1999Objective Geopotential minimum at 500 hPa GeopotentialObjective- charts at 200 hPa 1958–1998Objective- Northern Hemisphere and NE 1992–2001 USA Northern HemisphereObjective and NE USA 1969–1999 Closed contours ofObjective Geopotential Geopotential PV charts minimum at at 200 500 hPa Geopotential minimum at Northern 1979–1988 500 Atlantic hPa Europe 1979–1993 Southern Hemisphere southern South America Northern Hemisphere Conceptual model of COL Geopotential minimum and Conceptual model of COL Conceptual model of COLs and Closed contours of PV 63 6 51 61 54 53 56 55 58 25 26 (2005) ıguez Objective 1989–January 1999 Northern Atlantic Closed contours of PV ´ 57 (1989) (2002) et al. Main Characteristics of Major Studies on COL Climatologies, Showing the Different Periods of Study, Methodology Used, Period of Analysis (2002) (2007) (2005) 52 62 14 (1999) et al. et al. et al. et al. et al. ´ andez (1999) ´ u et al. (2002) (2000) (1998) Cuevas & Rodr ReferenceParker Approach Period Region Methodology Qi Novak Pizarro & Montecinos Objective 1979–1996 Coast of Chile Meridional geopotential gradient Kentarchos & DaviesHern Subjective 1990–1994 Northern Hemisphere Geopotential charts at 200 hPa Bell & Bosart (1989) Price & Vaughan (1992) Porc Campatella & Possia (2007) Wernli & Sprenger (2007) Smith Fuenzalida Nieto TABLE 1. and Areas Covered Nieto et al.: Identification and Climatology of Cut-off Lows near the Tropopause 269 generally only used for relatively short time pe- 17 days, although most COLs were short lived riods as their approaches are labor intensive. (74% of COLs lasted 3 days or less with 41% of In particular, PV92 was developed for a 1-year them lasting only 1 day). They also suggested period (October 1982 to September 1983) and that southern COLs have a shorter life span KD98 for a 5-year period (1990–1994). The au- than northern ones, a result that was confirmed thors looked for a closed cyclonic geopotential by KD98. Indeed, they determined a life span contour or any closed circulation in the wind for their subtropical COLs of between 2 and 4 vectors at the 200 hPa pressure level. Three days, whereas their polar COLs lasted longer types of COL were defined by PV92, namely: a) (most of them more than 5 days). Both studies polar vortex, b) polar, and c) subtropical. Polar concluded that, during their early stages, COLs COLs, which usually extend to mid-latitude re- are either quasistationary or move erratically. gions, were the most intensely studied systems. PV92 found that almost half of the systems According to the study, subtropical COLs are moved considerably,and KD98 concluded that smaller than the other two types and their life about 50% of the COLs that last more than 3 span is shorter. The authors concluded that, days move significant distances (>600 km) be- a priori, subtropical COLs do not play a sig- tween the genesis and lysis stages. The COLs nificant role in the tropospheric ozone bud- tended to move northward (or northeastward) get. They also identified six favored geographic as they began to decay in strength. Some inter- regions for the occurrence of COLs: Europe annual variability in the frequencies of occur- (the most favored region with 33% of the total rence was also found in KD98. However, the number), northern China–Siberia, the north short period of the KD98 hampers any gener- Pacific, northeastern United States and east- alization of results. ern Canada, western United States, and the Studies of COL climatology were also under- northeast Atlantic. Analysis of the annual cy- taken for parts of the Southern Hemisphere. In cle revealed that COLs occurred more often particular, a relatively long subjective analysis in summer than in winter, particularly in June was undertaken for southern Australia,57 using and July (30% reported by PV92 and 37% by a 14-year (1983–1996) climatology of geopo- KD98). The season with the least number of tential height. The study was carried out using COLs was clearly winter (9% in PV92 and 15% 3-hourly 500 hPa synoptic charts from the Aus- in KD98). According to PV92, the seasonal dis- tralian Bureau of Meteorology archives. The tribution of COLs showed very similar patterns authors showed that the interannual variation when the analysis was performed separately of COLs was particularly large. They indi- for the three types of cut-off low. Furthermore, cated that most COLs are discernible during KD98 confirmed the same basic annual cycles winter (from May to October, with an aver- for COLs below lat 40◦N (subtropical COLs) age frequency of 9.3 days per month). These and for COLs at higher latitudes, between 40◦ Australian COLs moved either southeastward and 70◦N (polar COLs). These authors sug- (45%) or eastward (36%) and were concen- gested that the seasonal variations in latitudinal trated in a region south of lat 30◦S. Some of temperature gradients and temperature differ- the COLs moved very quickly, up to 10◦ lati- ences between continental and oceanic regions tude in 24 h, and some moved slowly, about 2◦ may be responsible for the seasonal cycle in latitude during 1 day. the numbers and positions of COLs through seasonal variations in jet stream configuration. Objective Analysis Concerning the duration of COLs, KD98 con- cluded that the majority of COLs lasted only 2– In order to develop an objective method of 3 days with very few lasting more than 10 days. detecting cut-off lows, several conditions are PV92 found that the duration ranged from 1– required.38 Any objective technique should be 270 Annals of the New York Academy of Sciences based on as few parameters and as little dif- Europe to the Caspian Sea and central Asia. ferentiation as possible and should make use The authors also detected a maximum along of a number of grid points appropriate to the the southern branch of the jet stream extending computer resources available. Furthermore, it from the eastern Pacific into the southwestern should be intelligible in the sense that any dif- United States. Finally, both studies found con- ferentiated quantities related to the initial fields siderable monthly, seasonal, and interannual should be readily apparent. It is important to variability. COLs declined from 1950 to 1970 develop objective approaches that yield reason- but increased from 1971 to 1985. A strong sea- able agreement with manual analysis, although sonal cycle was also found, with a maximum to try to reproduce subjective results precisely incidence during summer months. is pointless because manual analysis is not gen- In recent years, updated COL climatologies erally repeatable. Finally, it should be possible for the Northern Hemisphere have been com- to apply any objective method easily to any piled for the period from 1953 to 1999 by Smith gridded data set. et al.58 and Novak.55 The method used in both A preliminary characterization of COLs in of these analyses relied essentially on the al- terms of their spatial and temporal distribution gorithm developed by Parker et al.53 and Bell was provided by Parker et al.53 and Bell et al.,54 et al.54 These recent studies55,58 used twice-daily who constructed a Northern Hemisphere (from (0000 and 1200 UTC) 500 hPa gridded geopo- around lat 20◦N through 70◦N) climatology of tential height from the NCEP–NCAR reanal- COLs using data at the 500 hPa geopotential ysis.59 One key objective of both studies was to level. Their analysis was based on data from the provide greater precision for the northeastern National Meteorological Center of the United United States, in particular for the warm sea- States. The gridded data set had a resolution of son (from May to September). Favored COL 10◦ × 10◦ for Parker et al.53 and 2◦ latitude × 5◦ areas for the cold season include the north- longitude for Bell et al.54 In each case, a given ern Pacific Ocean, the southwestern United grid point was identified as a potential COL States, north-central and eastern Canada, the point if it was a geopotential minimum with re- North Atlantic around Greenland, and an east– spect to at least six of the eight surrounding grid west belt stretching from the Straits of Gibral- points. Once this set of candidate COLs was tar across northern Africa and the Mediter- identified, only grid points that showed a dif- ranean into the Turkish plateau.58 Moreover, ference in the geopotential value of about 60 m these studies have found that the COL distri- in Parker et al.53 and 30 m in Bell et al.54 were bution is orographically and synoptically de- retained. The authors made use of localized pendent. Finally, the frequency of occurrence closed circulation centers for a 36-year period generally increases and shifts toward the equa- (1950–1985) in the case of Parker et al.53 and tor with the onset of winter, with the strongest for a 15-year period (1963–1977) in the case signals occurring with the approach of spring. of Bell et al.54 In both cases, detached small Recent work by Nieto et al.26 provided the cyclones at 500 hPa were found primarily at first multidecadal climatological study of COLs middle and high latitudes. The occurrence of over the Northern Hemisphere (from lat 20◦N COLs was found to be higher within the main to 70◦N) for the period 1958–1998. This work belt of westerlies from northeast Asia to the was based on the NCEP–NCAR reanalysis59 Gulf of Alaska near lat 50◦N, and from eastern using a 2.5◦ by 2.5◦ resolution and daily geopo- Canada/northeastern United States to south- tential height, zonal wind, and temperature at east of Greenland and west of the United King- 200 and 300 hPa. COLs were identified us- dom. Their occurrence was also considerably ing an objective method that was based on the higher throughout a band extending from the three main physical characteristics of the con- east–central Atlantic Ocean across southern ceptual model of COLs.2,26 The restrictive and Nieto et al.: Identification and Climatology of Cut-off Lows near the Tropopause 271 consecutive conditions imposed were as follows: were obtained from reanalysis from the Euro- 1) a COL was characterized by a vortex (cy- pean Centre for Medium-range Weather Fore- clonic circulation) surrounding a geopotential cast (ECMWF) (ERA-40),60 with the authors minimum at the 200 hPa level and was an iso- constructing a 6-hourly COL database on a lated low from the westerly circulation (using 2.5◦ × 2.5◦ resolution grid. COLs were de- horizontal wind); 2) the equivalent thickness tected with the objective algorithm described was characterized by a thickness ridge on the above, and the results were then post processed eastern flank of the center; and 3) the baro- manually. The relevant pressure fields were an- clinicity and hence the TFP had higher values alyzed subjectively in order to assign the COL eastward of the central point. This objective center to one grid point and to ensure that analysis, using a very long data set, provided two or more consecutive synoptic times of de- a more comprehensive assessment of the issues tection corresponded to the same COL. The identified by other authors in previous studies. gaps were neglected if they exceeded 12 h. The In fact, this work confirms and reinforces previ- database was used to extract the geographical ous results, namely that the three preferred ar- distribution and duration of COLs. A vertical eas of COL occurrence in the Northern Hemi- structure analysis was also performed, charac- sphere are located in a) southern Europe and terizing the COLs from the 200 hPa level down along the eastern Atlantic coast, b) the east- to the earth’s surface. As a consequence, the ern north Pacific, and c) north China to the COL depth was defined as the lowest level at Siberian region extending to the northwest Pa- which a local geopotential minimum was de- cific coast. It also confirms that the European tected during the COL life cycle. The summer area is the most active region in the entire months were the most active (with a peak in Northern Hemisphere. COLs are more fre- June). A large number of events (41%) lasted quent during summer (about 50% of the total) less than 1 day and the number of detected than during winter and at higher latitudes than COLs decreased monotonically with increas- at lower ones. A more exhaustive analysis of the ing life span. Geographically, the following pat- areas of occurrence revealed a two-nuclei distri- tern was determined: a local maximum was bution in Europe in all seasons apart from win- located over the western Iberian peninsula in ter, a summer displacement toward the ocean the Atlantic, a second maximum toward the in the American region, and a summer ex- north of Spain/southern France, and a third tension to the continent in the Asian region. one from the Ionian Sea to Anatolia and the Furthermore, about 47% of COLs formed a Black Sea. Furthermore, by distinguishing be- surface low during their cycle and 73% corre- tween COLs with a life span more than 1 day spond to maxima of PV.The life span of COLs (HCOL) and less than 1 day (LCOL), it was was reconfirmed as being generally less than determined that while HCOLs are preferably 3days(veryfewlastmorethan5days),and, located over the western Portugal coast, the dis- in general, COLs last longer when they occur tribution of LCOLs showed two lesser peaks, at higher latitudes. Less that 20% of COLs are one over the Iberian peninsula and one over stationary during their life cycle, with the ma- the western Black Sea. The results showed that jority being highly mobile. about 60% of the detected COLs had an as- Based in part on the technique described sociated closed geopotential contour at surface previously, Porcu´ et al.25 carried out a hybrid level. objective–subjective scheme to study the dis- Objective methods have also been used tribution of COLs for the decade 1992–2001 to study spatial and seasonal distributions ◦ ◦ in the European area (lat 30 W–45 Eand of COLs over the Southern Hemisphere. ◦ ◦ long 20 –60 N) restricted to the warm season In particular, Fuenzalida et al.61 studied (April to September). In this study,gridded data the occurrence of COLs at 500 hPa using 272 Annals of the New York Academy of Sciences four-times-daily NCEP–NCAR reanalysis considered as a COL. The typical intense sea- fields for the 31-year period from 1969 to sonal variability was found (ranging from two 1999. They used a mixed method based on to 15 events), with the austral spring being the an objective detection of lows combined with most frequent season. The second climatology a visual inspection. The region under analysis was developed using an algorithm based on covered the entire Southern Hemisphere a COL conceptual model.26 The selected do- between lat 10◦ and 60◦S. COLs tended to main extended from long 100◦Wto20◦Wand occur in a belt between lat 20◦ to 50◦S with a from lat 15◦Sto50◦S. The highest frequency maximum density at lat 38◦S and with a higher of occurrence was over the Pacific area (44%), frequency around the three main continental followed by the Atlantic (30%) and continen- areas and a lower frequency over the oceanic tal areas (26%). COLs were more frequent in regions. Using three equally spaced sectors the austral autumn (in particular over the coast covering the whole Southern Hemisphere, of Chile) and less frequent during summer. A the authors showed that only 10% of the shorter duration of COLs (2–3 days) was also COLs were found in the African sector, with detected, but over continental areas COLs were 48% in the Australian sector and 42% in the of longer duration. South American sector. The COLs showed The COL climatologies presented so far a strong seasonal cycle in South America are essentially based on geopotential height. and Africa (with a lower occurrence during A different approach is also possible because summer), while in the Australian area they COLs are characterized by an anomalous occurred fairly steadily throughout the year. lower tropopause, which induces higher val- The authors further noted that over South ues of PV within the COL core. This feature America, the dissipation of COLs was a more was used to identify COLs over the northern common feature than their generation, with Atlantic during the decade from 1989 to 1999 the opposite behavior occurring over the by Hernandez´ 56 and Cuevas and Rodr´ıguez.14 Australian continent. Close to South America, Both studies used the operational ECMWF 12- a statistically significant increase in the number hourly data at 315 K, 320 K, 325 K, and 330 K, of COLs over time was observed at the 500 hPa with a horizontal resolution of 1.25◦ × 1.25◦. level, with a higher frequency of COLs being The criterion used to locate COLs was the same generally observed after 1990. Typical life for both studies, namely that a COL was de- spans of COLs were found to be similar fined as a closed contour on isentropic surfaces to those found for COLs in the Northern of 2 PVU, with an inner maximum of at least Hemisphere, with most COLs lasting for 4 PVU, and covering a minimum extension a short time; only 10% lasted longer than of 30◦ × 30◦. The results obtained using this 5 days. The frequency distribution of intensity approach confirm that the highest frequency (maximum value of geopotential on each COL of COLs can be observed in the vicinities of track) showed no geographical variation. the Iberian peninsula during all seasons, with Using smaller areas, such as the subtropical two peaks of occurrence, one over southwest- coast of Chile62 or southern South America,63 ern Iberia and the other over southern Iberia two climatologies were developed using (this maximum did not appear in Ref. 14). NCEP–NCAR reanalysis during the periods The monthly occurrence was higher during the 1979–1996 and 1979–1988, respectively. The warm season, with June and July being the most first one identified COLs using the meridional active months, which was in agreement with geopotential gradient at 500 hPa, calculated the highest values of tropospheric ozone mea- as the difference between geopotential heights surements indicating STE. Over southeastern over lat 30◦–35◦S, long 80◦–75◦Wandlat35◦– Europe, two kinds of behavior were described 40◦S, long 80◦–75◦W. Negative values were concerning the movement of COLs, namely (1) Nieto et al.: Identification and Climatology of Cut-off Lows near the Tropopause 273 that they were either quasistationary or (2) that are more frequent in summer than in winter they moved in a cyclonic sense with a mean and that they occur preferably near the major gyre of 20◦ in 12 h and a maximum movement troughs of the large-scale circumpolar flow. Al- of 8.75◦ in longitude and 6.25◦ in latitude in most all the studies found commonly favored 12 h. As in previous climatologies, the interan- regions of occurrence, namely southern Eu- nual occurrence showed large variability, with rope and the eastern Atlantic coast, the eastern an absolute maximum (minimum) taking place North Pacific, and the north China–Siberian around 1993 (1998). region extending to the northwest Pacific coast, Finally, Wernli and Sprenger6 presented a with Europe being the region of most common climatology of potential vorticity (PV) stream- occurrence. Their duration is typically from 2– ers and cut offs for the Northern Hemisphere 3 days with few lasting more than 5 days. Anal- defined on isentropes from 295 K to 360 K. yses of larger data sets of several years or more Their analysis covered the period from 1979 also found very large interannual variability in to 1993 and was based on the shorter ERA- the occurrence of COLs at the annual and 15 reanalysis data set from the ECMWF. In seasonal scales, although without any signifi- their case, COLs were identified as closed iso- cant trends. Several major large-scale modes of lines of the two-PVU contour on the previ- variability in atmospheric circulation seem to ously mentioned isentrope surfaces. One of have an impact on the occurrence of COLs. In their main findings relating to the spatial distri- particular, blocking events, the North Atlantic bution of COL events revealed a pronounced Oscillation (NAO), the El Nino–Southern˜ Os- zonal asymmetry and a clear change of fre- cillation (ENSO), and the persistence of the quency maxima with altitude. For instance, at stratospheric polar vortex may influence the 300 K and during winter, most COLs were occurrence of COLs.3,64 Despite the few (and found over the western side of Canada and only empirical) studies that look for the influ- Siberia (near lat 50◦–60◦N), whereas on higher ence of these modes on COL occurrence, there isentropes the frequency maxima were shifted are physical reasons that provide support for further to the south and were generally found this relationship. Blocking events and COLs are at the downstream end of the North Atlantic both associated with the occurrence of impor- and Pacific storm tracks. With respect to con- tant disruptions of the upper tropospheric jets tinental areas, the Mediterranean stands out as well as being associated with strong merid- as a region with a particularly high number of ional components and even with the bifurcation COL episodes. In addition to the pronounced of the jet and the appearance of double jets.65,66 geographical variability on specific isentropic Blocking events and COLs, therefore, tend to surfaces, a distinct seasonal cycle can be dis- occur mainly during winter. The relationship of cerned. This cycle is attributed mainly to the the NAO or ENSO with the intensity and zon- seasonal shift of the isentropic surfaces them- ality of the jet flow is well known.67,68 The neg- selves. In fact, if the COL frequency is con- ative (positive) ENSO and NAO phases are re- sidered as a function of latitude, irrespective of lated to the appearance of weaker (stronger) jets the vertical altitude, a fairly robust pattern with and to a higher (lower) probability of COLs oc- almost no seasonal cycle emerges. curring. Furthermore, there are dynamic rea- sons for thinking that the occurrence of COLs Common Results of Previous could be influenced by the persistence of the Climatologies stratospheric vortex (the quasisteady summer- like state after the decay of the vortex with Taking into consideration all the studies sum- strong weakening of the westerlies favors COL marized above for the Northern Hemisphere at development, and the fact that remnants of the extratropical latitudes, it can be seen that COLs vortex survive as coherent PV structures for 274 Annals of the New York Academy of Sciences around 2 months for early breakup years69,70). ing as a basis the methodology developed by As a consequence, COL events appear to be Wernli and Sprenger.6 This analysis is applied more frequent during the late spring and in only to the ERA-40 data set from the ECMWF summer months for those years characterized covering the period from January 1958 to De- by an earlier vortex decay. cember 2002. The required fields (horizontal wind components, temperature, and geopoten- tial) are available every 6 h on 60 vertical levels from the surface up to 10 hPa and were inter- Climatologies Based upon ◦ the Conceptual Model and polated onto a regular grid with a 1 horizontal Potential Vorticity resolution. Secondary fields, such as potential temperature and PV, were computed on the Data original hybrid model levels. Finally, the PV field was interpolated to a stack of isentropic Three climatologies of COLs were calcu- levels from 300 to 350 K at 10 K intervals (see lated using two reanalysis data sets, namely the Fig. 11 for an illustrative example showing the ERA-40 and NCEP–NCAR. ERA-40 reanal- heights at which these isentropic surfaces are yses60 were produced by the ECMWF in the located). late 1990s, using a T106 model resolution with 60 vertical levels and the optimum interpola- Climatology Based on the tion data assimilation technique. The NCEP– Conceptual Model NCAR reanalysis59 is an intermittent data as- similation scheme performed with a T62 model COLs are identified by means of three con- with 28 vertical sigma levels and the opera- secutive and restrictive steps that are based on tional statistical interpolation (SSI) procedure the physical characteristics of the conceptual for assimilation. The first two climatologies model of COLs described in section 2 and by were built from 6-hourly data from the ERA-40 Nieto et al.26 (hereafter, NAL). The condition and from the NCEP–NCAR and based on the of a minimum of geopotential height field of conceptual model of COL developed by Nieto 200 hPa and the isolation from the general cir- et al.26 This approach uses geopotential, zonal culation pattern of westerlies in the upper tro- wind, and daily temperature data from 200 and posphere was the first imposed restriction used 300 hPa with a 2.5◦ by 2.5◦ resolution. Those in considering a grid point as a part of a pos- cut-off systems found north of lat 70◦Norsouth sible cut-off low. At each 6-h interval, a given of lat 20◦N were not included in the study. The grid point was therefore identified as a geopo- reason for excluding systems north of 70◦ was tential minimum if it is a minimum (within a in order to eliminate the main polar vortex, and 10 gpm threshold) with respect to at least six of the reason for excluding cut-off systems south the eight surrounding grid points. Once this set of 20◦N was the small probability of their oc- of COL points was chosen, we retained only curring, despite the fact that this could be physi- the grid points that showed changes in the di- cally meaningful as has been shown in previous rection of the 200 hPa zonal wind at either climatologies. The time periods covered by the of the two grid points situated immediately to analyses varied, depending on the availability of the north of that grid point. The next required data. While the ERA-40 data were only avail- condition for a COL was that it should have able from January 1958 to December 2002, a greater equivalent thickness (computed be- the NCEP–NCAR data set extended from Jan- tween two pressure levels, in this case 200 and uary 1948 to December 2006. The third clima- 300 hPa) to the east of its central point than this tology was developed using potential vorticity value at its center. Finally, the grid point imme- (PV) as the physical parameter of diagnosis, us- diately to the east of a candidate COL point Nieto et al.: Identification and Climatology of Cut-off Lows near the Tropopause 275

Figure 11. Potential temperature (corresponding to isentropic surfaces) and tropopause (two-PVU isosur- face) along the 50◦W meridian for January (left) and July (right). Of particular interest to this study are the isentropes between 300 and 350 K. The horizontal axis is the geographical latitude, the vertical axis the pressure (in hPa). was required to have a TFP higher than that cise and more extensive than those obtained at the COL point in order to locate the frontal in NAL, mostly because of the higher tem- baroclinic zone. The temperature at 200 hPa poral resolution and the extended time period was used to calculate the TFP in our analysis. considered. If all these conditions were fulfilled, the ana- lyzed grid point was considered to be a COL Case Study point. It should be noted that this algorithm is The possibility of wrongly identifying a COL essentially identical to the one previously devel- was minimized by imposing the consecutive oped by NAL but makes use of 6-hourly data and restrictive conditions described above, but instead of one single daily value. Some other the possibility nevertheless still existed of failing refinements were made to the NAL method in to identify systems that, in a subjective analysis, terms of the method of calculating the TFP, would have been considered as cut-off lows. thus achieving better resolution in temperature Figure 12 summarizes the procedure im- gradients. Some of the COL points detected posed at each step for a real example of a cut- were probably the same COL observed over off low. The method identified a cut-off low consecutive days. In NAL, several conditions on May 13, 1992 that lasted only 6 h. Subjec- were imposed to determine where the same tive analysis might identify this COL as end- COL had been identifed more than once, but ing on May 15, 1992, but by applying our in these climatologies the plots and statistics will methodology, days after May 13, 1992 were be presented for all points that fitted the COL not considered as cut-off low days because they conditions outlined above. It should be noted did not fit all our imposed conditions, which that, unlike in NAL, systems that lasted only must be satisfied simultaneously. In particular, 1 day or less were not excluded from the anal- the most restrictive conditions, concerning the yses. With these additional steps, we are confi- minimum thickness and the TFP, were not sat- dent that both the COL databases derived from isfied. On May 13, 1992, 1200 UTC, a mini- ERA-40 and NCEP–NCAR are now more pre- mum of geopotential over the southern Italian 276 Annals of the New York Academy of Sciences

Figure 12. Cut-off low detected on 13 May 1992 at 12 UTC that met all the criteria imposed by the objective method. The box shows the COL point detected and the underlining grid point fits the imposed condition. peninsula was detected at 200 hPa (left top Climatology panel of Fig. 12). The evolution of the COL at that moment corresponded with the upper- This section presents the two climatologies level trough stage and the initial tear-off stage, that were based on the conceptual model of ◦ but the circulation was already detached from COLs over the Northern Hemisphere (lat 20 – ◦ the westerlies, as shown by the zonal wind pat- 70 N). Figure 13 shows the seasonal frequency terns at 200 hPa (right top panel). The baro- with which a cut off is found over a grid point us- clinic boundary was detected to the east of the ing the ERA-40 data set (hereafter, ERANAL), selected COL point (bottom panels). Only 6 h and Figure 14 shows the equivalent results later, only two out of four conditions were ful- using the NCEP–NCAR data set (hereafter filled, which explained why this COL had a life NCEPNAL). The gray scale refers to the per- span shorter than 1 day using our new clima- centage per 1000 (‰) of COL points over a grid tology. Later, it will be shown that this COL point, counting all the hourly occurrences. This nevertheless complied with the conditions im- display differs significantly from those used in posed in the diagnosis of the PV. previous studies of spatial distribution, which Nieto et al.: Identification and Climatology of Cut-off Lows near the Tropopause 277

Figure 13. ERA-40 (1958–2002) climatology cut-off lows (ERANAL) in a 2.5◦ × 2.5◦ grid for winter, spring, summer and autumn. The gray shading gives the percentage per thousand (‰) of time for which a grid point is part of a cut off. showed only the frequencies of the first day of east United States, and (3) the north China– COL occurrence. Siberian region extending to the northwest Pa- The most important finding is that the ge- cific coast. ographical distribution of COLs is very simi- Another significant outcome is that for both lar in both climatologies and is also coherent climatologies, the position of the COLs changes with what was obtained in NAL, with most of coherently from season to season. COLs are the cut-off lows being concentrated in three much more common in summer [June to Au- main areas. These areas are (1) southern Eu- gust (JJA)] (lower left panel) than in winter [De- rope and the eastern Atlantic coast, includ- cember to February (DJF)] (upper left). We cal- ing the Mediterranean and northern Africa, culate that 43% (41%) of these COLs take place (2) the eastern North Pacific coast and north- during summer and 15% (17%) during winter, 278 Annals of the New York Academy of Sciences

Figure 14. NCAR-NCEP (1948–2006) climatology cut-off lows (NCEPNAL) in a 2.5◦ × 2.5◦ grid for winter, spring, summer, and autumn. The gray shading gives the percentage per thousand (‰) of time for which a grid point is part of a cut off. using ERANAL (NCEPNAL). During summer, bands detected over the western Pacific coast most COLs are found in a bipolar band with of China and over the eastern Pacific coast of two centers for both climatologies, one span- North America have similar extensions in both ning the middle North Atlantic to the Iberian climatologies, with a region of higher COL oc- peninsula and another centered over the east- currence for ERANAL over the mid Pacific ern Mediterranean, reaching values with peaks (>4‰). During winter the pattern is more sim- around 6.5‰. The maximum of occurrence ilar between climatologies than during sum- over the Atlantic area has a larger zonal exten- mer. For ERANAL the western Pacific coast sion for ERANAL than for NCEPNAL, with of China is the area of higher occurrence, and the former starting from the mid ocean and the COLs are more frequent over the northern At- latter starting near the Iberian peninsula. The lantic coast of North America than the eastern Nieto et al.: Identification and Climatology of Cut-off Lows near the Tropopause 279

Pacific coast of North America. Comparing the tropospheric) reservoir has not split from the three main regions, the eastern Pacific coast of main body, which of course has nothing to do North America is where COL occurrence is with the traditional view of cut offs. most unstable—limited to the Pacific coast dur- As was the case in WS, a clear distinction ing winter and then expanded and displaced between stratospheric cut offs and ephemeral westward during the remaining seasons. This diabatically produced PV anomalies is difficult is most probably because of the strong seasonal to obtain and would require the computation cycle that is imposed by the meridional dis- of trajectories to check for PV conservation. placement of the jet stream (the main produc- However, such an analysis is computationally tion mechanism that is responsible for cut-off expensive and has been performed in the study low formation) over the areas most prone to reported herein. As will be shown later, it ap- COL formation. Over the European region, pears that at the lower isentropes considered, the bipolar structure that is found in summer some of the PV cut offs identified do appear disappears during winter and COLs tend to oc- to be related to the surface topography. How- cur toward lower latitudes than during summer. ever, because of their stationary nature, they do The spring and autumn seasons are character- not get mixed up with the real stratospheric cut ized by a transitional structure pattern. offs. A simple, but rather rough, elimination of surface-induced cut offs is implemented in the Climatology Based on Potential Vorticity new algorithm. All cut offs must be character- ized by a minimum distance of 75 hPa from Stratospheric and tropospheric cut offs were the COL grid point with the highest pressure identified using the following algorithm: (a) value to the surface pressure, otherwise they are first, the PV on an isentropic surface is sep- neglected in the climatology. arated into parts with PV larger and smaller than 2 PVU; (b) based upon this distinction, Case Study a unique label is attributed to each connected In this section, the same case study of cut-off patch of high (>2 PVU) and low (<2PVU)PV formation that was assessed previously (Fig. 12) air, resulting in a clustering of stratospheric and is discussed, but this time within the framework tropospheric air masses on that isentropic sur- of the PV-detecting approach. Figure 15 shows face; (c) the largest patches of connected high an isentropic PV (IPV) map on the 320 K isen- and low PV air are discarded because they cor- trope. On May 13, 1992, 0600 UTC, a narrow respond to the stratospheric and tropospheric filament of high-PV air extended from Poland PV reservoirs, respectively; (d) the remaining southwestward to Greece and southern Italy patches are saved as stratospheric (for high PV) (left panel). While the northern stratospheric andtropospheric(forlowPV)cutoffs.Notethat high-PV reservoir remains a compact body, the this algorithm is somewhat different from the filament shows a tendency to cut off. This fea- one recently applied in Wernli and Sprenger6 ture is clearly discernible in the narrowing of (hereafter WS). A minor difference applied in the filament to the west of the Black Sea. In- this study is in relation to an area threshold. If a deed, 36 h later (right panel), the southern- connected tropospheric or stratospheric patch most parts of the filament were completely de- is smaller than 5 × 104 km2 or larger than coupled from the stratospheric reservoir and 1000 × 104 km2, it is discarded in this data showed as a distinct PV cut off. Since this was set. The former criterion guarantees that mi- high-PV air of stratospheric origin, it is known nor and most probably spurious cut offs with a as a stratospheric cut off. The wind vectors are weak dynamic impact are neglected, whereas also included in Figure 15. In accordance with the latter criterion is a “security check” to en- the invertibility and partitioning principles of sure that a large part of the stratospheric (or the PV perspective, as summarized by Hoskins 280 Annals of the New York Academy of Sciences

Figure 15. Isentropic potential vorticity (IPV) map on the 320 K isentrope for 13 May 1992 at 06 UTC (left) and 14 May 1992 at 18 UTC (right). Potential vorticity values are given in PVU (1 PVU = 1 × 10−6 m2 s−1Kkg−1). Additionally, the wind vectors for the 320 K isentrope are overlaid on the figure, with the wind speed (in m/s) shown on the lower right corner. The dynamical tropopause (two-PVU isosurface) is marked by the bold black line. Note the slightly different geographical region shown in the two panels. et al.,5 the stratospheric PV cut off is associated with a cyclonic wind field. Although a clear indication of the strato- spheric origin of the Mediterranean PV cut off is obtained from this Eulerian approach, especially if intermediate time steps are also considered, a Lagrangian approach reveals its northern source region even more clearly. To this end, air parcels within the stratospheric cut off (Fig. 15, right panel) were set as starting points for a backward trajectory calculation. Figure 16 shows the trajectories as calculated by ECMWF winds and by the 3D Lagrangian trajectory tool LAGRANTO71 that involves the calculation of extensive coherent ensembles of trajectories with a 6-h time step. Trajectories Figure 16. Backward trajectories using LA- emanating from the affected area are evalu- GRANTO from the stratospheric cut off in Figure 15. ated over the subsequent 72 h—a time pe- The trajectories start at 14 May 1992, 18 UTC and riod which includes a phase of upper-level PV extend 72 h backward in time. isolation. The air parcels moved in an almost zonal a location over Greece and were then “cap- flow pattern over Scotland and Denmark, be- tured” in the cyclonic flow around the PV cut fore turning southward. They quickly reached off. Nieto et al.: Identification and Climatology of Cut-off Lows near the Tropopause 281

Figure 17. Geopotential height at 13 May 1992, 06 UTC (left) and 14 May 1992, 18 UTC (right), and for the two levels 250 hPa (upper) and 500 hPa (lower).

Finally, it is worthwhile considering the tra- geopotential height at 250 hPa may be observed ditional view of this PV cut-off event, i.e., its to the west of the Black Sea, showing the tear- manifestation as a closed contour of geopoten- off stage. This small-scale cut off coincides with tial height. This is shown in Figure 17 where the the southern parts of the narrow PV filament geopotential height at 250 and 500 hPa is plot- in Figure 15. At the lower level (500 hPa), no ted for the two times shown in Figure 15 (May closed contours can be seen, which is indicative 13, 1992, 0600 UTC and May 14, 1992, 1800 of the shallow vertical extent of the 250 hPa cut UTC). For the first of these, a closed isoline of off in this stage prior to the cut-off phase. In 282 Annals of the New York Academy of Sciences sharp contrast to this, for the following day, a sufficiently strong upper-level PV cut off co- closed isolines of geopotential height are found incides with a geopotential cut off that extends at all levels, from 250 hPa down to 500 hPa. over a significant part of the troposphere. Hence, during the cut-off stage, this COL is characterized by a significant vertical depth. Climatology The above example clearly illustrates the fol- We now present a climatology of strato- lowing important features: (a) an upper-level spheric PV cut offs from 1958 to 2002. PV cut off might be associated with a geopo- Figure 18 shows the seasonal distribution of fre- tential cut-off low; (b) the opposite is not nec- quencies of stratospheric PV cut offs. The most essarily the case, as for example the narrow striking aspect is the clear deviation from zonal PV filament (the stratospheric PV streamer) is symmetry. During winter (upper left panel), also associated with a closed geopotential cut most stratospheric PV cut offs are found in off, albeit at a smaller scale; and (c) a substan- an almost zonal band that extends from the tial upper-level PV cut off is associated with a Azores to the eastern Mediterranean. Values geopotential cut off extending over a significant of peak frequency in this band reach up to part of the troposphere (at least from 250 down 4%. Further downstream, this band extends to 500 hPa in this case). Considering the mis- as far as the Pacific coast of China, although match between the two methods, an obvious with a considerably reduced amplitude. A sec- question arises in relation to the physical rele- ondary maximum is discernible over the west vance of either definition. The physical mean- coast of North America, extending westward ing of geopotential cut offs was discussed in an to the eastern Pacific and eastward almost to earlier section. Here, some additional attention the east coast. One possible explanation for is given to the meaning of PV cut offs. One this geographical distribution is linked to the key aspect of PV theory is the invertibility prin- North Pacific and Atlantic storm tracks. It is ciple. This states that the PV distribution in well known that the exit regions of these two the interior of a domain is sufficient to derive storm tracks6 are affected greatly by nonlinear the wind and temperature field in the domain, distortions of the two-PVU isoline on an isen- provided that some boundary conditions are tropic surface, i.e., by the breaking of Rossby specified and a suitable relationship between waves. As shown in Figure 15, the associated the geopotential height field and the wind field PV streamers quite often break up, forming is assumed (this relationship is often referred stratospheric PV cut offs. It is worthwhile men- to as the balance condition underlying the PV tioning that the PV cut-off maxima are found inversion). In this PV framework, it is relatively slightly to the south of the main region that straightforward to show that a positive upper- contains breaking Rossby waves. During spring level PV anomaly (for instance, a stratospheric (upper right panel), the geographical patterns PV cut off) is associated with a cyclonic wind remain essentially the same as in winter, with field and a decreased temperature below the an intriguing third maximum appearing over anomaly. It is particularly important that the Japan. Summer (lower left panel) is associated impact of the upper-level PV structure should with the highest PV cut-off frequencies, with not be restricted to these upper levels but in- peak values of 8% very often being surpassed. duces wind and temperature anomalies below Note also that the geographical pattern is much and above them as a consequence of the el- more zonally symmetric, being shifted signifi- liptical character of the PV inversion problem. cantly to the north compared to during winter. Through this mechanism, a sufficiently strong Given that the climatology is calculated on an upper-level PV cut off must coincide with some isentropic surface (320 K in this case), the sea- vertically coherent anomaly of horizontal wind sonal shift of the PV cut off simply reflects the and temperature. In summary, it is clear that seasonal shift of the isentropic surface (see also Nieto et al.: Identification and Climatology of Cut-off Lows near the Tropopause 283

Figure 18. ERA-40 (1958–2002) climatology of stratospheric PV cut offs on the 320 K isentrope for winter, spring, summer, and autumn. The gray shading gives the percentage (%) of time for which a grid point is part of a PV cut off.

Fig. 11). Indeed, as a result of this seasonal shift, this southward shift, a westward shift of maxima it is essential that other isentropic surfaces are can be seen relative to their equivalent positions considered, in order to obtain a comprehensive for the 320 K surface. For instance, the spring view of the distribution of cut-off formation. maximumoverthewestcoastoftheUnited Finally,autumn (lower right panel) is again very States (at 320 K) can now be found in the mid similar to spring. Pacific (at 340 K) and the autumn maximum The stratospheric PV cut-off climatology on over Europe and the Mediterranean (at 320 K) the 340 K surface is presented in Figure 19 is now shifted to the mid Atlantic. It should also in which it can be seen that the maxima are be noted that the summer distribution pattern located further to the south (consider the in- on the 340 K surface is significantly different tersection of the 2-PVU isoline with the 320 from the much more zonal pattern found on and 340 K isolines in Fig. 11). In addition to the 320 K surface. 284 Annals of the New York Academy of Sciences

Figure 19. ERA-40 (1958–2002) climatology of stratospheric PV cut offs of the 340 K isentrope for winter, spring, summer, and autumn. The gray shading gives the percentage (%) of time for which a grid point is part of a PV cut off.

◦ ◦ Comparison and Conclusions latitude band between 20 Nand70N. The ERANAL (gray areas) and NCEPNAL (con- Preliminary inspection of the two climatolo- tinuous black line) climatologies agree quite gies described in Section 4 shows good agree- remarkably, with the few differences restricted ment in terms of their spatial distribution. A to their extent; the main areas of occurrence, more rigorous comparison between these ap- namely the European area, the Asian area, proaches can be obtained by plotting in a sin- the North American Pacific, and Atlantic ar- gle figure the seasonal percentage of COLs eas were detected equally using both data sets. detected in the areas of maximum COL oc- The detection of these areas of main COL oc- currence (Fig. 20). This analysis is limited to currence in the stratospheric WS climatology the common period of input data, i.e., Jan- depends on the isentropic level under consider- uary 1958–December 2002 and the common ation. Several isentropic levels (310, 320, 330, Nieto et al.: Identification and Climatology of Cut-off Lows near the Tropopause 285

Figure 20. Comparison between seasonal COL climatologies. Gray areas are ERANAL climatology, the black line denotes NCAR-NCEP NAL climatology, and squares, asterisks, rhombuses, and dots mark the longitudinal axes of main areas of occurrence from WS climatologies at 310, 320, 330, and 340 K isentropic levels, respectively. and 340 K) were plotted using different symbols model, whereas the WS methodology attributes to indicate the longitudinal axes of main COL the same COL event to many more grid points occurrence from the WS climatology. It should surrounded by a PV line. In a typical event, be noted that the plot accounts only for areas of therefore, we have only one or two grid points occurrence without representing the actual fre- indicating a COL when applying the ERANAL quency of COLs. Such fields have been shown and NCEPNAL methods, but we can easily ob- before in Figures 12, 13 and 18, 19. The very tain 10 or more grid points using WS. large differences in counts between these two Further comparison between areas of occur- approaches (almost an order of magnitude) is rence (Fig. 20) shows that the higher the latitude not from the actual number of individual COLs of the COL region, the lower the isentropic but results from the different approaches used level that is appropriate for the comparison. In in counting them. ERANAL and NCEPNAL the Asian and North American Atlantic sec- (Figs. 12 and 13) represent a COL by a lim- tors, the main areas of COLs were commonly ited number of grid points that satisfy the four detected at 310 and 320 K, while in the Eu- restrictive criteria imposed by the conceptual ropean and the North American Pacific areas, 286 Annals of the New York Academy of Sciences

Figure 21. Mean seasonal potential temperature (K) corresponding to geopotential fields at 200 and 300 hPa for the extratropical Northern Hemisphere derived from NCAR-NCEP reanalysis. Main areas of COL occurrence are plotted over the latitude axis. Continuous black segments indicate the isentropic level where COLs are identified for different latitudes.

COLs were detected using 330 and 340 K. This the strength and meander of the jet stream. applies for all seasons with the exception of Figure 22 shows the seasonal distribution of winter when COLs occurring at higher lati- COLs for the three climatologies described in tudes were identified on lower isentropes. So, this paper, using several isentropic levels for during winter, COLs in the European and the WS climatology (310, 320, 330, 340, and North American Pacific areas were detected at 350 K). In this comparison, we use only one 310, 320, and 330 K. In summary,for each sea- COL point to represent each hourly cut-off low son and for different areas of COL occurrence, (and not all the points that fit the conditions different isentropic levels must be used to ob- imposed by the methods). The well-known sea- tain a good representation of the main areas of sonal pattern (maximum during summer and COL occurrence. Given that we used two lev- minimum during winter) is shown in all the els (200 hPa and 300 hPa) to detect ERANAL analyzed climatologies, with some slight dif- and NCEPNAL, the possible potential temper- ferences in the amplitude of the seasonal cy- ature intervals, as shown in Figure 21, were cle. The best agreement is obtained between those between 310 and 350 K, depending on ERANAL, NCEPNAL, and WS for the 320 K the different areas of occurrence and on the level. Of all the seasons, the best agreement oc- season. In general, the most appropriated isen- curs in winter and the worst in summer when tropic levels to analyze each latitude band are: higher relative differences are found between from 30◦Nto60◦N, 320 and 330 K; for lati- the two approaches. The agreement between tudes lower than 40◦N, 340 K; and for latitudes theseasonalcyclesisalsopresentinthemonthly higher than 40◦N, 310 K. Analysis of the sea- distribution of occurrence (Fig. 23), with a com- sonal cycle allows the comparison of the two mon maximum in July and common minima in approaches for COL identification. Most pre- December or January, and with a higher am- vious studies of COLs in the Northern Hemi- plitude of the monthly cycle for the 340-K level sphere agree that they are more frequent during than for the other climatologies. the warm season, which is consistent with the The study reported herein reviews and ex- relationship between COL development and pands previous multidecadal climatologies of Nieto et al.: Identification and Climatology of Cut-off Lows near the Tropopause 287

Figure 22. Seasonal distribution of the number of COL points for the three climatologies (five isentrope levels for WS, 310, 320, 330, 340, and 350 K).

Figure 23. Monthly distribution of the number of COL points for the three climatologies (two isentrope levels for WS, 320, and 340 K). cut-off low systems in the Northern Hemi- model of cut-off lows and the other based on sphere. Cut-off lows were identified from data an analysis of IPV. The application of the al- taken from the NCEP–NCAR and ERA-40 re- gorithms to the extended data sets improves analyses, and the study makes use of two objec- our ability to perform more accurate studies tive methods, one based on imposing the four of the seasonal and interannual variability of main physical characteristics of the conceptual these systems. Comparison between the results 288 Annals of the New York Academy of Sciences derived from the two approaches shows that, References in general terms, the agreement between the methods is good for extratropical areas. Appar- 1. Palmen,´ E. & C.W. Newton. 1969. Atmospheric Circu- ent differences in the percentage time for which lation Systems: Their Structure and Physical Interpretation. a grid point is part of a cut off is only a result Academic Press. New York. 603 pp. of the particular method for counting the grid 2. Winkler, R. & V.Zwatz-Meise. 2001. Manual of syn- optic satellite meteorology. Conceptual models, Vers. points that form a COL. For the COL data sets 6.0. (Available at Central Institute for Meteorology based on the conceptual model, the number and Geodynamics, Hohe Warte 38, 1190 Vienna, of COL points detected is one order of mag- Austria). nitude lower than for the COL data set based 3. Gimeno, L., R.M. Trigo, P. Ribera & J.A. on PV. The same COL could be represented Garc´ıa. 2007. Editorial: special issue on cut-off low systems (COL). Meteorol. Atmos. Phys. 96: 1–2 by only a few very significant grid points for (doi:10.1007/s00703-006-0216-5). the approach based on the conceptual model 4. Kurz, M. 1998. Synoptic Meteorology. Deutscher Wet- or by many points for approaches based on terdienst, Offenbach am Main, Germany 200 pp. PV, as the PV approach accounts for an area 5. Hoskins, B.J., M.E. McIntire & A.W. Robertson. surrounded by a PV contour. That there is a 1985. On the use and significance of isentropic poten- different counting methodology does not have tial vorticity maps. Q. J. R. Meteor. Soc. 111: 877–946. 6. Wernli, H. & M. Sprenger. 2007. Identification and an influence on the main aspects of the clima- ERA-15 climatology of potential vorticity streamers tology, such as the geographical distribution of and cutoffs near the extratropical tropopause. J. At- COLs and their seasonal and monthly variabil- mos. Sci. 64: 1569–1586. ity. In summary, the comparison between the 7. Schwierz, C., S. Dirren & H.C. Davies. 2004. Forced climatologies reinforces the view that the con- waves on a zonally aligned jet stream. J. Atmos. Sci. 61: 73–87. ceptual model is well represented by the ob- 8. Appenzeller, C., H.C. Davies & W.A. Norton. 1996. jective analysis technique developed by Nieto Fragmentation of stratospheric intrusions. J. Geophys. et al.26 and that the climatology based on IPV Res. 101: 1435–1456. analysis by Wernli and Sprenger6 is in fairly 9. Bamber, D.J., P.G.W. Healey, B.M.R. Jones, et al. close agreement. The two methodologies are 1984. Vertical profiles of tropospheric gases: chem- complementary and each contributes to a bet- ical consequences of stratospheric intrusions. Atmos. Environ. 18: 1759–1766. ter characterization of the physical properties 10. Holton, J., P. Haynes, M. McIntyre, et al. 1995. of COLs. Stratosphere–troposphere exchange. Rev. Geophys. 33: 403–439. 11. Sprenger, M., H. Wernli & M. Bourqui. 2007. Acknowledgments Stratosphere–troposphere exchange and its relation to potential vorticity streamers and cutoffs near the This work was supported by the Span- extratropical tropopause. J. Atmos. Sci. 64: 1587–1602 ish Ministry of Education and Science under (doi:10.1175/JAS3911.1). Grant CGL2007-65891-C05-01. R.N. ac- 12. Oltmans, S.J., H. Levy II, J.M. Harris, et al. 1996. knowledges the Portuguese Ministry of Sci- Summer and spring ozone profiles over the North ence through the Grant SFRH/BPD/22178/ Atlantic from ozonesonde measurements. J. Geophys. 2005 by the FCT. The authors thank the Res. 101: 29179–29200. 13. Gimeno, L., E. Hernandez,´ A. Rua´ & R. Garc´ıa. anonymous reviewer for the helpful comments 1998. Surface ozone in Spain. Chemosphere 38: 3061– and suggestions. M. S. and H. W. thank 3074. MeteoSwiss for providing access to the 14. Cuevas, E. & J. Rodr´ıguez. 2002. Statistics of cut- ECMWF reanalysis data. off lows over the North Atlantic (in Spanish). Proc. Third Asamblea Hispano- Portuguesa de Geodesia yGeof´ısica, Comision Espanola˜ de Geodesia y Ge- Conflicts of Interest ofisica, Valencia, Spain, pp. 1–3. 15. Kentarchos, A.S., G.J. Roelofs, J. Lelieveld & E. The authors declare no conflicts of interest. Cuevas. 2000. On the origin of elevated surface ozone Nieto et al.: Identification and Climatology of Cut-off Lows near the Tropopause 289

concentrations at Izana˜ Observatory during the last tems over Europe using Meteosat imaginery. Meteorol. days of March 1996: a model study. Geophys. Res. Lett. Atmos. Phys. 96: 141–157 (doi:10.1007/s00703-006- 27: 3699–3702. 0225-4. 2007). 16. Wirth, V.1995. Diabatic heating in an axisymmetric 31. Huber-Pock, F. & C. Kress. 1981. Contributions cutoff cyclone and related stratosphere–troposphere to the problem of numerical frontal analysis. Pro- exchange. Quart. J. Roy. Meteor. Soc. 121: 127–147. ceedings of the Symposium on Current Problems of 17. Wirth, V.& J. Egger. 1999. Diagnosing extratropical Weather-Prediction. Vienna, June 23–26, 1981. Pub- synoptic-scale stratosphere-troposphere exchange: a lications of the Zentralanstalt fur¨ Meteorologie und case study. Quart. J. Roy. Meteor. Soc. 125: 635–656. Geodynamik, p. 253. 18. Price, J.D. & G. Vaughan. 1993. The potential for 32. Huber-Pock, F. & C. Kress. 1989. An opera- stratosphere–troposphere exchange in cut-off low tional model of objective frontal analysis based on systems. Quart. J. Roy. Meteor. Soc. 119: 343–365. ECMWF products. Meteorol. Atmos. Phys. 40: 170– 19. Bourqui, M.S. 2006. Stratosphere-troposphere ex- 180. change from the Lagrangian perspective: a case study 33. Sabo, P. 1992. Application of the thermal front pa- and method sensitivities. Atmos. Chem. Phys. 6: 2651– rameter to baroclinic zones around cutoff lows. Me- 2670. teorol. Atmos. Phys. 47: 107–115. 20. Gouget, H., G. Vaughan, A. Marenco & H.G.J. Smit. 34. Renard, R.J. & L.C. Clarke. 1965. Experiments in 2000. Decay of a cut-off low and contribution to numerical objective frontal analysis. Mon. Wea. Rev. stratospheretroposphere exchange. Quart. J. Roy. Me- 93: 547–556. teor. Soc. 126: 1117–1141. 35. Zwatz-Meise, V. & F. Hufnagl. 1990. Some results 21. Knippertz, P. & J.E. Martin. 2005. Tropical plumes about the relation between an objective front param- and extreme precipitation in subtropical and tropical eter and cloud bands in satellite images and its con- West Africa. Quart. J.Roy.Meteor.Soc. 131: 2337–2365. nection to classical cold front models. Meteorol. Atmos. 22. Garc´ıa-Herrera, R., D.G. Puyol, E.H. Mart´ın, et al. Phys. 42: 77–89. 2001. Influence of the North Atlantic Oscillation on 36. Nieto, R., L. Gimeno, J.A. Anel,˜ et al. 2007. Analysis the Canary Islands precipitation. J. Climate 14: 3889– of the precipitation and cloudiness associated with 3903. COLs occurrence in the Iberian Peninsula. Meteorol. 23. Tripoli, G.J., C.M. Medaglia, S.M. Dietrich, et al. Atmos. Phys. 96: 103–119 (doi:10.1007/s00703-006- 2005. The 9–10 November 2001 Algerian flood: a 0223-6). numerical study. Bull. Amer. Meteor. Soc. 86: 1229– 37. Jansa,A.,A.Genoves&J.A.Garc´ ´ıa-Moya. 2000. 1235. Western Mediterranean cyclones and heavy rain. 24. Kotroni, V., K. Lagouvardos, E. Defer, et al. 2006. Part I: numerical experiment concerning the Pied- The Antalya 5 December 2002storm: observations mont flood case. Met. Apps. 7: 323–333. and model analysis. J. Appl. Meteor. Clim. 45: 576– 38. Hewson, T.D. 1998. Objective fronts Meteorol. Appl. 590. 5: 37–65. 25. Porcu,´ F., A. Carrassi, C.M. Medaglia, et al. 2007. 39. Ebel, A., H. Hass, H.J. Jakobs, et al. 1991. Simulation A study on cut-off low vertical structure and precip- of ozone intrusion caused by tropopause fold and itation in the Mediterranean region. Meteorol. Atmos. COL. Atmos. Environ. 25A: 2131–2144. Phys. 96: 121–140 (doi:10.1007/s00703-006-0224- 40. Langford, A., C. Masters, M. Proffitt, et al. 1996. 5). Ozone measurements in a tropopause fold associated 26. Nieto, R., L. Gimeno, L. de la Torre, et al. 2005. with a cut-off low system. Geophys. Res. Lett. 23: 2501– Climatologies features of cut-off low systems in the 2504 (and correction in Geophys Res Lett 24: 109– Northern Hemisphere. J. Climate 18: 3085–3103. 110). 27. Matsumoto, S.K., K. Ninomiya, R. Hasegawa & Y. 41. Qi, L., Y. Wang & L.M. Leslie. 2000. Numerical Miki. 1982. The structure and role of a subsynoptic simulations of a cut-off low over southern Australia. cold vortex on the heavy precipitation. J. Meteor. Soc. Meteorol. Atmos. Phys. 74: 103–115. Japan 60: 339–353. 42. Qi, L. & L.M. Leslie. 2001. Cut-off low pressure sys- 28. Hill, E.F. & K.A. Browning. 1987. Case study of a tems over southern Australia: a numerical modeling persistent mesoscale cold pool. Meteor.Mag. 116: 297– study and sensitivity experiments. Aust. Meteor. Mag. 309. 50: 183–194. 29. Keyser, D. & M.A. Shapiro. 1986. A review of the 43. Buckley, L., M. Leslie, W.Sullivan, et al. 2007. A rare structure and dynamics of upper level frontal zones. East Indian Ocean autumn season tropical cut-off Mon. Wea. Rev. 114: 452–499. low: impacts and a high-resolution modelling study. 30. Delgado, G., A. Redano,J.Lorente,˜ et al. 2007. Cloud Meteorol. Atmos. Phys. 96: 61–84 (doi:10.1007/s00703- cover analysis associated to cut-off low pressure sys- 006-0221-8). 290 Annals of the New York Academy of Sciences

44. Ancellet, G., M. Beekmann & A. Papayannis. 1994. 58. Smith, B.A., L.F. Bosart, D. Keyser & D. St. Jean. Impact of a cut-off lows development transport of 2002. A global 500hPa cutoff cyclone climatology: ozone in the free troposphere. J. Geophys. Res. 99: 1953–1999. Preprints, 19th Conf. on Weather Anal- 3451–3468. ysis and Forecasting, Amer. Meteor. Soc., San Anto- 45. Barsby, J. & R.D. Diab. 1995. Total ozone and syn- nio, TX, pp. 1–14. optic weather relationships over southern Africa and 59. Kalnay, E. et al. 1996. The NCEP/NCAR 40-Year surrounding oceans. J. Geophys. Res. 100: 3023–3032. Reanalysis Project. Bull. Amer. Meteor. Soc. 77: 437– 46. Kentarchos, A.S., T.D. Davies & C.S.A. Zerefos. 471. 1998. Low latitude stratospheric intrusion associated 60. Uppala, S., P. Kallberg, A. Hernandez,´ et al. with a cut-off low. Geophys. Res. Lett. 25: 67–70. 2004. ERA-40: ECMWF 45-year reanalysis of the 47. Kentarchos, A.S., G.R. Roelofs & J. Lelieveld. 1999. global atmosphere and surface conditions 1957– Model study of a stratospheric intrusion event at 2002. ECMWF Newsletter 101: 2–21. lower midlatitude associated with the development 61. Fuenzalida, H.A., R. Sanchez´ & R.D. Garreaud. of a cut-off low. J. Geoph. Res. 104: 1717–1727. 2005. A climatology of cutoff lows in the South- 48. Baray, J.L., S. Baldy, R.D. Diab & J.P.Cammas. 2003. ern Hemisphere. J. Geoph. Res. 110: D18101 Dynamical study of a tropical cut-off low over South (doi:10.1029/2005JD005934). Africa and its impact on tropospheric ozone. Atmos. 62. Pizarro, J. & A. Montecinos. 2000. Cut-off cyclones Environ. 37: 1475–1488. off the subtropical coast of Chile. Preprints 6th Int. 49. Ravetta, F. & G. Ancellet. 2000. Identification of dy- Conf. on Southern Hemisphere Meteorology and namical processes at the tropopause during the de- Oceanography, pp 278–279. cay of a cut-off low using high resolution airbone li- 63. Campetella, C.M. & N.E. Possia. 2007. Upper-level dar ozone measurements. Mon. Wea. Rev. 128: 3252– cut-off lows in southern South America. Meteorol. 3267. Atmos. Phys. 96: 181–191 (doi:10.1007/s00703-006- 50. Griffiths, M., M.J. Reeder, D.J. Low & Ra. A. 0227-2). Vincent. 1998. Observation of a cut-off low over 64. Nieto, R., L. Gimeno, L. de la Torre, et al. 2007. In- southern Australia. Quart. J. Roy. Meteor. Soc. 124: terannual variability of cut-off low systems over the 1109–1132. European sector: The role of blocking and the North- 51. Price, J.D. & G. Vaughan. 1992. Statistical studies of ern Hemisphere circulation modes. Meteorol. Atmos. cutoff-low systems. Ann. Geophys. 10: 96–102. Phys. 96: 85–101 (doi:10.1007/s00703-006-0222-7). 52. Kentarchos, A.S. & T.D. Davies. 1998. A climatol- 65. Croci-Maspoli, M., C. Schwierz & H.C. Daviesa. ogy of cutoff lows at 200 hPa in the Northern Hemi- 2007. A multifaceted climatology of atmospheric sphere, 1990–1994. Int. J. Climatol. 18: 379–390. blocking and its recent linear trend. J. Climate 20: 53. Parker, S.S., J.T. Hawes, S.J. Colucci & B.P. Hayden. 633–649. 1989. Climatology of 500 mb cyclones and anticy- 66. Koch, P., H. Wernli & H.C. Davies. 2003. An event- clones 1950–85. Mon. Wea. Rev. 117: 558–570. based jet-stream climatology and typology. Int. J. Cli- 54. Bell, G.D. & L.F.Bosart. 1989. A 15-year climatology matol. 26: 283–301. of Northern Hemisphere 500 mb closed cyclone and 67. Hurrell, J.W. 1995. Decadal trends in the North At- anticyclone centers. Mon. Wea. Rev. 117: 2142–2163. lantic Oscillation: regional temperatures and precip- 55. Novak, M.J., L.F. Bosart, D. Keyser, et al. 2002. Cli- itation. Science 269: 676–679. matology of warm season 500 hPa cutoff cyclones and 68. Trenberth, K.E., G.W. Branstator, D. Karoly, et al. a case study diagnosis of 14–17 July 2000. Preprints, 1998. Progress during TOGA in understanding and 19th Conf. on Weather Analysis and Forecasting, modelling global teleconnections associated with Amer. Meteor. Soc., San Antonio, TX, pp. 68–71. tropical sea surface temperatures. J. Geophys. Res. 56. Hernandez,´ A. 1999. Un Estudio Estad´ıstico sobre 103(C7): 14291–14324. Depresiones Aisladas en Niveles Altos (DANAs) en el 69. Waugh, D.W. & P.P. Rong. 2002. Interannual vari- Sudoeste de Europa basado en Mapas Isentropicos´ ability in the decay of lower stratospheric arctic vor- de Vorticidad Potencial. IV Simposio Nacional de tices. J. Meteor. Soc. Japan. 80: 997–1012. Prediccion,´ Instituto Nacional de Meteorolog´ıa, Se- 70. Black, R.X., B.A. McDaniel & W.A.Robinson. 2006. rie Monogr., No. SM 351, Ministerio de Medio Am- Stratosphere-troposphere coupling during spring on- biente, 235 pp. set. J. Climate 19: 4891–4901. 57. Qi, L., L.M. Leslie & S.X. Zhao. 1999. Cut-off 71. Wernli, H. & H.C. Davies. 1997. A Lagrangian- low pressure systems over southern Australia: cli- based analysis of extratropical. I: The method and matology and case study. Int. J. Climatol. 19: 1633– some applications. Quart. J. Roy. Meteor. Soc. 123: 467– 1649. 489.