A barrier jet is frequently found along the northwestern coast of Taiwan in the prefrontal southwesterly flow regime during the Taiwan Area Mesoscale Experiment (TAMEX). It has a maximum wind speed of 14 m s−1 at approximately 1 km above the surface with a vertical wind shear approximately 10 × 10−3 s−1 below and 4 × 10−3 s−1 above that altitude. During TAMEX, the southwesterly monsoon flow strengthens over Taiwan when the low-level pressure trough/surface front moves toward the southeastern China coast. The barrier jet is a result of the stably stratified airflow past an island obstacle under a small–Froude number [<O(1)] flow regime. During the occurrence of a barrier jet, a windward pressure ridge is observed along the southwestern coast due to island blocking. In low levels, the incoming southwesterly flow decelerates off the southwestern coast and moves around the island. Along the western coast, the deflected airflow accelerates northward with a large cross-contour wind component down the pressure gradient along the western coast, resulting in a barrier jet over northwestern Taiwan. The barrier jet is the strongest when the surface windward ridge–leeside trough pressure pattern is most significant and weakens after a surface front arrives.
Strong winds or jet streams in the lower troposphere have generated considerable interest because of their importance to air pollutant transport, thunderstorm development, wind energy production, and aviation safety. A strong low-level mountain-parallel flow as a result of blocking is generally referred to as a barrier jet (Schwerdtfeger 1975; Parish 1982; Overland and Bond 1993). In cold-air damming events (Forbes et al. 1987; Bell and Bosart 1988; Doyle and Warner 1991), the gradient winds aloft have a significant cross-mountain easterly wind component with the arrival of a high pressure center north of a north–south-oriented mountain range. A cold dome and high pressure anomaly develop on the eastern slope because of cold advection and orographic blocking in low levels. The cold dome is held against the mountain by the Coriolis force induced by the mountain-parallel jet and the easterly wind component aloft (Xu 1990). The locally enhanced baroclinity across the sloped inversion results in a strong horizontal pressure gradient perpendicular to the mountains that forces a mountain-parallel geostrophic wind speed maximum below the sloped inversion.
Overland and Bond (1995) suggested that the flow characteristics of a mountain-parallel jet depend on thestrength and the stability of the upstream flow, and on the scales (height, width, and length) of the mountain range. For airflow past a long mountain range, the Burger number [B = (N/f)(H/L)] characterizes the scaled mountain slope where N, f, H, and L are the Brunt–Väisälä frequency, the Coriolis parameter, the averaged height of the mountain, and the mountain half-width, respectively (Pierrehumbert and Wyman 1985; Overland and Bond 1995). It can be written as Ro/Fr, where Ro and Fr are the Rossby number (Ro = U/fL) and the Froude number (Fr = U/NH), respectively, and U is the upstream wind speed. For B ≪ 1, the flow is quasigeostrophic as the flow proceeds over the mountain. If B > 1, the mountain is hydrodynamically steep and the influence of the mountain is independent of the mountain half-width (Overland and Bond 1995); the mountain ridge is wall-like and the alongshore momentum balance is ageostrophic. If 0.1 ⩽ B ⩽ 1.0, the flow is semigeostrophic and it is modified by the slope of the mountain, primarily over the mountain. For typical mesoscale mountain ranges, such as the Alps (Pierrehumbert and Wyman 1985) and the central mountain range of Taiwan (Fig. 1), H ∼ 2.5 km, L ∼ 50 km, N ∼ 10−2 s−1, f ∼ 10−4 s−1, B is greater than 1. For steep mountains (B > 1), the Fr and Ro can be used to characterize the behavior of the flow (Pierrehumbert and Wyman 1985; Xu 1990; Trüb and Davies 1995). The extent of upstream influence is trapped within the Rossby radius of deformation (Pierrehumbert and Wyman 1985; Overland and Bond 1995). The Coriolis force would inhibit the upstream influence far upstream of the obstacle by forcingan adjustment to geostrophy. Xu (1990) found that the cold dome shrinks and the interface slope increases as the Fr increases or as the Ro decreases. If the Fr is large enough, the cold dome will either vanish or not be in a steady state.
Two-thirds of the island of Taiwan is mountainous. The central mountain range is oriented north-northeast to south-southwest with an average terrain height of about 2.5 km (Fig. 1). During the early summer rainy season (May–June) over Taiwan, the southwest monsoon flow prevails in low levels. Both the dynamic blocking of the low-level airflow and the thermally induced local circulations are significant (Wang 1986; Kuo and Chen 1990; Chen and Li 1995a). The Taiwan Area Mesoscale Experiment (TAMEX), conducted jointly by scientists of Taiwan and the United States during May–June 1987, provides a unique opportunity to study the blocking of the low-level flow by the island topography (Kuo and Chen 1990). In a TAMEX case study, Chen and Li (1995b) found locally strong winds existing at low levels along the northwestern coast under the prevailing southwest monsoon flow. They suggest that the strong coastal winds are due to the blocking of the southwesterly flow by the island obstacle. Li et al. (1997) show that during TAMEX IOP 13, under favorable large-scale conditions, the convergence between the barrier jet and the westerly flow coming from the southeastern China coast plays an important role in focusing heavy rainfall along the northwestern coast of Taiwan.
Based on the analyses of previous case studies [i.e., Chen and Li (1995a,b); Li et al. (1997)], we define the barrier jet as a strong southwesterly flow along the northwestern coast if the low-level flow accelerates down the pressure gradient along the coast and the wind speed has a local maximum (≥10 m s−1) between 0.5 and 1.5 km over northwestern Taiwan. In this study, we will survey all the barrier jets during TAMEX to examine the general characteristics of the jets in terms of their structure, horizontal extent, and evolution. We will also investigate the synoptic-scale settings for the occurrences of these jets. Based on these analyses, the linkage of the barrier jet to the synoptic-scale flow patterns and local pressure distributions will be discussed.
2. Data sources
The low-level wind data were obtained from routine upper-air soundings over coastal areas in southeastern China and high-resolution TAMEX soundings (Fig. 1). The routine upper-air sounding data along the southeastern China coast were available at 12-h intervals. During TAMEX, there were 12 rawinsonde and 10 pibal stations over the Taiwan area. The high-resolution TAMEX sounding observations were usually made at 6-h intervals during 10 May–15 June 1987 and at 3-h intervals during 13 intensive observing periods (IOPs) (Cunning 1988). In this study, high vertical resolution (∼5 hPa) wind data at four sounding stations (station 46695, 46685, 46751, and ship station RCHY, Fig. 1) over northwestern Taiwan are used to examine the structure of the barrier jet including the horizontal extent and vertical wind profiles. In addition, aircraft data presented by Chen and Li (1995b) are used to analyze the flow acceleration along the western coast for a barrier jet. The data processing procedures were given by Chen and Li (1995b).
During TAMEX, the surface data were obtained from 74 surface observation stations and 21 wind tower stations (Cunning 1988). Wind speed, wind direction, and temperature were recorded at most surface stations every hour, increasing to every half-hour during IOPs (Cunning 1988). The Central Weather Bureau (CWB) stations (a total of 25 stations, Fig. 1) recorded routine 3-h surface data throughout the entire experiment (Chen and Schumann 1990). Tower stations (Fig. 1) measured wind speed and direction every 10 min at a height of about 10 m above the ground.
3. General characteristics of barrier jets along the northwestern coast during TAMEX
In this section, we first present a barrier jet case during TAMEX IOP 3 (21–22 May 1987). After that, we survey barrier jets during TAMEX and analyze their generalcharacteristics in terms of the horizontal extent, vertical structure, and evolution. The spatial structure of the barrier jets is estimated from the sounding data. The evolution of barrier jets is discussed based on the stages classified by Chen and Li (1995a) from the surface airflow and sea level pressure patterns.
a. A case study of the barrier jet during IOP 3
Chen and Li (1995b) analyzed the low-level airflow and pressure patterns over Taiwan for the 21–22 May (IOP 3) case. At 1200 UTC 21 May, a low-level pressure trough/surface front moved toward the southeastern China coast (Fig. 2a). Rising motion was diagnosed along the southeastern China coast from the synoptic-scale data with fair weather conditions over Taiwan (Chen and Li 1995b). The ambient flow impinged onthe central mountain range of Taiwan with a noticeable angle. A windward ridge–leeside trough sea level pressure pattern developed over Taiwan as a result of blocking of the southwesterly flow by the island obstacle. The D value, which is equal to the actual altitude minus the pressure altitude, is a measure of pressure perturbation (D = p/ρg) from a reference state. For every point along the 900-m flight leg, if the flight altitude is slightly lower or higher than 900 m, the pressure at the 900-m level is calculated from flight-level data based on hydrostatic equation. Then, the 900-m pressure perturbation (p) from the average pressure at the 900-m level is computed and converted to the D value. The subjectively analyzed horizontal distribution of the D value at the 900-m flight level exhibits a windward pressure ridge–leeside trough pattern (Fig. 2b). The windward ridge is caused by adiabatic cooling as the incoming flow experiences forced ascent (Smith 1982); the leeside trough is a result of adiabatic warming on the lee side when the airflow aloft moves across the mountain range (Smith 1979, 1982; Sun et al. 1991). At the 900-m level, the flow deceleration is evident off the southwestern coast (Fig. 2b) as the southwesterly flow encounters high pressure there. The low-level airflow splits off the southwestern coast and moves around the island, consistent with the theoretical studies of the airflow past an isolated island for a low–Froude number [Fr < O(1)] flow regime [e.g., Smolarkiewicz et al. (1988); Smith (1989); Sun et al. (1991)]. The Froude number is estimated by Vn/NH, where Vn is the upstream wind speed normal to the central mountain range. Discussion on the calculations of Fr is given in section 4. For Fr ≫ 1, the airflow approaches the three-dimensional potential-flow solution and is approximated by the linear theory (Smith 1980, 1989). As the Froude number passes below O(1), flow splitting may occur on the windward side (Smith 1989; Smolarkiewicz and Rotunno 1990) with a stagnation point at the lower boundary (Smith 1988). When Fr → 0, the airflow approaches the horizontal potential flow solution (Drazin 1961) and all the fluid is diverted laterally around the mountain.
Along the western coast, the low-level airflow at the 900-m level is almost parallel to the coast line. It has a cross-contour wind component down the windward pressure ridge and accelerates downstream (Fig. 2b). A barrier jet with a maximum wind speed (∼15 m s−1) is observed along the northwestern coast (Fig. 2b). The wind speed reduces significantly to 8 m s−1 immediately north of the island. To investigate the flow acceleration down the pressure ridge along the coast, an estimation of the acceleration, the pressure gradient force, and the Coriolis force is made from the aircraft data along the flight track between points A and B (Fig. 2b). For a coordinate system along the northwestern coast, which is almost parallel to the central mountain range (about 23° from the north), the equation of motion for the alongshore wind component can be written as
where υs and υn are the wind components parallel and normal to the northwestern coast. The Coriolis parameter (f) is 5.9 × 10−5 s−1. Terms I–III and FR represent acceleration, pressure gradient force, Coriolis force, and frictional force, respectively. The pressure gradient force is calculated from the D value along the flight tracks. The acceleration term in the left-hand side of Eq. (1) includes four terms (i.e., ∂υs/∂t, υs∂υs/∂s, υn∂υs/∂n, w∂υs/∂z).
The barrier jet reaches the maximum wind speed near the 900-m flight level and it is also almost parallel to the flight track along the northwestern coast (Fig. 2b). The υn wind component along the flight track is less than 10% of υs (Table 1). Allowing for a 10% error, the υn∂υs/∂n term at the 900-m level can be neglected. Under steady-state assumption and neglecting vertical advection, term I at the 900-m level can be estimated by υs(Δυs/Δs), where Δs is the distance between points A and B (∼260 km), υs is the mean alongshore wind speed, and Δυs is the difference of the alongshore wind speed between A and B. The frictional force term is computed as a residual. Note that the local change term (∂υs/∂t) could affect the estimated magnitude of term I because the barrier jet is strengthening during the flight period (Chen and Li 1995b). Based on the sounding data at stations 46685, 46751, and RCHY (Fig. 1) over northwestern Taiwan, the local change term, averaged in the layer of 0.5–1.5 km during the flight period, is about 0.1–0.4 × 10−4 m s−2, which is small as compared to the υs∂υs/∂s term (2.4 × 10−4 m s−2). Ignoring horizontal advection in the direction normal to the flight track and using the steady-state assumption probably introduces a 10%–20% error in our estimation of term I. At the 900-m level, the inertial advection term and the pressure gradient force are the dominant terms in Eq. (1) (Table 1). This suggests that the flow accelerating downstreamalong the northwestern coast is mainly caused by the pressure gradient force
In integrated form, this gives the Bernoulli equation
where υso is the initial alongstream velocity and ΔΦ is the alongstream geopotential difference. The force balance for the barrier jet along the northwestern coast of Taiwan is similar to gap winds in the Strait of Juan de Fuca (Overland and Walter 1981). For TAMEX barrier jets, the along-stream pressure difference is primarily orographically induced, whereas for gap winds in the Strait of Juan de Fuca, the along-stream pressure difference is both imposed by the synoptic situation and generated locally by orography.
At the 2.5-km level, the flow acceleration along the west-northwestern coast is not significant [Fig. 24 in Chen and Li (1995b)]. Along the eastern coast, winds beneath the ridge top of the mountain range are weak because the southwesterly flow is blocked by the mountain range. Analysis of the aircraft data off the northwestern coast at the 2.5-km level shows that the alongshore pressure gradient force is nearly balanced by the Coriolis force in the alongshore direction and the flow acceleration is quite small as compared to either the pressure gradient or the Coriolis force (Table 1). During the flight period, a low-level pressure system (L1) is located to the north of the island (Fig. 2a). The pressure gradients at the 2.5-km level off the northwestern coast are related to synoptic-scale pressure patterns. It is apparent that at the 2.5-km level, the winds off the northwestern coast represent large-scale flow and are not significantly affected by the island topography. These results show that the barrier jet develops along the northwestern coast below the average height (∼2.5 km) of the central mountain range as a result of the interaction between the low-level southwesterly flow and the island topography. Because the aircraft measurements at the 2.5-km level are made off the western and northwestern coasts rather than above the ridge tops, we are unable to determine whether or not vertical propagating mountain waves exist (Smith 1979) at this time.
b. Barrier jets during TAMEX and their spatial structure
Because the dynamics of the barrier jet was not considered in the experiment design of TAMEX, no upper-air observations were available along the northwestern coast of Taiwan. We analyzed the time series of the averaged winds for the 0.5–1.5-km layer observed at station 46685 (25.0°N, 121.43°E, Fig. 1) for the entire TAMEX period. This station, which is located about 30km east of the northwestern coast (Fig. 1), is the only TAMEX sounding station close to the strong mountain-parallel winds observed by the aircraft (Fig. 2b). As will be shown later (section 3c), wind observations at this station also exhibit a notable υn wind component during barrier jet occurrences as the deflected airflow moves around the northern corner of the central mountain range. Therefore, the υs wind component rather than the total wind speed measured by station 46685 is used to identify barrier jets during TAMEX. It is hoped that in the future, continuous measurements of vertical wind profiles along the northwestern coast will be made to better describe the structure and evolution of the barrier jet.
During TAMEX, the magnitude of υs averaged in the 0.5–1.5-km layer at station 46685 is greater than 10 m s−1 at 0600 UTC 13 May, 1200 UTC 16 May, 1800 UTC 21 May, 0000 UTC 22 May, 1200 UTC 22 May, 0000 UTC 27 May, 0000 UTC 2 June, 0000 UTC 7 June, 0000 UTC 22 June, 1200 UTC 22 June, 0000 UTC 23 June, 0600 UTC 23 June, 0000 UTC 24 June, and 1200 UTC 24 June, respectively (Fig. 3a). For all these times, the wind profiles at this station show strong southwesterly winds with υs greater than 10 m s−1 at the 1-km level (Fig. 3b). Thus, there are seven barrier jet events (13–14 May, 16–17 May, 21–22 May, 26–27 May, 1–2 June, 7–8 June, and 22–25 June) during TAMEX. In contrast to the nocturnal low-level jet over the Great Plains (Blackadar 1957; Helfand and Schubert 1995; Higgins et al. 1997), there is no apparent preferred period during the diurnal cycle for barrier jet occurrences over the northwestern coast of Taiwan. The lastbarrier jet event (22–25 June) during TAMEX lasted for more than two days with a persistent windward ridge–leeside trough pressure pattern at the surface. Over northern Taiwan, Chen (1985) shows that nocturnal heavy rainfall events dominate with 73% in the nighttime hours of 2100 to 0800 LST. Continuous observations of wind profiles along the northwestern coast and numerical simulations are needed to study the effects of the diurnal heating cycle on the strength and evolution of the barrier jet.
Despite the fact that station 46685 is not located at the center of the maximum wind speed area observed by the aircraft, the averaged profile of υs for the above observational times shows a maximum of approximately 14 m s−1 at the 1-km level (Fig. 4a). The vertical shear is about 10 × 10−3 s−1 below the level of the jet core (∼1 km) and 4 × 10−3 s−1 above, respectively. The standard deviation of υs at the 1-km level is 5 m s−1 (Fig. 4a), indicating large variations in the strength of the barrier jets. The strongest TAMEX barrier jet occurs during 23–25 June. For this particular case, υs averaged between the 0.5-km and the 1.5-km levels at station 46685 reaches 17 m s−1 (Fig. 3a) and the maximum υs is 23 m s−1 at the 1-km level (Li et al. 1997).
Figure 4b shows the composited vertical profile of virtual potential temperature θυ, averaged for the above observational times during TAMEX. The θυ increases slightly with respect to height in the lowest 1 km (Fig.4b). The rate of this increase is larger in the 1.0–1.3-km layer than that in the lowest 1 km. The θυ continues to increase with height above 1.3 km. The vertical profile of virtual potential temperature indicates that the depth of the weakly stable boundary layer is about 1 km. The alongshore wind maximum occurs at the top (∼1 km) of the weakly stable boundary layer along the northwestern coast. Analysis of pressure field for the flight period of 21 May shows that the pressure difference between point A and B (Fig. 2b) is about 0.7 hPa (6.7 m) at the 900-m level. During the same period, the sea level pressure near point A, averaged over stations 46730 and 46741 (Fig. 1), is 1007 hPa and the sea level pressure near point B, averaged over stations 46757 and 46690 (Fig. 1), is 1005.9 hPa. The sea level pressure difference between points A and B is about 1.1 hPa, which is 0.4 hPa greater than at the 900-m level. Although the pressure gradient along the northwestern coast is the largest at the surface due to island blocking, the friction drag (FR) in the boundary layer (Stull 1988) would reduce the wind speed in the lowest levels.
To examine the horizontal extent of the barrier jets, we compare the averaged υs wind component in the layer of 0.5–1.5 km at station 46685 to those at the nearby sounding stations (stations 46695, 46751 and RCHY, Fig. 1) during TAMEX. For the seven barrier jet cases, υs at station 46685 always has the largest values among these four stations (Figs. 3a and 5). Station 46695 (25.63°N, 122.07°E) is located about 90 km northeast of station 46685 (Fig. 1). The averaged υs in the 0.5–1.5-km layer at station 46695 is always less than 10 m s−1 (Fig. 5a). For all seven barrier jet cases during TAMEX, the averaged ratio of the magnitude of υs at station 46695 to that at station 46685 is about 1:2, indicating that barrier jets weaken considerably after moving awayfrom the island. At station 46751 (24.20°N, 120.65°E), about 120 km south-southwest of station 46685, the averaged magnitude of υs is less than that at station 46685 (Figs. 3a and 5b). Only for the 21–22 May, the 7–8 June, and the 22–25 June cases, υs is greater than 10 m s−1 at station 46751 (Fig. 5b). For the seven barrier jet cases, the magnitude of υs at station 46751 is only 63% of that at station 46685. These results suggest that along the direction of the barrier jet, the jet has its maximum intensity over northwestern Taiwan near station 46685. These results are consistent with the aircraft and sounding observations for IOP 3 (Fig. 2b).
The width of the barrier jet is difficult to determine from TAMEX data because there is only one sounding station (RCHY, 24.50°N, 119.75°E) within the northern Taiwan Strait (Fig. 1). In addition, the sounding observations at this station only cover five barrier jet cases (13–14 May, 16–17 May, 21–22 May, 26–27 May, 1–2 June, and 7–8 June) during TAMEX. The ship station, RCHY, is about 90 km west of the coast (Fig. 1). For these five cases, the magnitude of υs averaged in the layer of 0.5–1.5 km at station RCHY is less than 10 m s−1 except at 1200 UTC 22 May (Fig. 5c). For these five cases, the averaged magnitude of υs at station RCHY is about 61% of that at station 46685. For 1200 UTC 22 May, the magnitude of υs averaged in the layer of 0.5–1.5 km at station RCHY is 10.5 m s−1 (Fig. 5c), about 4 m s−1 smaller than that at station 46685 (Fig. 3a).
Theoretical analysis indicates that for the flow regime with a Froude number less than 1, an appropriate height scale for the disturbance induced by a steep mountain is the gravity height scale Hs [Hs = Vn/N (Smith 1989)]. For the coastal mountain area, the disturbance will grow seaward to a horizontal distance given by a Rossby radius LR based on this height (Overland and Bond 1995):
For the TAMEX barrier jet cases, Vn is about 4–8 m s−1. Thus, the Rossby radius LR is about 60–120 km. The estimated Rossby radius is consistent with our observations that the magnitude of the alongshore wind component at RCHY, about 90 km offshore, is only about 60% of that at station 46685. Although it is difficult to define the exact width and length of the barrier jet from limited data, it is apparent that the barrier jet along the northwestern coast is localized in nature with limited horizontal extent as compared to the subsynoptic-scale LLJ (Chen and Yu 1988).
One of the unique features of the barrier jet over northwestern Taiwan is that the jet is located north of the major mountain peaks, in contrast to mountain-parallel winds observed along the windward side of long mountain ranges (Schwerdtfeger 1975; Forbes et al. 1987; Bell and Bosart 1988; and others). For those cases, the force balance along the jet is between the pressuregradient and frictional forces, whereas the force balance in the mountain-normal direction is largely between the Coriolis force and the pressure gradient generated by the accumulation of cold air against the mountain (Schwerdtfeger 1975; Bell and Bosart 1988; Xu 1990). Because the airflow and orography of Taiwan are three-dimensional, the dynamics of the barrier jet along the northwestern coast of Taiwan is different from the classical theoretical studies of barrier winds from 2D models (Pierrehumbert and Wyman 1985; Xu 1990; Trüb and Davies 1995). With an extra degree of freedom, the airflow is allowed to go around the orography under a low–Froude number [<O(1)] flow regime (Pierrehumbert and Wyman 1985; Smolarkiewicz et al. 1988; Smith 1989; Sun et al. 1991). In the case of circular 3D mountains, the airflow is qualitatively different from 2D flow for the same flow parameters because of horizontal streamline splitting (Miranda and James 1992). Other dynamic processes such as vortex shedding (Schär and Smith 1993; Sun and Chern 1993) also occur in 3D flow. Based on the linear theory for a nonrotating fluid, stagnation is predicted to occur at the surface and aloft simultaneously as the Fr passes below 0.77 for a bell-shaped mountain (Smith 1989). For a long mountain ridge perpendicular to the flow, stagnation at the surface is predicted to occur as the Fr passes below a value between 0.77 and 1; stagnation aloft causing wave breaking begins at an even higher Fr. For the island of Taiwan with an elongated topography, the Fr for the prefrontal southwesterly monsoon flow is small [<O(1)]. The incoming southwesterly flow decelerates off the southwestern coast and moves around the island. The northern branch of the splitting airflow moves down the windward orographically induced pressure ridge and accelerates downstream, resulting in a coastal jet along the northwestern coast. The cold air produced by orographic blocking is allowed to move around the orography. The central mountain range is apparently not long enough to force a mountain-parallel jet. Based on limited aircraft data for the 21–22 May barrier jet case, the force balance along the coastal jet over northwestern Taiwan is dominated by the inertial advection term and the pressure gradient force.
The principal component (PC) analysis shows that there are three dominant modes for sea level pressure patterns and two dominant modes for the surface winds over Taiwan during TAMEX (Chen and Li 1995a). These pressure and wind patterns appear in sequential order and represent the general characteristics of the surface airflow and sea level pressure patterns in response to the low-level flow change from southwesterlies to northeasterlies during frontal passages over Taiwan. Based on the timing of the maxima/minima in the PC scores of these dominant modes for eight frontal episodes (13–14 May, 16–17 May, 23–24 May, 27–28May, 1–2 June, 7–8 June, 14–15 June, and 24–25 June), Chen and Li (1995a) classified the evolution of surface airflow and sea level pressure during a frontal passage into six stages (Fig. 6). The first PC mode of the low-pass-filtered sea level pressure is characterized by a higher (lower) pressure along the southwestern coast and a lower (higher) pressure along the northeastern coast during the periods with a positive (negative) PC score. The first PC mode of the low-pass-filtered surface winds is characterized by strong southwesterlies (northeasterlies) along the western coast (depending on the sign of the corresponding PC score). The PC score of the first PC mode for sea level pressure and the first mode for surface winds are highly correlated with a correlation of 0.94, suggesting that these two modes occur almost simultaneously. During the positive (negative) peaks of the PC score of the first PC mode for the low-pass-filtered sea level pressure, the surface weather pattern exhibits a windward ridge along the southwestern (northeastern) coast and a lee side trough along the northeastern (southwestern) coast with southwesterlies (northeasterlies) along the western coast. The positive (negative) peaks of the PC score for this mode are classified as stage 2 (6). The second PC mode for the low-pass-filtered sea level pressure represents the mesolow pattern with higher pressure on the southwestern coast and lower pressure over the southeastern coast. The positive peaks of the PC score for this mode are classified as stage 3. The third PC mode for the low-pass-filtered sea level pressure exhibits a lower (higher) pressure over the northwestern coast and a higher (lower) pressure along the eastern and southeastern coast during the periods with a positive (negative) PC score. The positive (negative) peaks of the PC score for this mode occur during the intensification of the southwesterly (northwesterly) flow before the occurrence of positive (negative) peaks of the PC score for the first PC mode of the surface winds. The positive (negative) peaks of the PC score for this mode are classified as stage 1 (5). The second PC mode of the low-pass-filtered surface winds represents the transition period between the southwesterly flow and the northeasterly flow regimes. The positive peaks of the PC score for this mode occur when the surface front is over Taiwan and are classified as stage 4. The time intervals between successive stages 1–6 are 22, 12, 7, 9, and 14 h, respectively.
Except for the 14–15 June case, the barrier jets occur along the northwestern coast in the prefrontal southwesterly flow for these frontal episodes during TAMEX (Fig. 3a). For the 14–15 June case, the low-level southwesterly flow is absent in the prefrontal environment because of the presence of a tropical cyclone over the northern South China Sea (Chen and Hui 1990). Thus, no barrier jet is observed in the prefrontal atmosphere along the northwestern coast of Taiwan. In this section, we will analyze the evolution of the barrier jet along the northwestern coast for these seven frontal episodes.
The wind profile for station 46734 (23.55°N, 119.62°E, Fig. 1) within the Taiwan Strait shows that the wind speed of the low-level southwesterly flow increases significantly from stage 1 to stage 2 (Figs. 7a,b). At sea level, a windward ridge develops along the southwestern coast with a leeside trough along the eastern coast during this period (Figs. 6a,b). The windward pressure ridge reaches the maximum intensity at stage 2 (Fig. 6b). The vertical profile of the alongshore wind component (υs) at station 46685 over northwestern Taiwan shows that the barrier jet develops between stages 1 and 2 (Fig. 7d). It has a maximum value at the 1-km level during stages 1–3. The magnitude of υs reaches its maximum (∼14 m s−1) at stage 2 (Fig. 7d) when the sea level windward ridge–leeside trough pressure pattern is most significant (Fig. 6b). At this stage, υs at the 1-km level for station 46685 over northwestern Taiwan is about 7 m s−1 greater than that at station 46734 within the Taiwan Strait (Fig. 7). The significant flow acceleration downstream at the 1-km level is consistent with the large pressure gradient along the western coast (Fig. 6b) as found for the IOP 3 case (Fig. 2b).
At stage 3, the surface front arrives over northwesternTaiwan (Fig. 6c) and the barrier jet weakens (Fig. 7d). As found in the IOP 13 (24–25 June) case, the barrier jet is present along the northwestern coast ahead of the surface front in the prefrontal southwesterly flow regime (Li et al. 1997). At stage 3, the incoming southwesterly flow has the largest υn above the 2-km level (Fig. 7a) with a mesolow along the southeastern coast (Fig. 6c). The airflow aloft moves across the southern portion of the central mountain range, where the mountain peaks are lower (Fig. 1), and adiabatically descends on the lee. As a result, a mesolow forms along the southeastern coast. At stage 4, the surface front is over Taiwan with a shallow, northeasterly flow in the postfrontal region. The barrier jet disappears as the surface front arrives (Figs. 6 and 7). The evolution of the low-level winds over northwestern Taiwan and the sea level pressure patterns over the island further shows that the barrier jet occurs in the prefrontal southwesterly flow regime. It is caused by the blocking of the southwesterly flow by the island obstacle. As the upstream southwesterly flow intensifies, the windward pressure ridge along the western coast continues to rise. Thus, the intensity of the barrier jet increases.
During stages 5 and 6, the northeasterlies strengthen over Taiwan with a windward ridge along the northeastern coast of Taiwan because of cold advection in low levels and orographic blocking of the northeasterly flow by the central mountain range. Without upper-air observations along the northeastern coast, it is not certain whether or not a barrier jet exists there during these stages. Analysis of the 13–14 May 1987 case by Chen and Hui (1992) reveals that under postfrontal northeasterly flow, a coastal wind maximum (∼15 m s−1) occurs along the southeastern China coast. The low-level northeasterlies impinge on the coastal mountains along the southeastern China coast with a pressure ridge along the coast. These features are similar to the cold-air damming situation on the eastern slopes of Appalachians (Forbes et al. 1987). However, the warm overrunning air riding up over the dammed cold air and crossing the mountain is absent because at the 850-hPa-level winds have a westerly wind component along the southeastern China coast.
4. Upstream conditions for barrier jet events
To examine the characteristics of the upstream flow for the barrier jet events, we examine the following parameters for the upstream low-level southwesterly flow during TAMEX: the wind component normal to the mountain (υn), Brunt–Väisälä frequency (N), and the Froude number (Fr). During TAMEX, routine sounding data at 12-h intervals are available at mandatory and significant levels for stations 59134 (24.40°N, 118.00°E) and 59316 (23.30°N, 116.60°E) along the southeastern China coast, and at 5-hPa intervals for station 46734 (23.55°N, 119.62°E) over the southern Taiwan Strait (Fig. 1). The average of the υnwind component at the 850-hPa level for stations 59134 and station 59316 and N for the layer of 950-850 hPa at station 46734 are used to estimate Fr of the upstream flow (Fig. 8). Although hilly terrain with mountain peaks of about 1 km is present over southeastern China, the 850-hPa winds at stations 59134 and 59316 for all barrier jet cases during TAMEX are consistent with synoptic-scale pressure patterns (Fig. 9). Thus, the 850-hPa winds observed at these two stations represent the upstream synoptic-scale airflow reasonably well. The reason for using N at station 46734 is that we do not have sufficient vertical resolution to compute N from mandatory- and significant-level sounding data at stations 59134 and 59316. Significant-level sounding data are not always available and are not adequate for our calculations of the stability. The mountain height H is taken as 2.5 km. The data obtained during active convection or heavy rain periods at these stations are excluded in the computations.
Figure 8 presents the time series of the Fr for the southwesterly monsoon flow regime during TAMEX. The Brunt–Väisälä frequency N is about 1.0–1.8 × 10−2 s−1 (not shown). The variation of the Froude number is mainly related to the magnitude of υn. The barrier jet occurs along the northwestern coast when the upstream Froude number is about 0.2–0.5 (Fig. 8a). We also estimate Fr from the sounding data in the layer of 0.5–1.5 km at station 46810 (Fig. 1) over the northern South China Sea (Fig. 8b). For most cases, Fr computed from station 46810 is underestimated (Fig. 8b). This occurs if LLJ along the southeastern China coast is oriented east-northeast to west-southwest. Under this situation, the airflow impinging on the central mountain range is the LLJ along the southeastern China coast from the west-southwest (e.g., Figs. 9a and 9b). Station 46810 is south of the LLJ axis with much weaker winds. If the 850-hPa trough is oriented northeast–southwest extending to the southeastern China coast with southwesterlies along the southeastern China coast and over the northern South China Sea (e.g., Fig. 9e), Fr computed from station 46810 is comparable to that computed along the southeastern China coast (Fig. 8b). We only find one situation for which winds along the southeastern China coast during a barrier jet event cannot represent the upstream condition. This occurs when the 850-hPatrough axis is over the Taiwan Strait near the end of a barrier jet event [e.g., Fig. 3 in Li et al. (1997)]. Because the variation of N is rather small compared with the variations of υn, the good correlation between Fr computed from sounding data along the southeastern China coast and Fr computed from station 46810 suggests that the strength of the southwesterly monsoon flow over the northern South China Sea is highly correlated with the strength of the LLJ along the southeastern China coast. It appears that southwesterly monsoon flow over the northern South China Sea is modulated by the passage of the low pressure systems over the southeastern China coast. In May and early June, as an upstream upper-level short-wave trough approaches with strengthening westerlies over the Tibetan Plateau, low pressure systems frequently form in the lee side of the plateau and move southeastward (Chen and Chen 1995).
For the IOP 3 case, the barrier jet occurs when the upstream southwesterly flow strengthens as a low-level trough approaches the southeastern China coast (Fig. 2). We also examine the low-level synoptic-scale winds andsubjectively analyze geopotential height patterns for the remaining six barrier jet cases. For all these cases the barrier jet occurs when an 850-hPa trough moves toward the southeastern China coast (Fig. 9) with a LLJ (Chen and Yu 1988; Chen and Chen 1995) ahead of a low-level pressure trough. The barrier jet is a result of blocking of the stably stratified flow by the island obstacle under a small–Froude number [< O(1)] flow regime.
A comparison between the barrier jet and the LLJ is summarized in Table 2. The barrier jet maximum altitude is at the top of a weakly stable boundary layer (∼1 km), about 1.2 km lower than the LLJ height. The vertical wind shear is about 10 × 10−3 s−1 below the barrier jet level and 4 × 10−3 s−1 above, respectively. The shear magnitudes below the barrier jet level are about twice the shear below the LLJ (Table 2). The barrier jet is parallel to the topography, whereas the orientation ofthe LLJ varies from case to case depending on the orientation of the subsynoptic-scale 850-hPa pressure trough (Chen and Yu 1988; Chen and Chen 1995). With limited data, the exact length and width of the barrier jet are difficult to determine. Based on the aircraft data and sounding observations during TAMEX, it is apparent that the barrier jet is localized in nature. The barrier jet has a maximum speed near the northwestern coast. The alongshore wind speed decreases about 50% (40%) 90 km downstream (120 km upstream). Using scale analysis (Overland and Bond 1995), the Rossby radius of deformation is on the order of 100 km, which characterizes the offshore width of the barrier jet. Based on the sounding data at station RCHY, about 90 km offshore of the northwestern Taiwan coast, the speed of the alongshore wind is approximately 40% less than that at land station 46685.
A schematic diagram for the jet formation based on our analysis is presented in Fig. 10. When a large-scale low pressure system moves eastward toward the southeastern China coast, a strong southwesterly flow is present ahead of the pressure trough upstream of Taiwan (Fig. 10a). The barrier jet occurs along the northwestern coast of Taiwan when the subsynoptic-scale LLJ impinges on the central mountain range. A windward ridge develops along the southwestern coast with a pressure trough on the lee side as a result of stably stratified airflow past an island obstacle under a small–Froude number [<O(1)] flow regime. The low-level incoming southwesterly flow decelerates off the southwestern coast when it encounters the orographically induced high pressure there. The flow splits offshore of the coast and moves around the island (Fig. 10b). Along the western coast, the deflected airflow accelerates northward with a large cross-contour wind component down the pressure gradient, resulting in locally strong alongshore winds (or the barrier jet) over northwestern Taiwan.Above the average height of the central mountain range, the windward pressure ridge is not significant. The incoming southwesterly flow is only slightly deflected by the mountain range (Fig. 10c).
During the early summer rainy season over Taiwan, the blocking of the low-level southwesterly flow by the island obstacle is significant. In the past decade, the subsynoptic-scale LLJ was studied extensively, but the orographic influence on the flow around the island is not well understood. Our analysis shows that the barrier jet is a local feature over northwestern Taiwan and is different from the well-known low-level jet defined by Chen and Yu (1988). The barrier jet has a maximum wind speed (∼14 m s−1) at about 1 km above the surface with a vertical wind shear approximately 10 × 10−3 s−1 below and 4 × 10−3 s−1 above. During TAMEX, the southwesterly monsoon flow strengthens over Taiwan when the low-level pressure trough/surface front moves toward the southeastern China coast. The barrier jet occurs under the prefrontal southwesterly monsoon flow with a windward ridge–leeside trough pressure pattern. This is a result of the stably stratified airflow past an island obstacle under a small–Froude number [<O(1)] flow regime. The barrier jet reaches its maximum intensity when the orographic blocking of the LLJ by the island obstacle is most significant. It weakens as a surface front arrives over northwestern Taiwan and dissipates when the surface front is over the island.
Although this study shows a close relationship between the barrier jet and the pressure gradient along the western coast, a detailed diagnostic analysis of the force balance from the primitive equation is needed to fully understand the physical processes responsible for the formation of the jet. This requires dense data coverage not only along the shore but also in the direction normal to the shore. A high-resolution mesoscale model will also be used in the future to simulate the barrier jet and to assess the physical mechanisms responsible for its formation.
We would like to thank all participants involved in the planning and execution of TAMEX. The authors benefitted from discussions with Drs. W.-C. Lee (NCAR), G. M. Barnes, and D. E. Stevens. Part of the computing resources is supported by the Scientific Computing Division of the National Center for Atmospheric Research, which is sponsored by the National Science Foundation. The authors wish to thank Dr. J. Doyle and the other, anonymous, reviewer for their constructive comments. This work is supported by the National Science Foundation Grant ATM-9421060.
Corresponding author address: Dr. Yi-Leng Chen, Department of Meteorology, SOEST, University of Hawaii at Manoa, Honolulu, HI 96822.