1. Introduction

In the context of global climate change, extremeweather and climate events show an obvious increasein both frequency and intensity. As a result, bothmeteorological and geological disasters, such as flooding and mudslides, which are closely associated withextreme rainfall, are causing serious damage to human lives and properties(IPCC, 2007). The low-leveljet(LLJ)provides favorable background circulation and abundant water vapor for extreme rainfall events, and has therefore attracted widespread attention inthe research community since the 1930s, with scientists having carried out a large amount of studies onLLJ since the 1950s in particular. The LLJ is widelydistributed across each continent of the earth, suchas North and South America, Africa, Asia, Oceania, and Antarctica(Stensrud, 1996). Goualt(1938) and Farquharson(1939)were among the first to discoverAfrican LLJ events in the late 1930s. Means(1952)used the concept of LLJ in his work on thunderstormsover the central area of the United States, and foundthat the LLJ plays an important role in squall lines and other extreme weather events. Blackadar(1957), Bonner(1968), and Uccellini and Johnson(1979)allcarried out extensive research on LLJ formation and evolution, LLJ distribution and structure, as well asthe relationships between LLJ and other weather and climate phenomena over North America. Similarly, Findlater(1969, 1977)completed a series of observational analyses on the characteristics of the Somali jet and its relationship with summer monsoonal rainfallin western India. Meanwhile, since the 1980s, Chinese scientists have focused on analyzing the relationship between LLJ and rainstorms, and their findingshave provided useful reference for heavy rainfall prediction in South China(e. g. , Tao et al. , 1980; Yu et al. , 1983).

Generally, the LLJ is regarded as a fastmovingribbon of air with wind speed greater than 12 m s^-1in the boundary layer or the lower troposphere(usually below 700 hPa). Conventional sounding observations only include twicedaily measurements, and theirspatial distribution is uneven. Therefore, it is hardto capture all LLJ events and obtain detailed information on LLJ evolutionary processes, especially forstronger LLJs(Mitchell et al. , 1995). In view of thisfact, most LLJ studies are based on rawinsondes. Forexample, with 6-hourly rawinsonde and hourly windprofiler observations, Bonner(1968) and Mitchell et al. (1995)each presented the geographical distribution ofthe Great Plains LLJ over North America, its seasonalactivities, horizontal and vertical structures, and diurnal variation characteristics, and were able to concludethe climatology of the warm season Great Plains LLJ. Arritt et al. (1997)investigated anomalous LLJ activities and their contributions to the 1993 summer floodin America through high-resolution hourly wind observations from the National Oceanic and AtmosphericAdministration(NOAA)Profiler Network. In order tostudy the environmental configuration of the LLJ, sci-entists have used regular weather observations(Chen and Yu, 1988). However, these observations rarelymatch each other in terms of observation time and location, so errors will inevitably arise. In addition, single-point rawinsonde measurements cannot capturehorizontal shear and environmental conditions(Chen et al. , 2005). High-resolution reanalysis data providepowerful support for detailed LLJ analysis. For example, using such data, Higgins et al. (1997) and Weaver and Nigam(2008)respectively examined the effects ofthe Great Plains LLJ on summer rainfall and water vapor transport, and the influence of the North AtlanticOscillation and El Ni~no{Southern Oscillation on seasonal and annual variations of the LLJ and the subse-quent adjustments of regional hydroclimate. Further-more, a vast field campaign, the South America LLJExperiment", was carried out in the period 2002{2003aimed at underst and ing the role of the LLJ in moisture and energy exchange between the tropics and extratropics and related aspects of regional hydrology, climate, and climate variability(Vera et al. , 2006).

During the past half-century, LLJ studies havecovered many aspects, including the structural and evolutionary characteristics(Bonner, 1968; Findlater, 1969; Chen et al. , 2005), formation and developmentmechanisms(Blackadar, 1957; Holton, 1967; Uccellini and Johnson, 1979; Pan et al. , 2004), and LLJ relationship with rainfall and other weather and climate phenomena(Tao et al. , 1979; Higgins et al. , 1997; Saulo et al. , 2007). The research methods have evolved fromobservational analysis at the very beginning, to theoretical studies, and more recently to numerical experimental investigations(Holton, 1967; Mitchell et al. , 1995; Ting and Wang, 2006). The mesoscale numerical model is the most widely used tool, as it cannot only provide high spatial and temporal resolutiondata for LLJ research, but also examine the alreadyexisting LLJ formation theories and further explorethe possible influencing mechanisms through numerical sensitivity experiments. For example, Pan et al. (2004) and Zhang et al. (2006)conducted studies concerning the influence of topography, l and surface heatflux, and l and sea contrast on the formation and evolution of the Rocky Mountains and Appalachian Mountains LLJs with mesoscale models so as to quantifythe relative roles of the above factors. In addition, Qian et al. (2004) and Zhao(2012)respectively analyzed the interactions between precipitation-relatedlatent heat release and LLJ development, and the local topographical effects on the LLJ during the Meiyuseason in South China.

The relationship between LLJ and strong precipitation has been intensively studied during the pastseveral decades. The findings have been effective inincreasing the precision of rainfall prediction(Tao et al. , 1980), and enhanced our underst and ing of extremerainfall and flooding events, which provides the scientic basis for the prediction and prevention of suchdisastrous events. In addition to precipitation, theLLJ is closely related to air pollution, wind energyutilization, aviation safety, s and storm transport, and forest fire(Uccellini, 1980; Liechti and Schaller, 1999;Archer and Jacobson, 2005). As these relationshipshave come to light, more and more attention has beenpaid to corresponding studies in these fields by natural and social scientists. Hence, LLJ research has boththeoretical and practical meanings.

At present, information regarding the LLJ overmainl and China is inadequate, and our knowledge onLLJ-related events is limited(Sai and Miao, 2012). Meanwhile, most Chinese scientists have focused onthe relationship between LLJ and precipitation, and the lack of observations or di±culties in data sharinghave resulted in relatively few studies on LLJ structural and evolutionary mechanisms. Therefore, manymore studies are still needed to underst and this important weather phenomenon. In fact, in recent years, theamount of observations has increased, which providesan ever-improving and favorable platform for LLJ research. However, before we embark on such research, adetailed review of LLJ-related studies from both home and abroad is useful. Accordingly, in the present paper, we focus on reviewing three major aspects of LLJresearch: LLJ classification, definition, distribution, and structure; LLJ formation and evolutionary mechanisms; and the relationships between LLJ and rain-fall as well as other interdisciplinary fields. Finally, we discuss the shortcomings of LLJ studies in China and speculate on the future prospects of several LLJresearch avenues.

2. LLJ classification, definition, distribution, and structure2. 1 LLJ classification and definition

Based on the maximum wind speed height, theLLJ can be divided into the free atmosphere LLJ(850{600 hPa) and the boundary layer LLJ(below 850 hPaor 1500 m)(Sai and Miao, 2012). Free atmosphereLLJs always appear with synoptic systems, e. g. , thestrong baroclinic LLJ at the edges of the tropical cyclone or the southwest vortex east of the Sichuan basin, and the LLJ in local circulation resulting from strongconvection. Boundary layer LLJs include both theless-baroclinic jets without obvious synoptic systems, and those jets accompanying rainstorms over the transition zone between high and low pressure systems, such as the LLJ east of the Rocky Mountains and theSomali jet in East Africa. Of course, according to thecoverage area of strong winds, the LLJ can also bedivided into large-scale, synoptic-scale, and mesoscalejets(Yu, 1986). We do not discuss each type of LLJin this paper, but review their general characteristics and formation mechanisms.

So far, there is no universally accepted LLJ definition because of the differences in LLJ height, coveragearea, maximum wind speed, horizontal and verticalshears, and so on. Bonner(1968)set three criteriafor an LLJ based on the maximum wind speed and vertical shear. Criterion 1(2, 3)is that the wind atthe level of maximum wind must equal to or exceed12(16, 20)m s^-1 and must decrease by at least 6(8, 10)m s^-1 to the next highest level with minimumwind or to the 3-km level, whichever is lower. Thisdefinition is extensively accepted in both North and South American LLJ research(Higgins et al. , 1997;Pan et al. , 2004; Vera et al. , 2006). Besides, scientists also set different limitations for LLJ maximumwind speed and its height, and vertical shear, basedon topographical distribution and background circulation conditions. For example, the vertical jetlikestructures in the lowest 1. 5 km were identified withpositive shear below and negative shear above at 300m height intervals in Zhang et al. (2006). Such a broaddefinition of LLJ is necessary because a single stationis unable to frequently capture the core of LLJs.

Following the LLJ criteria of Bonner(1968), relaxed the height of maximum windspeed up to 600 hPa. With this definition, the sounding data in their study showed a double-peak structure, with a primary maximum at 900{925 hPa and a secondary maximum at 825{850 hPa. Over Taiwan, the primary LLJ wind speed maximum correspondsto barrier jets within the boundary layer due to terrain blocking, which has a higher frequency. The 850{700-hPa jets are movable and closely linked to Meiyu-frontal heavy rainfall. Du et al. (2012)also foundsimilar double-peak LLJs at Qingpu station(Shanghai)from the wind profiler radar data, which correspond to the 500{800-m boundary layer jet and 2100{2200-m synoptic system related LLJs. Before this, most studies on mainl and China LLJs only set a limitation on the maximum wind speed at a certain pressure level because of the lower-resolution data and norestriction on vertical shears(e. g. , Yu et al. , 1983; Zhai et al. , 1999; Xu et al. , 2001; Qian et al. , 2004).

Stensrud(1996)made a distinction between theLLJ and the low-level jet stream. He pointed out thatthe LLJ represents the air currents with obvious vertical shear, whereas no limit is given for the low-leveljet stream(strong vertical shear is not necessary forit to exist), and this classification is accepted by mostresearchers. Zhang et al. (2007)compared the southwest LLJ and low-level southwesterly maximum windover China and revealed their synoptic and climatological differences. Most LLJ research from abroadsets a limit on vertical shear to avoid confusion withlow-level high-speed streams. The LLJ and low-leveljet stream do show obvious differences in their distribution, horizontal and vertical structure, diurnal variations, formation mechanisms, as well as their effectson rainstorms. In this paper, unless otherwise stated, the LLJ refers to high-speed air currents with obvious vertical shear in both the boundary layer and thelower troposphere.

2. 2 LLJ distribution and activity

The LLJ is widely spread over every continent ofthe earth, but tends to locate to the east of large mountains or in areas with l and -sea contrast. It always liesalong topographic or coastal lines. Stensrud(1996)prepared a h and -drawn map of LLJ distributions overthe earth based on previously published data. Recently, Rife et al. (2010)produced a map of the globaldistribution of diurnally varying LLJs based on 40-kmhigh-resolution hourly reanalysis data(Fig. 1). Allthe known nocturnal LLJs of Stensrud(1996)appearclearly in Fig. 1, and several newly identified areas ofintense LLJ activity are also present, such as over theTibet and the Tarim basin in China. The intensityof the LLJ in the Northern Hemisphere is in generalstronger than that in the Southern Hemisphere due toa larger continental area.

Fig. 1. Mean nocturnal LLJ(NLLJ)index(shaded) and 500-m above-ground-level(AGL)winds(arrows)at localmidnight in the period 1985{2005 for(a)July and (b)January, calculated from CFDDA hourly analyses. The insetshows the locations of NLLJs analyzed for the study . This figure is taken from Rife et al. (2010).

Figure options

LLJs occur throughout the year, but maintaina higher frequency during the warm season in eachhemisphere, with greater intensity and coverage area(Findlater, 1977; Virji, 1981; Paegle et al. , 1987). Using North America as an example, Weaver and Nigam(2008)found that the Great Plains LLJs are mostactive and strong in July according to 25-yr highresolution reanalysis data. The LLJs over SouthwestChina show similar activities(Zhang et al. , 2007). Meanwhile, LLJ activities are regulated by the largescale climate background, showing distinctive annualvariations(Weaver and Nigam, 2008). For example, the North American LLJ was found to have occurredmore frequently during the period 2000{2002 than1997{1999 under the simultaneous modulation of ElNino and the Pacific Decadal Oscillation(Song et al. , 2005).

2. 3 LLJ structure

A significant jet core with large horizontal shearalways exists at the horizontal level of the LLJ, whichruns parallel to the topography or coastline. In NorthAmerica, the LLJ mainly appears over the GreatPlains to the east of the Rocky Mountains. Synopticscale low pressure systems always exist to the left ofthe LLJ and large-scale high pressure systems providefavorable background conditions to its right. The pressure gradient between high- and low-pressure systemsleads to consistent increase of air currents(Fig. 2a and 2b of Rife et al. , 2010). The LLJs to the east of theTibetan Plateau, Africa, and the Andes Mountains inSouth America all show similar horizontal structure and synoptic configuration(Findlater, 1977; Vera et al. , 2006; Zhang et al. , 2007; Liu, 2012; Liu et al. , 2012). Bonner(1968)conducted a systematic studyof the LLJ horizontal structure with sounding databased on the LLJ coordinate system. His results indicated that divergence exists in the upstream side and convergence in the downstream side of LLJs. Thus, air stream should be sinking as it moves into the jetmaximum, and then rising downstream from the jet, leading to an increased likelihood of nocturnal thunderstorms in the downstream section of the jet. InChina, the LLJs south of the Yangtze River basin showsimilar structures; namely, highly warm and moist rising air stream and intense convergence to the left sideof the jet, and always accompanied by rainstorms(Tao et al. , 1980).

Fig. 2. Composite characteristics for strong(≥ 90th percentile)NLLJ events for the Great Plains region(site 1 in Fig. 1)for July 1985{2005. (a)Terrain elevation within the region(meters MSL) and (b)mean NLLJ index(shaded) and 500-m AGL winds(arrows). The thick white line denotes the location of the cross-section shown in(c), and the whitecircle denotes the point at or near the jet core, and marks the location of the time-height plot shown in(d). (c)Crosssection of the mean wind speed along the white line in(b). (d)Mean time-height evolution of wind speed within the jetcore, denoted by the white circle in(b). This figure is taken from Rife et al. (2010).

Figure options

In the vertical direction, the wind profile has typical jet-like characteristics, with maximum wind and strong vertical shear. We take North America as anexample: the wind maximum(> 16 m s^-1)appearsbetween 250 and 1000 m and then quickly decreasesto 6 m s^-1 at 3000 m(Fig. 2c). The above verticalstructure is continuously found in areas with frequentLLJ activity(Findlater, 1977). As mentioned earlier, Chen et al. (2005) and Du et al. (2012)found doublepeak LLJs in Taiwan and Shanghai, corresponding toboundary layer and synoptic-scale jets. Because ofthe close relationship with synoptic system forcing, the synoptic-scale LLJ appears much higher(850{800hPa). In fact, the two kinds of jets are ubiquitous(Stensrud, 1996). Tao et al. (1980)pointed out thatdifferent scale LLJs are not isolated; most studies showonly one jet peak because they choose different LLJcriteria.

2. 4 Diurnal variation

In the early 1930s, Goualt(1938) and Farquharson(1939)noticed nocturnal intense wind events; and then Gifford(1952), Lettau(1954), and Blackadar(1955)further proved the existence of this phenomenon. The United States of America first carriedout a field experiment aimed at boundary layer winds, and Hoecker(1963)demonstrated the detailed diurnal variation of the LLJ with hourly observation data. The wind speed increased and reached its maximumat 0500 CST(Central St and ard Time), and then itweakened after sunrise with the well-organized jet coreseparating into several centers. After sunset, the windstrengthened and recovered into a complete jet coreat 2300 CST(Hoecker, 1963). Diurnal variation characteristics of LLJs are widely observed in areas withfrequent LLJ activity. Although differences exist inthe LLJ onset time, maximum wind speed, and jet coreheight over different areas, the wind speeds on averageall reach their maximum at midnight to early morning and minimum around noon time(Fig. 2d and 3a). In addition, the LLJ maximum wind presents adistinct clockwise rotation during the whole day(Fig. 3). In view of this diurnal cycle, Blackadar(1957) and Holton(1967)each proposed explanations based oninertial oscillation theory and attributed it to upslope and downslope winds caused by the day- and -night radiation difference. Furthermore, Bonner and Paegle(1970)found that the veering of thermal winds due tothe differences between topographic and atmosphericthermal characteristics can also affect the diurnal variation of the LLJ. Related research is reviewed in detailin Section 3.

Fig. 3. (a)Time-height cross-section of horizontal wind and isotach(dotted; every 2 m s-1)from the wind profilerobservations at Fort Meade for the period 1200 LST(Local St and ard Time)19 June{1200 LST 20 June 2001. Shadingdenotes the layers of horizontal winds exceeding 10 m s-1. A full barb is 5 m s-1. (b)Hodograph at hourly intervals taken at 500 m AGL for the period 2100 LST 19 June{2000 LST 20 June 2001. The arrow denotes horizontal windvectors given near sunset and at the time of peak magnitude. This figure is taken from Zhang et al. (2006).

Figure options

Meteorological interest in the diurnal cycle of theLLJ lies in the fact that it is closely associated withthe diurnal variation of precipitation. When the LLJstrengthens during nighttime, vertical shear increases and the wind becomes supergeostrophic. This leadsto a less stable air column and is favorable for the development of convection. As a result, thunderstorms and rainstorms tend to develop or strengthen duringnighttime(Tao et al. , 1980). Higgins et al. (1997)pointed out that LLJ events are associated with enhanced precipitation over the northern central UnitedStates and the Great Plains and decreased precipitation along the Gulf Coast and East Coast. Becauseof the LLJ, in excess of 25% more precipitation fallsover the Great Plains during the nighttime hours th and uring daytime hours in the warm season, and theoverall moisture budget is considerable, with low-levelinflow from the Gulf of Mexico increasing on averageby more than 45% over nocturnal mean values. Withthe help of numerical simulation results, Liu(2012)found that the rainfall over the Yangtze-Huai Riverbasin has double peaks; namely, rainfall is strongest inearly morning and early evening, but weakest at noon and midnight. Meanwhile, the LLJ shows a diurnalcycle with intensification in the morning and weakening or disappearance around noon. The convergenceresulting from the clockwise rotation of wind direction during this diurnal variation provides a reasonableexplanation for the early morning rainfall peak overthis region. The minimum rainfall downstream of theYangtze River basin is mainly caused by the eastwardwarm advection from the Tibetan Plateau, which inhibits local convection after noon(Chen, 2009). Xu and Chen(2013)also indicated that the LLJ forms 12h prior to the occurrence of nocturnal rainstorms innorthern Zhejiang Province.

3. Mechanisms of LLJ formation and evolution

Scientists have carried out a large amount of workon the above-described LLJ structural and evolutionary characteristics. In this section, we summarize fivekey aspects of the mechanisms of LLJ formation and evolution.

3. 1 Inertial oscillation

Turbulent motion and boundary layer depth exhibit distinct diurnal variation as the solar radiationvaries throughout the day. Vertical turbulent mixing weakens after sunset and a shallow inversion layerforms above the surface. The horizontal wind decouples from the l and surface within the residual layer;that is, the wind is no longer affected by the surface friction. Therefore, the horizontal wind, whichis subgeostrophic, will evolve into being geostrophic. Because of the combined effects between Coriolis force and inertial centrifugal force, an inertial oscillation appears under the interactions of these two forces with aperiodicity of 17 h(Blackadar, 1957; Stensrud, 1996;Hao et al. , 2001). Meanwhile, the wind field endures a transition from subgeostrophic to quasigeostrophic and finally supergeostrophic. Blackadar(1957)showedthat inertial oscillation is an important factor for theformation of the LLJ. Later, Bonner and Paegle(1970)verified this viewpoint through a simple conceptualmodel.

Van De Wiel et al. (2010)extended Blackadar'sconcept of nocturnal inertial oscillation by includingfrictional effects within the nocturnal boundary layer. Figure 4 shows a schematic illustration of the windvelocity inertial oscillation profiles with different assumptions. Initially, the wind velocity gradually increases with height(Fig. 4a). Under the assumptionof frictionless effects in the boundary layer, the windvelocity will maintain an inertial oscillation aroundthe geostrophic wind vector based on the boundarylayer equations for the mean average wind components, which is shown by the inertial circles(dashedlines in Fig. 4a). The envelope marked with an"×" shows the vertical profile of whole wind velocitywith different oscillation radii at the moment of thestrongest LLJ. The initial velocity profile in Fig. 4bis exactly the same as that in Fig. 4a; however, thenocturnal wind speed profile describes an oscillationaround the nocturnal equilibrium wind vector ratherthan around the geostrophic wind vector, when including frictional effects. In comparison with Blackadar'sconcept model, the velocity profile in Van De Wiel etal. (2010)is more similar to observations, having atypical jet-profile and intense vertical shear. Inertialoscillation is one of the most accepted mechanisms forthe explanation of boundary layer LLJ formation withobvious diurnal variation.

Fig. 4. Schematic illustration explaining(a)Blackadar's(1957)inertial oscillation around the geostrophic equilibrium; and (b)as in(a), but with the nocturnal wind oscillating around the nocturnal equilibrium profile. M represents the wind speed(m s-1). This figure is taken from Van De Wiel et al. (2010).

Figure options

3. 2 Topographic thermal and dynamic forcing

The diurnal oscillation of horizontal temperature gradients over sloping terrain and the subsequent geopotential height variation induced upslope and downslope motions can also account for theLLJ formation. Solar radiation is strong after noon, and therefore the negative temperature gradient frommountains to valleys leads to the reverse in terms ofthe pressure gradient. This induces upslope winds onthe eastern side of sloping terrain. The air streamgradually veers with the influence of Coriolis force and inertial oscillation, and finally shows an obvious southerly component after several hours(in nighttime). This strengthens the previously existing southwesterly wind and leads to the formation of the LLJ. The above activities are reversed at nighttime. Thedownslope wind resulting from the opposite temperature gradient veers and northerly wind forms during daytime, which heavily counteracts the prevailingsouthwesterly wind and finally makes the LLJ disappear(Holton, 1967; Jiang et al. , 2007). Meanwhile, therotation of thermal wind due to the thermal contrastover the sloping terrain also promotes the formation and disappearance of the LLJ(Bonner and Paegle, 1970). Under the background of thermal differencesdue to terrain, most LLJs over the east of the TibetanPlateau and Rocky Mountains form as the gradientwind and thermal wind adjust with the change of surface temperature(Stensrud, 1996; Sai and Miao, 2012)(Note that the mechanisms of l and -sea thermal contrasts are similar to those over sloping terrain, and therefore we do not explain them in this paper). Inaddition, the positive feedback between the inversionlayer and intense vertical shear is another importantfactor for LLJ formation and development. Duringnighttime, radiative cooling is significant over the surface layer of the terrain, and the inversion forms whenthe air over the terrain is advected over the valley(Hao et al. , 2001). This stable layer structure inhibits thedevelopment of turbulence within the boundary layer and favors the retention of momentum. When the top of the inversion layer coincides with the maximumwind peak, turbulence can exist beneath and help theinversion layer to sustain and develop upward. Hence, the existence of the inversion layer provides favorableconditions for an increase of wind speed within theboundary layer. Conversely, the strong vertical shearhelps to sustain and develop the inversion layer. Thepositive feedback between them promotes LLJ formation and enhancement(Blackadar, 1957; He and Wu, 1989).

Topographic dynamic blocking effects cannot beneglected either. During the warm season in NorthAmerica, the easterly wind of the Atlantic subtropicalhigh can reach the eastern side of the Rocky Mountains, and then the air current veers northward whenit climbs up the mountain. Under the control of potential vorticity conservation, the air mass graduallyspeeds up as it travels northward, and then the LLJforms(Wexler, 1961). This kind of LLJ is regarded asa "barrier jet", which is closely associated with largescale background circulations. Chen et al. (2005)revealed several characteristics of the barrier jet, such asits appearance at relatively lower levels; being withoutor with weak vertical shear, and less movable; and possessing no obvious diurnal variation. Therefore, thebarrier jet can be ascribed into "low-level high-speedair streams". In addition, sloping terrain can significantly reduce air mass transport when the air currentmoves across isobars, which ensures the persistenceof LLJ events(Holton, 1967). In fact, the high and low pressure system configurations due to heating differences along sloping terrain and the "narrow pipeeffect" both help the formation and development ofLLJs(Pamperin and Stilke, 1985; Chen et al. , 2006).

3. 3 Coupling between upper- and lower-level jets

LLJs provide favorable conditions for the development of convective systems through transfer of momentum, heat, and water vapor, whereas upper-level jets provide favorable convergent or divergent situations through circulation configuration. At the sametime, the direct and indirect forced circulations induced by the upper-level jet can also promote threedimensional mass and momentum transport, and thus is coupled with the LLJ. The interaction betweenupper- and lower-level jets is one of the most important factors making the organized rainstorms form atthe exit region of the upper-level jet; and the upperlevel jet also promotes the formation, development, and enhancement of the LLJ(Uccellini, 1980; Ding, 2005). Using a straight upper-level westerly jet asan example, directly and indirectly forced circulationsrespectively form at the entrance and the exit region while moving under the effect of inertial rotation(Ding, 2005). The low-level backflow branch of theindirect circulation leads to an increase of horizontalpressure gradient force and thus enhances northwardisallobaric wind. The enhancement of the isallobaricwind and westerly wind component results in the LLJformation, which is embedded in the low-level backflow. At this time, the upper-level jet continuouslytransports cold and dry air masses eastward and theLLJ transports warm and moist air masses northward;the coupling between the upper- and lower-level jetsproduces favorable conditions for deep convection systems and rainstorms(Uccellini and Johnson, 1979). According to the convergence and divergence characteristics of upper-level jets, LLJs can also develop onthe right side of the exit region of upper-level jets(Si et al. , 1982; Xiao and Chen, 1984). Brill et al. (1985)proved the existence of this transverse indirect circulation at the exit region of upper-level jets, and thestrengthening effects from the upper-level jet and thediabatic heating on the LLJ. From the viewpoint of inertial gravity wave instability, Chen(1982)discussedthe coupling process between upper- and lower-leveljets. Given enough water vapor supply and a conditionally unstable atmosphere to the south of theupper-level jet, the nongeostrophic wind caused by advection will induce the development of the LLJ duringits adaptation process to the south of the upper-leveljet entrance region. With the help of numerical sensitivity experiments, Saulo et al. (2007)demonstratedthe interactions among the LLJ, mesoscale convectionsystems, and the upper-level jet, and exhibited a positive feedback process among them through a conceptual model. Generally, the upper-level jet related to anLLJ appears at a much higher level and does not possess obvious diurnal variation(Uccellini et al. , 1987).

3. 4 Synoptic system forcing

LLJs do not appear every day, despite topographic forcing and inertial oscillation always being inexistence. Whether or not the wind speed can reachthe necessary level of the criteria that define an LLJmay also depend to a certain degree on midlatitudesynoptic system forcing. By conducting a numericalsimulation of coastal secondary circulation, Uccelliniet al. (1987)found that the low-level wind speed significantly increased and reached a peak intensity of30 m s^-1 in response to the parcels' vertical acceleration under a baroclinic environment. When the parcelmoved toward the low-pressure system from the northeast direction, it experienced external forcing causedby the pressure gradient changes. Although this forcing was mild in the horizontal direction, the rapid upward acceleration became prominent in a baroclinicenvironment and with the effect of the coastal front. The significant enhancement of ageostrophic wind resulted in the formation of an LLJ during the parcel'supward acceleration.

When the LLJ accompanying rainstorms over theYangtze River basin or South China occurs, the western Pacific subtropical high and a low-pressure system(e. g. , southwest vortex)always exist to the east and west side of the LLJ. A large horizontal pressuregradient forms within the transition zone between thehigh- and low-pressure systems, and thus it increasesthe wind speed, which is favorable for the developmentof the LLJ(Xu et al. , 2004). Through the transformation and decomposition of ageostrophic wind in thenatural coordinate systems, Wang and Zhang(2012)gave four factors influencing the ageostrophic characteristics of the LLJ; namely, the non-constant windfield; inhomogeneous wind speed in the flow direction;curved streamline; and the atmospheric baroclinicity. The changes in the intensity and location of the western Pacific subtropical high and the lee systems fromthe Tibetan Plateau both have important influenceson the above four factors, especially the geostrophicdeviation caused by the curved streamline. The occurrence of LLJs in April over Southeast China mainlyresults from the northward shifting of the western Pacific subtropical high, and the occurrence of LLJs inJuly results from the intensification of detouring flowaround the Tibetan Plateau(Wang et al. , 2013). Inaddition, the LLJ is closely related to the formationof fronts, leeside cyclones, and leeside troughs(Ding, 2005).

3. 5 Positive feedback from diabatic heating

Diagnostic studies and numerical simulationsboth indicate that diabatic heating plays an important role in the development and strengthening ofmesoscale LLJ events(Huang, 1981; Uccellini et al. , 1987; Nicolini et al. , 1993). Chen and Yu(1988) and Si(1994)showed that the secondary circulation associated with rainstorms is one important factor involvedin the intensification of the LLJ. When a rainstormhappens, upward motion and latent heat release enhance the upper-level divergence, and the surface pressure decreases. Meanwhile, two secondary circulationsform on both the north and south sides of the rainfall area. The southern indirect circulation featuressoutherly winds, which gradually intensify and finallybecome the LLJ following the impacts of the pressuregradient and Coriolis forces. Qian et al. (2004)carriedout a case study of torrential precipitation during 11{17 June 1998 using the Pennsylvania State University NCAR Mesoscale Model(MM5). Their results demonstrated that the main source of water vapor for thisheavy precipitation event over the Yangtze River basinwas from the Bay of Bengal, and the moisture wastransported by a southwesterly LLJ southeast of theTibetan Plateau. Although the LLJ was largely manipulated by large-scale forcing, the mesoscale circulation that resulted from Meiyu condensational heating acted to increase the maximum wind speed of theLLJ. The intensified LLJ further promoted moisturetransport and thus formed a positive feedback and sustained the Meiyu precipitation system. Similarly, Zhao(2012)performed a case study of a localized extreme heavy rainfall event during mid-summer in central China with the Weather Research and Forecasting(WRF)model. He found that both the LLJ and simulated rainfall intensity were significantly decreasedwhen the latent heat release was shut down in the simulation, which proved that diabatic heating was one ofthe most important factors enhancing the LLJ.

In addition, Tao et al. (1980)found that thedownward transport of upper-level momentum canalso increase lower-level wind speed and lead to LLJevents, according to their diagnostic analysis. Convective mixing tends to make wind speed uniform inthe troposphere, and therefore promotes the downward transport of upper-level momentum, and formsa strong wind field over the convection area. It is important to note that this kind of low-level wind speedenhancement can only be mesoscale, and is not suitable for synoptic-scale jets. However, they also pointedout a shortcoming of this theory in so far as that it ishard to explain how the momentum concentrates atcertain levels. At the time of writing, we could find noresearch results that provide a reasonable explanationfor this possible mechanism of mesoscale LLJ formation.

In fact, it is clear that no one factor explains allthe phenomena and characteristics of LLJs from theabove listed mechanisms. It is the synergistic effect ofeach element that promotes the formation and development of the LLJ.

4. Relationships between LLJ and rainfall as well as other interdisciplinary fields4. 1 Rainfall

Among all the weather systems, LLJs show theclosest relationship with rainfall, which therefore attracts the most attention from meteorological researchers. Due to the influence of the East Asiansummer monsoon and the existence of complex topography, China consistently suffers from torrentialrainfall events. Statistical results show that the correlation coe±cient between LLJ and rainstorm can beup to 80% in China(Tao et al. , 1979). In the 1950s, Tao(1965)used the position of the LLJ to make rainfall forecast, and then a large number of studies appeared in the 1980s, focusing on the relationship between LLJ and torrential rainfall events(e. g. , Sun and Zhai, 1980; Huang, 1981; Yu, 1986). At present, theLLJ is still a key indicator in rainfall prediction overEast China. Because of the sharp vertical shear belowthe jet core, this region is always accompanied by potential instability. Therefore, the LLJ not only transports heat and moisture during precipitation, but alsofrequently stimulates the mesoscale fluctuation propagating along the jet core due to its strong instability. This kind of sharp increase of wind speed can triggermesoscale convective systems and thus lead to torrential rainfall events(Sun and Zhai, 1980).

Many studies have been carried out to investigatethe unstable conditions caused by the LLJ. Accordingto the numerical solutions of the linear and inviscidBoussinesq approximation equations, Mastrantonio etal. (1976)derived the LLJ-generated gravity waves. Tao et al. (1979)indicated that the minor-cycle windpulsation over the jet core may reflect the effects ofgravity waves triggered by ageostrophic movements. The adaptation theory of atmospheric motion showsthat the geostrophic deviation can excite inertial gravity waves, and the intensity of waves is proportional tothe magnitude of geostrophic deviation. Along withthe dispersion of unstable energy caused by inertialgravity waves, super-geostrophic wind gradually weakens and the severe rainstorm disappears(Yu, 1986). In addition, Kuo and Seitter(1985)pointed out thatthe geostrophic LLJ can produce unstable mesoscaledisturbances that resemble frontal cloud b and s and squall lines. Limaitre and Brovelli(1990)also indicated that the basic flow of the LLJ can generatebaroclinic symmetric instability, which may generate anarrow mesoscale unstable line during the tilting motions. Both theoretical analysis and a case study werecarried out by Zhang and Zhou(2003)concerning theinertial and symmetric stabilities for the front-left region of a southwesterly LLJ and the back-right regionof an upper-level strong northerly flow. Their resultsshowed that the existence of the maximum inertial stability in the front-left region of the LLJ is favorable forthe accumulation of moist thermal energy, and conditional symmetric instability or convective instabilitycan be expected for the development of slanted convective instability in this region.

The LLJ is the major water vapor transport channel for rainfall events(Findlater, 1969; Saulo et al. , 2007). Roads et al. (1994)found that in most areas of the United States, moisture flux convergencefrom model outputs exhibits a good correlation withrainfall observations, which is likely due to the moisture transport by the LLJ. Higgins et al. (1997)further examined the summer rainfall and moisture transport over the central United States with observations and high temporal and spatial resolution assimilateddatasets. Their results demonstrated that the GreatPlains LLJ is extremely important for the moisturebudget throughout the summer season. The impact ofthe LLJ on the overall moisture budget during summeris considerable, with low-level inflow from the Gulf ofMexico increasing on average by more than 45% overnocturnal mean values. Similarly, the peak precipitation episode of the 1993 summer flood over the GreatPlains was associated with a sustained period of highincidence of strong LLJs(over 20 m s^-1)(Arritt et al. , 1997).

4. 2 Air pollution

LLJs, especially those occurring below 200 m, significantly affect the vertical wind shear within thelayer from the ground surface to the jet core. As a consequence, LLJs control the exchange process of pollutants between the surface and the atmosphere and determine the air quality over the region. Studies on therelationship between the LLJ and air pollution havegradually increased in recent years because of rapidurbanization and increasing attention on environmentissues. For example, Banta et al. (1998)indicated thatthe transmission effect by nocturnal LLJs shows significant influences on sustained pollution events in urbanareas. Ryan(2004)pointed out that O3 concentrationsare enhanced when southwest LLJs occur, with an average peak of 82. 5 ppbv over the mid-Atlantic states, with 44% of these days exceeding the 8-h Code Orangethreshold(85 ppbv) and 22% exceeding the Code Redthreshold(105 ppbv). When southwest LLJs are notassociated with high O3, it is typically due to thunderstorm formation or cloud cover in advance of frontalboundaries. Airborne observations of trace gases, particle size distributions, and particle optical propertieswere made by Taubman et al. (2004)at a constant altitude along a transect from New Hampshire to Maryl and on 14 August 2002, the final day of a multi-dayhaze and ozone(O3)episode over the mid-Atlantic and northeastern United States. These observations, together with chemical, meteorological, and dynamicalanalyses, suggested that the influence of the Appalachian lee trough and LLJ during this episode redirected the westerly synoptic flow in a more southerlydirection during the day and evening, respectively. Asa result, the polluted air masses transported from theindustrialized Midwest mixed with the urban plumesof the eastern seaboard, which reinforced the air pollution over this region.

4. 3 Wind energy utilization

Because of the strong wind speed and sharp vertical and horizontal wind shear, LLJs have an importantrole in harnessing wind energy as well as wind turbineprotection. Considering the huge potential dem and for renewable energy, especially wind energy, in thecoming years, a project named the Lamar Low-LevelJet Program(LLLJP)was established in 2003 as ajoint effort among the U. S. Department of Energy and other research centers and companies. The purpose ofthe project is to develop an underst and ing of the influence of nocturnal LLJs on the inflow turbulence environment and to document any potential operationalimpacts on current large wind turbines and the LowWind Speed Turbine(LWST)designs of the future(Kelley et al. , 2004). Baidya Roy et al. (2004)explored the possible impacts of a large wind farm in theGreat Plains on local meteorology with the regional atmospheric modeling system(RAMS)model. Their results showed that the wind farm significantly decreasesthe wind speed at the turbine hub-height level. Meanwhile, turbulence generated by rotors creates eddiesthat can enhance vertical mixing of momentum, heat, and moisture, usually leading to warming and dryingof the surface air and reduced surface sensible heatflux. As a result, this process changes the impact ofthe LLJ on regional climate. This effect is most intense in the early morning hours when the boundarylayer is stably stratified and the hub-height level windspeed is the strongest due to the nocturnal LLJ. Thefrequent nocturnal LLJ events over the Great Plainsmake this region quite favorable for wind energy collection and utilization(Storm et al. , 2008). However, the presence of LLJs can significantly modify vertical shear and nocturnal turbulence in the vicinitiesof wind turbine hub heights, and therefore the LLJposes a potential threat to turbine rotors. Hence, ourknowledge on LLJs will to some extent determine theprecise assessment of wind energy resources, as well asthe reliable prediction of power generation and robustdesigns of wind turbines(Storm et al. , 2008).

4. 4 Aviation safety and other interdisciplinary fields

LLJs aligned in the direction of an airport runway can cause wind shear during aircraft take-off and l and ing. Wind shear(a sudden change in the winddirection and /or wind speed)results in a change inthe lift of the aircraft. The risk is particularly highfor departing aircraft entering the jet at a steep angle in the initial climbing phase. The sudden gainin headwind in an LLJ, immediately followed by anabrupt loss, may be likened to the experience of passing through a microburst, although in this instancethere is clearly no convective activity to cause any microburst(Lau and Chan, 2003). The change in vertical buoyancy caused by the vertical wind shear greatlyaffects the smoothness of the aircraft's flight, and especially impacts upon safety during take-off and l and ing(Li, 2006). Weather information and warnings onwind shear are thus very important to aircraft operation and safety. Many air crash events have beencaused by sudden dramatic wind shear during flight.

Furthermore, Fromm and Servranckx(2003), Tang et al. (2004), and Liechti(2006)have carried outextensive research on the relationships between LLJs and s and storms, forest fires, and bird migration, and obtained meaningful results. For example, Tang et al. (2004)demonstrated that the position and intensity ofthe LLJ can be an important prediction index for theintensity and affected area of a s and storm. The results from Fromm and Servranckx(2003) and Liechti(2006)respectively indicated that LLJs can accelerate the spread of forest fire, and also act as an orientation reference for the seasonal migration of birds, helping to reduce their energy consumption. In addition, the LLJ can change the regional climate inAntarctica through its influence on sea-ice distribution(Schwerdtfeger, 1975), and LLJs also appear to be essential in prescribing the larger-scale circulations nearthe South Pole when associated with drainage flows(Parish and Bromwich, 1991). Overall, LLJ-relatedresearch provides valuable points of reference that canbe applied to improving people's quality of life, environmental protection, and resource development and utilization. Further studies on LLJ events will enrichour underst and ing of meteorological phenomena and thus help us to more successfully forecast and preventassociated meteorological disasters.

5. Future challenges and prospects for LLJ research

Currently, a majority of LLJ research focuses onthe Great Plains of North America, the Andes Mountains in South America, and the Somali region of EastAfrica. Comparatively inadequate studies have beencarried out to explore LLJ characteristics and theirmechanisms of evolution in East Asia, especially forLLJs over complex terrain to the east of the TibetanPlateau.

5. 1 Coarse LLJ definition

The definitions for East Asian LLJs, especiallythe LLJs over mainl and China are relatively coarse. Most studies do not consider vertical shear, but onlyuse the maximum wind speed at a single pressure level(usually below 600 hPa)as the LLJ selection criterion. For example, the LLJs in Wang et al. (2003)satisfythe following criteria: the wind speed at a single station should be ≥ 12 m s^-1 at 700 or 850 or 925 hPa;the wind direction should be ≥ 180± and 6 265±. Ina study on the characteristics and formation mechanisms of the Beijing summer boundary layer LLJ, Sun(2005)defined the LLJ as: the wind should be easterlyor southerly and the speed should be ≥ 12 m s^-1 below 1500 m at 2000, 0200 or 0800 LST; meanwhile, thesynoptic-scale LLJ was excluded. Similar definitionswere widely adopted in early studies of LLJs and rainstorms in China(Yuan, 1981; Yu et al. , 1983). TheLLJs with such criteria may belong to the low-leveljet stream, and research using these coarse definitionsmay mislead our underst and ing to a certain extent onLLJ characteristics, the relationship between LLJ and rainfall, as well as other weather and climate phenomena. Therefore, we need clear LLJ selection criteria infuture studies so as to correctly underst and LLJs inChina.

5. 2 Lack of observations and inadequate quality control

Sounding stations are sparsely distributed inChina, and regular soundings are released only twicedaily at 0000 UTC and 1200 UTC, which do not tallywith the strongest and weakest times of LLJ occurrence. Hence, we cannot use these sounding datato capture detailed LLJ characteristics. In fact, theamounts of other types of sounding data over Chinahave increased in recent years. However, they arenot in st and ardized data formats, and data postprocessing problems and barriers to data sharing makethese data less well used. Because the cost of fieldexperiments is relatively high and it is hard to ensure quality control of the observational data theyyield, they are seldom launched. Furthermore, Chinastill does not have high spatial and temporal resolution regional reanalysis data for scientific researchdue to limitations in data observation and data assimilation techniques. The above-mentioned factorsgreatly limit LLJ research in China, and our knowledge on LLJ-related issues remains inadequate as aresult. However, on the positive side, recent increasesin regular observations and wind profiler radar dataprovide good opportunities for LLJ studies. For example, with half-hourly data from a wind profiler radar atthe Qingpu site during the warm season of 2008{2009, Du et al. (2012)developed a climatology of the LLJsover Shanghai and exhibited the LLJ types, temporal evolution, and statistical relationship with rainfallevents.

5. 3 Few thorough explorations of LLJ characteristics and formation mechanisms

Most LLJ studies in China focus on the relationships between LLJ and rainstorms or the East Asianmonsoonal system(Tao et al. , 1979; Si, 1994; Zhang et al. , 2002), and very little attention has been givento LLJ structural and evolutionary mechanisms. Although some early studies were carried out on the status of LLJ distributions, horizontal and vertical structures, as well as diurnal variations, they did not capture enough detailed information due to the too-coarsetemporal and spatial resolutions at that time(Li et al. , 1981; Sun, 1986). Besides, as mentioned, research onthe mechanisms of LLJ evolution and other key factors over China is inadequate. For example, the terrain of East Asia is complex, characterized by obviousthree-step topography:(1)the Tibetan Plateau;(2)the Hengduan Mountains, Yunnan-Guizhou Plateau, Loess Plateau, and Inner Mongolian Plateau; and (3)the Southeast Hills, North China Plains, and Northeast Plain. Unfortunately, most studies do not consider this three-step structure and instead focus onlyon the Tibetan Plateau to test the effect of topographyon the LLJ over China(Chen and Qian, 1993; Liu and Jiao, 2000). Obviously, this approach means that it ishard to explain the relative contributions from eachterrain type. That said, on the positive side, higherresolution numerical models provide a good platformupon which LLJ research can build. Scientists canadopt numerical experiments to further investigate theindividual and collective contributions of each factorinfluencing LLJs over East Asia.

5. 4 Limited studies in interdisciplinary fields

Compared to the large amount of research onthe relationship between LLJ and precipitation processes, limited studies have been carried out in otherinterdisciplinary fields. With the rapid economic and social development in recent decades, the issues ofenvironmental pollution and new energy explorationbecome especially urgent. The frequent appearanceof haze and s and storms can impose serious harm onpeople's health and quality of life. Besides, the dependence on foreign energy inhibits China's comprehensive and sustainable development potential in economic terms. The excessive use of fossil fuels accelerates global warming and leads to the frequent occurrence of extreme weather and climate events. Therefore, we need to underst and the interactions betweenLLJs and air pollution, wind energy utilization, as wellas other interdisciplinary fields, in order to deal withthe problems encountered during the process of economic development. Such studies are of great realisticsignificance.

6. Summary

In this article, we have summarized and assessedthe current knowledge of LLJ-related subjects over thepast five decades from both home(China) and abroad. We focused especially on three aspects: LLJ classification, definition, distribution, and structure; LLJ formation and evolution mechanisms; and the relationship between LLJ and rainfall, as well as other interdisciplinary fields. The shortcomings of LLJ studiesin China are discussed, and the future prospects forseveral LLJ research avenues are speculated upon.

LLJ research in China currently has the problemssuch as coarse LLJ definitions, lack of observations and inadequate quality control of data, too few thorough explorations of LLJ characteristics and formationmechanisms, and limited studies in interdisciplinaryfields. Therefore, we need to pay attention to the following aspects in future studies. (1)Making clearregulations regarding LLJ criteria. We should notonly consider the horizontal maximum wind speed, butalso give strict requirements on vertical and horizontalshears. (2)Strengthening observational data collection and organization, increasing observation in specific layers, and carrying different types of field experiments. Critical quality control of observations shouldalso be carried out, such as on wind profiler data. (3)Based on previous studies, investigating LLJ structure and evolutionary mechanisms, especially by exploringthe impact of complex terrain in East Asia on LLJdevelopment. (4)Performing work in interdisciplinaryfields so as to underst and related issues, and providea basis for weather forecasting and disaster prevention and mitigation. As observations increase, study methods develop, and more scientists invest their time inLLJ research, we will undoubtedly gain a deeper underst and ing of LLJs in the future. Such knowledgecan then be applied to providing better guidance insocial and economic contexts, as well as reducing theamount of serious losses caused by related disastrousevents.

Acknowledgments.   The authors are thankfulto Prof. Da-Lin Zhang for his enthusiastic and rig-orous comments, and suggestions with respect to thewriting of this article.