1. Introduction
The impact of weather on the operation and safety of aviation is a worldwide issue (Humphreys 1930; WMO 1989). Although the direct causes of accidents are most commonly related to human error, weather is often a primary contributing factor for aviation accidents (Helmreich 1997; Kulesa 2003). A number of weather phenomena, such as mountain waves and thunderstorms, have been shown to produce hazardous circumstances that may lead to fatal aircraft accidents (Wurtele 1970; Fujita and Byers 1977; Wilson et al. 1984; Smith 1986; Haddad and Park 2010; Keller et al. 2015). In particular, weather-induced intense rainfall, high wind shear or strong downdrafts at low altitudes near airports can significantly affect the aircraft takeoff/landing safety (Kessler 1985). The effective documentation and prediction of aviation weather hazards are thus critically important for the prevention of aviation accidents (Chun et al. 2017).
Tropical cyclones (TCs) are not only one of the most life-threatening and destructive natural phenomena on Earth but are also well known to strongly affect air traffic and airport operation in areas where TCs pass by (Breslin 2016; Goodman and Small Griswold 2019). A mature TC is an approximately circular, strong cyclonic vortex, and its associated precipitation is typically characterized by an organized, banded feature called “rainbands” or “spiral bands” (i.e., tropical cyclone rainbands, TCRs) (Senn and Hiser 1959; Willoughby et al. 1984; Marks 2003; Yu and Chen 2011). It is well recognized that the inner core, which is approximately within 100–200 km or 2–3 times the radius of maximum wind (RMW) from the TC center, is the most hazardous region for TCs because it contains the most intense swirling winds and eyewall convection (Anthes 1982; Willoughby 1988; Rozoff et al. 2006; Wang 2009; Houze 2010). Currently, both TC location and movement can be monitored and predicted effectively by Doppler radar and satellite observation systems and numerical models. It is thus practically possible for meteorological forecasters to issue an appropriate, lead-time warning for the approach of hazardous, inner-core circulations of TCs. This weather alert usually allows aviation controllers to direct aircraft to a safer space or hold airplanes on the ground over a sufficient time period beforehand (Goodman and Small Griswold 2019).
In contrast to the inner core of TCs, both convective phenomena in the outer region of TCs and their potential impacts on aviation activities have not received much attention. The outer vicinity of TCs exhibits weaker swirling winds, and the moist convection in this region is not significantly filamented or constrained by the inner-core vortex (Rozoff et al. 2006). However, the outer region of TCs tends to possess larger convective available potential energy (CAPE) and lower humidity than the inner-core environment (Frank 1977; Bogner et al. 2000; Yu and Chen 2011; Molinari et al. 2012). These environmental conditions facilitate intense convection, making the structural characteristics of outer TCRs (OTCRs) resemble severe thunderstorms such as squall lines (Houze 2010; Eastin et al. 2012; Yu and Chen 2011; Yu and Tsai 2013; Tang et al. 2014; Moon and Nolan 2015). More recently, a comprehensive investigation of OTCRs by Yu et al. (2018, hereafter YU18) analyzed a large set of 50 rainband cases through dual-Doppler observations and identified a frequent similarity (58%, 29 rainband cases) between OTCRs and squall lines. These squall-line-like OTCRs are generally characterized by convective precipitation, an obvious convergence zone between the band-relative rear-to-front flow and front-to-rear flow at low levels and a surface cold pool signature. The processes responsible for the initiation of OTCRs have been partially addressed in the literature. Limited research evidence suggests that the origin of OTCRs is probably related to different scenarios and forcings, such as the outer propagation of inner-core convective activities, the intensification of convectively generated cold pools and the potential interaction of inner-core vortex circulation with its outer environmental flow (Willoughby et al. 1984; Yu and Cheng 2014; Li et al. 2017; Yu et al. 2019; Li et al. 2019).
Unlike the inner-core, hazardous region with a quasi-circular geometry that is well recognized and can be appropriately located given a known TC center, our awareness and understanding of aviation weather hazards caused by OTCRs is relatively less adequate. In particular, the detailed aspects of OTCR-produced severe weather conditions and how they affect aircraft safety have been neither described nor elaborated in the literature. It should be noted that turbulence associated with OTCRs can lead to very rough flights, which are well recognized by pilots and scientists who flew into TCs (Chapter 10, Houze 2014). Schaefer et al. (1992) noted that moderate turbulence is frequently located in transverse waves emanating from the OTCRs. It is possible that the impacts on aviation for both the inner and outer regions of TCs would be equally important.
On 23 July 2014, a commercial aircraft (model: ATR72–212A) with 2 pilots, 2 cabin crew, and 54 passengers on board executed regular public transport service (flight number GE222) from Kaohsiung (KH) International Airport located in southern Taiwan to Ma-Gong (MG) Airport on Penghu Island off the southwestern coast of Taiwan (Fig. 1). The aircraft crashed near MG Airport as it struggled to land in the stormy weather caused by the OTCRs of Typhoon Matmo (2014). Unfortunately, 4 flight crew and 44 passengers were killed in this airplane accident. As noted in the investigation report of this accident by the Taiwan Aviation Safety Council (TASC 2016), during the landing of the aircraft, meteorological conditions near MG Airport, including thunderstorm activities, heavy rain and significant changes in visibility, wind direction and speed, were among the contributing factors for the cause of the aircraft incident. The objective of this study is to use various available observations to document the detailed aspects of the airflow and precipitation of OTCRs related to this accident and to identify how these mesoscale structural characteristics contribute to the occurrence of hazardous weather conditions that may impact aviation safety. Over the accident area, there is relatively good, persistent coverage of temporal and spatial high-resolution measurements from two ground-based Doppler radars at Chi-Gu (CG) and MG Airport (see Fig. 1). These Doppler radar observations provide an unparalleled depiction of the finescale rainband features of the OTCRs and their relationship with the aerial incident.
(a) Best track of Typhoon Matmo (2014) from the Central Weather Bureau of Taiwan. The position of the typhoon center is indicated by solid gray circles every 3 h, with dark (light) shading indicating moderate (weak) intensity of the typhoon. The entire flight track for GE222 is indicated by the red curve. The locations of the Chi-Gu (CG) Doppler radar and Kaohsiung (KH) International Airport are denoted by the triangle and the solid circle, respectively. The CG radar is located at (x, y) = (0, 0). The arrow highlights the location of Penghu Island. (b) Enlarged map of Penghu Island. The runway of Ma-Gong Airport is indicated by the thick black line. The track of GE222 within the domain and the crash location of the aircraft are indicated by the red curve and red solid star, respectively. The location of the Ma-Gong (MG) Doppler radar is indicated by the triangle, and the locations of the surface station at Penghu (PH) and the automated weather observing system (AWOS) station are indicated by squares.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
(a) Best track of Typhoon Matmo (2014) from the Central Weather Bureau of Taiwan. The position of the typhoon center is indicated by solid gray circles every 3 h, with dark (light) shading indicating moderate (weak) intensity of the typhoon. The entire flight track for GE222 is indicated by the red curve. The locations of the Chi-Gu (CG) Doppler radar and Kaohsiung (KH) International Airport are denoted by the triangle and the solid circle, respectively. The CG radar is located at (x, y) = (0, 0). The arrow highlights the location of Penghu Island. (b) Enlarged map of Penghu Island. The runway of Ma-Gong Airport is indicated by the thick black line. The track of GE222 within the domain and the crash location of the aircraft are indicated by the red curve and red solid star, respectively. The location of the Ma-Gong (MG) Doppler radar is indicated by the triangle, and the locations of the surface station at Penghu (PH) and the automated weather observing system (AWOS) station are indicated by squares.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
(a) Best track of Typhoon Matmo (2014) from the Central Weather Bureau of Taiwan. The position of the typhoon center is indicated by solid gray circles every 3 h, with dark (light) shading indicating moderate (weak) intensity of the typhoon. The entire flight track for GE222 is indicated by the red curve. The locations of the Chi-Gu (CG) Doppler radar and Kaohsiung (KH) International Airport are denoted by the triangle and the solid circle, respectively. The CG radar is located at (x, y) = (0, 0). The arrow highlights the location of Penghu Island. (b) Enlarged map of Penghu Island. The runway of Ma-Gong Airport is indicated by the thick black line. The track of GE222 within the domain and the crash location of the aircraft are indicated by the red curve and red solid star, respectively. The location of the Ma-Gong (MG) Doppler radar is indicated by the triangle, and the locations of the surface station at Penghu (PH) and the automated weather observing system (AWOS) station are indicated by squares.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
2. Data
As described in the Introduction section, the primary datasets used to document the detailed features of the airflow and precipitation of Matmo’s OTCRs and their connection to the occurrence of aviation weather hazards are provided by two ground-based Doppler radars available in the surrounding area of the flight accident (locations in Fig. 1). One is the S-band (10 cm) operational Doppler radar of the Central Weather Bureau at CG located at the coast of southwestern Taiwan, approximately 65 km southeast of Penghu Island (Fig. 1a). The other is the C-band (5 cm) operational Doppler radar of the Weather Wing of the Chinese Air force at MG Airport. As indicated in Fig. 1b, this radar site is located ~1 km immediately adjacent to the eastern flank of the runway of MG Airport. MG Airport has a single runway oriented north-northeast to south-southwest, designated R20 and R02, respectively. The altitudes of the CG and MG radar sites are 38 and 48 m MSL, respectively. The CG (MG) radar is operated with a temporal interval of 7.5(10) min between each volume and a maximum observational range of 188 (120) km. The observational range of the non-Doppler scanning mode for the CG radar can be extended to ~460 km. The detailed characteristics of the CG and MG radars are described in Yu and Cheng (2013) and YU18. These two Doppler radars provide a continuous and comprehensive view of precipitation and airflow information over the coastal area of southwestern Taiwan during the occurrence of the aircraft accident.
Other data sources used in this study include 1-min temporal resolution surface observations from the Penghu (PH) station located on the western coast of Penghu Island and the automated weather observing system (AWOS) station located close to the northern runway threshold (see Fig. 1b) and 1–4-s-temporal-resolution flight-level aircraft data retrieved from the flight data recorder (FDR) of GE222. The FDR data used for analysis herein include altitude, longitude, latitude, and horizontal wind information. In addition, the National Centers for Environmental Prediction (NCEP) Climate Forecast System Reanalysis (CFSR) data (0.5° × 0.5°) (Saha et al. 2010) are also analyzed to provide a general context of large-scale, environmental flow for this particular event.
3. Flight overview and Matmo’s rainbands
Flight GE222 departed KH International Airport and headed for Penghu Island at 0945 UTC 23 July 2014. Shortly after ~1.3 h, the aircraft failed to land at MG Airport and crashed into the residential area approximately 625 m east-northeast of the runway threshold at 1106 UTC. The entire flight track for GE222 (in red) is indicated in Fig. 1a. During the flight period, the cyclone center of Typhoon Matmo had already made landfall on the southeastern coast of China (Fig. 1a), and its intensity continued to weaken, with a maximum wind speed less than 33 m s−1 (i.e., a weak typhoon).
The low-level plan position indicator (PPI) scans of radar reflectivity from the CG radar valid at 0945 UTC (i.e., the take-off time) and 1108 UTC (i.e., close to the accident time) indicate a highly asymmetric pattern of precipitation with prominent rainbands confined to the southeastern quadrant and outer region of the typhoon (Fig. 2). The observed asymmetry in the precipitation field has been recognized as a common rainfall distribution for a westward-moving typhoon passing over Taiwan, as its outer circulation interacts with the summer southwesterly monsoon active over the South China Sea (Yu and Cheng 2013, 2014). Consistent with this scenario, the large-scale environment for the present case was characterized by strong southwesterly monsoonal flow (~10–20 m s−1) prevailing at low levels over the South China Sea, as revealed by the wind analysis of the NCEP-CFSR data valid at 1200 UTC 23 July 2014 (Fig. 3a). A flow confluence between the outer circulation of Matmo and the southwesterly monsoon was evident over the oceanic region near Penghu Island and off the southwestern coast of Taiwan, where low-level convergence prevailed (Fig. 3b). The presence of this larger-scale confluent zone corresponded to the vigorous activities of the OTCRs observed there, as shown in Fig. 2.
The low-level PPI scan (0.5° elevation) of radar reflectivity (dBZ, color shading) from the non-Doppler scanning mode of the CG radar at (a) 0945 and (b) 1108 UTC. The locations of outer tropical cyclone rainbands (OTCRs) associated with Matmo are marked with R1, R2, and R3. The corresponding typhoon center located ~50–80 km inland of southeastern China is also indicated. Range rings (km) with respect to the typhoon center are also indicated. The location of the CG radar is indicated by the triangle. In (b), the thick dashed line indicates the band-normal segment for calculating the time–distance sections of radar reflectivity and radial velocity from the CG radar shown in Figs. 4 and 10.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
The low-level PPI scan (0.5° elevation) of radar reflectivity (dBZ, color shading) from the non-Doppler scanning mode of the CG radar at (a) 0945 and (b) 1108 UTC. The locations of outer tropical cyclone rainbands (OTCRs) associated with Matmo are marked with R1, R2, and R3. The corresponding typhoon center located ~50–80 km inland of southeastern China is also indicated. Range rings (km) with respect to the typhoon center are also indicated. The location of the CG radar is indicated by the triangle. In (b), the thick dashed line indicates the band-normal segment for calculating the time–distance sections of radar reflectivity and radial velocity from the CG radar shown in Figs. 4 and 10.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
The low-level PPI scan (0.5° elevation) of radar reflectivity (dBZ, color shading) from the non-Doppler scanning mode of the CG radar at (a) 0945 and (b) 1108 UTC. The locations of outer tropical cyclone rainbands (OTCRs) associated with Matmo are marked with R1, R2, and R3. The corresponding typhoon center located ~50–80 km inland of southeastern China is also indicated. Range rings (km) with respect to the typhoon center are also indicated. The location of the CG radar is indicated by the triangle. In (b), the thick dashed line indicates the band-normal segment for calculating the time–distance sections of radar reflectivity and radial velocity from the CG radar shown in Figs. 4 and 10.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
The large-scale wind distributions at 1 km MSL from the NCEP-CFSR data valid at 1200 UTC 23 Jul 2014. (a) Wind speed (m s−1) and (b) convergence (×10−4 s−1) are indicated by color shading.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
The large-scale wind distributions at 1 km MSL from the NCEP-CFSR data valid at 1200 UTC 23 Jul 2014. (a) Wind speed (m s−1) and (b) convergence (×10−4 s−1) are indicated by color shading.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
The large-scale wind distributions at 1 km MSL from the NCEP-CFSR data valid at 1200 UTC 23 Jul 2014. (a) Wind speed (m s−1) and (b) convergence (×10−4 s−1) are indicated by color shading.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
Three OTCRs (highlighted by R1, R2, and R3 in Fig. 2a) located ~200–300 km from the typhoon center were well identified and exhibited highly organized, elongated features of high reflectivities (>45 dBZ). These rainbands were oriented at some small angles to the circle about the TC center, similar to the typical geometry of OTCRs observed previously (Anthes 1982; Yu et al. 2019). Over the duration of 5.5 h from 0700 to 1230 UTC 23 July encompassing the flight period of GE222 (0945–1106 UTC), the weather on Penghu Island was significantly influenced by the passage of these three rainbands as they propagated northeastward from the offshore region west of Penghu Island to the western/northwestern coast of Taiwan. This scenario can be best depicted by the time–distance section of low-level reflectivities taken from a band-normal segment (marked in Fig. 2b) from the CG radar, as shown in Fig. 4. MG Airport (location highlighted by the vertical solid line in Fig. 4) clearly experienced an intermittent, heavy precipitation caused by R1, R2, and R3, with relatively weaker rainfall in between. According to the flight arrival information provided from MG Airport, seven other commercial aircraft successfully landed during this period. Unlike GE222, these flights during their landing effectively avoided the hazardous timing of the most intense precipitation observed within R1–3 (Fig. 4). Three of these flights (FE081, FE031, and B7647) landed near the leading edge of R1–3 but they actually experienced less threatening wind conditions compared to GE222, as will be elaborated in section 4.
Time–distance section of low-level PPI scan (0.5° elevation) of reflectivities (dBZ, color shading) from the CG radar along the band-normal segment indicated in Fig. 2b during 0700 to 1230 UTC 23 Jul. Three observed OTCRs are marked with R1, R2, and R3. The vertical solid line indicates the position of the runway (R20) of MG Airport. The horizontal dashed line highlights the accident time of GE222. The arrival times of several commercial aircraft that landed at MG Airport during the analysis period are marked by hollow circles with flight numbers.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
Time–distance section of low-level PPI scan (0.5° elevation) of reflectivities (dBZ, color shading) from the CG radar along the band-normal segment indicated in Fig. 2b during 0700 to 1230 UTC 23 Jul. Three observed OTCRs are marked with R1, R2, and R3. The vertical solid line indicates the position of the runway (R20) of MG Airport. The horizontal dashed line highlights the accident time of GE222. The arrival times of several commercial aircraft that landed at MG Airport during the analysis period are marked by hollow circles with flight numbers.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
Time–distance section of low-level PPI scan (0.5° elevation) of reflectivities (dBZ, color shading) from the CG radar along the band-normal segment indicated in Fig. 2b during 0700 to 1230 UTC 23 Jul. Three observed OTCRs are marked with R1, R2, and R3. The vertical solid line indicates the position of the runway (R20) of MG Airport. The horizontal dashed line highlights the accident time of GE222. The arrival times of several commercial aircraft that landed at MG Airport during the analysis period are marked by hollow circles with flight numbers.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
At the take-off time of GE222 (i.e., 0945 UTC), R1 nearly made landfall on the northwestern coast of Taiwan, while the leading edge of R2 just passed over Penghu Island (Figs. 2a and 4). As noted in the investigation report of GE222, bad weather observed at MG Airport during this time was actually beyond the landing circumstances for an aircraft (TASC 2016), indicating the significant impact of the intense convection of R2 on aviation safety. With time, weather conditions near MG Airport improved temporarily as R2 soon moved away from Penghu Island. At 1106 UTC, R2 already touched the western coast of Taiwan (Fig. 2b); however, Penghu Island at this time experienced the arrival of intense precipitation associated with R3, the target rainband influencing the flight GE222 (Figs. 2b and 4).
Figures 5a–d illustrate more details about the flight trajectory of GE222 and its relative location to the observed R2 and R3 during which the aircraft approached Penghu Island and attempted to land at MG Airport. For reference, the time series of the FDR-recorded flight altitudes and corresponding along-track reflectivity values extracted from the low-level PPI scans (0.5° elevation) of the CG radar are also shown in Fig. 5e. Note that the along-track reflectivity values should be reasonably representative of the precipitation intensities encountered by GE222 because the heights of radar beams for the extracted reflectivities (0.4–1.3 km MSL) are well within the range of flight altitudes (0–2 km MSL).
The flight track of GE222 (thick solid curve) superposed on the low-level PPI scan (0.5° elevation) of radar reflectivity (dBZ, color shading) from the CG radar at four selected times: (a) 1024, (b) 1039, (c) 1054, and (d) 1109 UTC. In (a)–(d), the positions of GE222 over the duration from approximately 1000 UTC to the analysis time of each panel are also highlighted by solid gray circles every 5 min. The observed OTCRs are marked (i.e., R2 and R3), and the location of the CG radar is denoted by the triangle. (e) The time series of the FDR-recorded flight altitudes (solid curve) and corresponding along-track reflectivity values (dashed curve) extracted from the low-level PPI scans (0.5° elevation) of the CG radar. For reference, the analysis times of (a)–(c) and the crash time (i.e., 1106 UTC) are marked by red arrows.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
The flight track of GE222 (thick solid curve) superposed on the low-level PPI scan (0.5° elevation) of radar reflectivity (dBZ, color shading) from the CG radar at four selected times: (a) 1024, (b) 1039, (c) 1054, and (d) 1109 UTC. In (a)–(d), the positions of GE222 over the duration from approximately 1000 UTC to the analysis time of each panel are also highlighted by solid gray circles every 5 min. The observed OTCRs are marked (i.e., R2 and R3), and the location of the CG radar is denoted by the triangle. (e) The time series of the FDR-recorded flight altitudes (solid curve) and corresponding along-track reflectivity values (dashed curve) extracted from the low-level PPI scans (0.5° elevation) of the CG radar. For reference, the analysis times of (a)–(c) and the crash time (i.e., 1106 UTC) are marked by red arrows.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
The flight track of GE222 (thick solid curve) superposed on the low-level PPI scan (0.5° elevation) of radar reflectivity (dBZ, color shading) from the CG radar at four selected times: (a) 1024, (b) 1039, (c) 1054, and (d) 1109 UTC. In (a)–(d), the positions of GE222 over the duration from approximately 1000 UTC to the analysis time of each panel are also highlighted by solid gray circles every 5 min. The observed OTCRs are marked (i.e., R2 and R3), and the location of the CG radar is denoted by the triangle. (e) The time series of the FDR-recorded flight altitudes (solid curve) and corresponding along-track reflectivity values (dashed curve) extracted from the low-level PPI scans (0.5° elevation) of the CG radar. For reference, the analysis times of (a)–(c) and the crash time (i.e., 1106 UTC) are marked by red arrows.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
After take-off, GE222 flew northbound and remained at altitudes of 1.8–2.1 km MSL heading toward the oceanic area of Penghu Island (Figs. 5a,e). When the aircraft initially approached Penghu Island, it passed through the southern end of R2 (Fig. 5a), where a local maximum of the along-track reflectivity (~40 dBZ) at ~1007 UTC was also evident, as shown in Fig. 5e. After the transverse, GE222 flew with a circle-like, holding flight pattern in regions ~30 km northeast of MG Airport and waited for approach clearance for runway R20 (Fig. 5a), due to the low visibility of ~800 m at the airport during the passage of R2. Note that because of different landing assistance systems, runway R20 (R02) required a higher (lower) visibility limitation of 1600 (800) m for aircraft to land; see TASC (2016). Shortly after, the pilots of GE222 then chose to execute an R02 approach, and the aircraft started to track southbound after 1030 UTC over regions of relatively weak echoes (~25–30 dBZ) ahead of R3 (Figs. 5b,e). At pretty much the same time, weather conditions and visibility improved at MG Airport, as it was just located in the gap region of precipitation between R2 and R3 (Figs. 5b,c). Subsequently, the flight crew of GE222 decided to take an R20 approach again for landing, so the aircraft turned back to the north with decreasing altitude and arrived near the northern end of R3 (Figs. 5c,e).
During the final stage of the flight, GE222 headed southwestward (i.e., roughly parallel to the orientation of runway R20, Fig. 1b) and kept descending immediately along the rear flank of the leading edge of R3, where strong reflectivities (>40–45 dBZ) prevailed (Fig. 5d). A fatal deviation of the flight track to the east from the runway (Fig. 1b) was recorded by the FDR as GE222 arrived at the northern shore of Penghu Island and flew into the zone of the most intense precipitation (~50 dBZ) observed in the middle segment of R3 (Figs. 5d and 1b). As will be demonstrated later, the eastward deviation of the flight track was consistent with the presence of strong westerly winds produced locally by R3. At almost the same time, the visibility near the runway, based on the AWOS measurements, was reduced very rapidly to ~250 m due to the arrival of the heavy rainfall of R3. The flight crew of GE222 was unable to visually locate the runway within the extremely low visibility environment before the aircraft descended to the minimum descent altitude1 (MDA, 300 ft; ~90 m) for runway R20 (TASC 2016). In the last minute, the pilots attempted to execute a go-around procedure until the aircraft hit the ground. At the accident time (1106 UTC), the along-track reflectivity was also observed to reach a peak value of ~48 dBZ (Fig. 5e). It is clear that R3, one of the OTCRs from Matmo, represents a critical, hazardous precipitation system during the occurrence of this unfortunate flight accident.
4. Finescale structures of OTCRs
This section focuses primarily on the finescale structural analysis of R3, followed by a brief description of the other two OTCRs (i.e., R1 and R2). As shown in Fig. 5, R3 was characterized not only by very intense precipitation but also by a sharp gradient of reflectivities along its leading edge. The flight-level winds recorded by the FDR of GE222 further show a prominent low-level wind alternation across the leading edge of the rainband, as shown in Fig. 6. Strong south-southwesterly flow (20–25 m s−1) prevailed in the prerainband region, which was associated with environmental, monsoonal southwesterly flow originating over the South China Sea, as shown in Fig. 3. The low-level winds became more westerly (~18–20 m s−1) within the rainband, particularly inside the zone of the most intense precipitation (maximum > 50 dBZ) near Penghu Island. The cross-band wind shift was also evident but relatively gentle over the northern segment of R3 with discrete reflectivity elements (>40 dBZ). Considering the runway oriented roughly parallel to the prerainband south-southwesterly flow, the existence of the westerly flow observed immediately to the rear of the leading edge of R3 is expected to yield a significant wind component perpendicular to the runway (i.e., crosswind). The intensity of the calculated crosswind could reach a maximum of ~18 m s−1, a magnitude close to or slightly beyond the landing crosswind limitation of the aircraft model of GE222 [(Avions de Transport Régional) ATR 2010].
Flight-level winds recorded by the FDR as GE222 flew from the prerainband vicinity of R3 to the region to the rear of the leading edge of R3. The low-level PPI scan (0.5° elevation) of radar reflectivity (dBZ) from the CG radar valid at 1109 UTC is indicated by gray shading. Wind flags correspond to 25 m s−1, full wind barbs correspond to 5 m s−1, and half barbs correspond to 2.5 m s−1. Line segment A–B marks the location of the vertical cross sections shown in Fig. 9.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
Flight-level winds recorded by the FDR as GE222 flew from the prerainband vicinity of R3 to the region to the rear of the leading edge of R3. The low-level PPI scan (0.5° elevation) of radar reflectivity (dBZ) from the CG radar valid at 1109 UTC is indicated by gray shading. Wind flags correspond to 25 m s−1, full wind barbs correspond to 5 m s−1, and half barbs correspond to 2.5 m s−1. Line segment A–B marks the location of the vertical cross sections shown in Fig. 9.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
Flight-level winds recorded by the FDR as GE222 flew from the prerainband vicinity of R3 to the region to the rear of the leading edge of R3. The low-level PPI scan (0.5° elevation) of radar reflectivity (dBZ) from the CG radar valid at 1109 UTC is indicated by gray shading. Wind flags correspond to 25 m s−1, full wind barbs correspond to 5 m s−1, and half barbs correspond to 2.5 m s−1. Line segment A–B marks the location of the vertical cross sections shown in Fig. 9.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
The more detailed views of the airflow and precipitation of R3 near Penghu Island can be further revealed by high-resolution observations from the CG and MG radars. It is possible to perform the dual-Doppler wind synthesis using the CG and MG radar observations for the present study. However, based on the consideration of synthesized geometries, the cross-beam angles between the two radars over the region of interest (i.e., near Penghu Island) are very small (Fig. 1), which is expected to produce large uncertainties and errors in the dual-Doppler-derived winds. In view of this, this study focuses only on single Doppler radar analysis. Figure 7 shows the low-level PPI scans (0.5° elevation) of reflectivity and radial velocity from the CG radar at two consecutive times (1101 and 1109 UTC) encompassing the accident time of GE222 (i.e., 1106 UTC). The low-level PPI scan (0.5° elevation) of radial velocity from the MG radar at 1057 UTC is also shown in Fig. 8 for reference. The heights of the radar beams from the 0.5° PPI scan of the CG (MG) radar calculated to be ~850 (~100) m MSL over regions of Penghu Island are low enough to provide useful information on the low-level flow encountered by GE222.
The low-level PPI scan (0.5° elevation) of (a) radar reflectivity (dBZ, color shading) and (b) radial velocities (m s−1, color shading) from the CG radar valid at 1101 UTC 23 Jul 2014. Corresponding surface winds measured by the PH and AWOS stations are also shown for reference. The runway of MG Airport is indicated by the black thick line. Full wind barbs correspond to 5 m s−1; half barbs correspond to 2.5 m s−1. The azimuth (°) and radial distance (km) from the CG radar are indicated by thin solid lines and dashed lines, respectively. (c),(d) As in (a) and (b), but for analyses valid at 1109 UTC 23 Jul 2014.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
The low-level PPI scan (0.5° elevation) of (a) radar reflectivity (dBZ, color shading) and (b) radial velocities (m s−1, color shading) from the CG radar valid at 1101 UTC 23 Jul 2014. Corresponding surface winds measured by the PH and AWOS stations are also shown for reference. The runway of MG Airport is indicated by the black thick line. Full wind barbs correspond to 5 m s−1; half barbs correspond to 2.5 m s−1. The azimuth (°) and radial distance (km) from the CG radar are indicated by thin solid lines and dashed lines, respectively. (c),(d) As in (a) and (b), but for analyses valid at 1109 UTC 23 Jul 2014.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
The low-level PPI scan (0.5° elevation) of (a) radar reflectivity (dBZ, color shading) and (b) radial velocities (m s−1, color shading) from the CG radar valid at 1101 UTC 23 Jul 2014. Corresponding surface winds measured by the PH and AWOS stations are also shown for reference. The runway of MG Airport is indicated by the black thick line. Full wind barbs correspond to 5 m s−1; half barbs correspond to 2.5 m s−1. The azimuth (°) and radial distance (km) from the CG radar are indicated by thin solid lines and dashed lines, respectively. (c),(d) As in (a) and (b), but for analyses valid at 1109 UTC 23 Jul 2014.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
As in Fig. 7b, but for the low-level PPI scan (0.5° elevation) of radial velocity (m s−1) from the MG radar valid at 1057 UTC 23 Jul 2014. The leading edge of the rainband is indicated by the thick dashed line. The location of the MG radar is indicated by the triangle.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
As in Fig. 7b, but for the low-level PPI scan (0.5° elevation) of radial velocity (m s−1) from the MG radar valid at 1057 UTC 23 Jul 2014. The leading edge of the rainband is indicated by the thick dashed line. The location of the MG radar is indicated by the triangle.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
As in Fig. 7b, but for the low-level PPI scan (0.5° elevation) of radial velocity (m s−1) from the MG radar valid at 1057 UTC 23 Jul 2014. The leading edge of the rainband is indicated by the thick dashed line. The location of the MG radar is indicated by the triangle.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
A discontinuity in radial velocities from positive to negative values was observed to be coincident with the leading edge of the rainband that was marked by a very sharp horizontal gradient of reflectivities (Figs. 7a–d). The aforementioned westerly flow inside the rainband was evidently highlighted by a local area of enhanced negative radial velocities (i.e., approaching the radar) immediately behind the leading edge of the rainband (Fig. 7b). The area of enhanced negative radial velocities (i.e., westerlies) moved rapidly eastward with the leading edge of the rainband and soon arrived over the runway during the landing of GE222 (Fig. 7d). The peak intensity of the negative radial velocities found at 1101 UTC could have exceeded 18 m s−1 but slightly weakened (16–18 m s−1) immediately after the flight accident at 1109 UTC. Generally, positive but weaker radial velocities (<10 m s−1) observed in the prerainband region were consistent with the environmental south-southwesterly flow prevailing there, as shown in Fig. 6.
The velocity patterns of the MG radar seen in Fig. 8 consistently indicated the development of rainband-induced westerly flow that was characterized by a local area of pronounced negative radial velocity. With the advantage of the observational geometry of the MG radar relative to the runway, the intensity of the low-level crosswind associated with the westerly flow could be appropriately estimated from the values of radial velocity along the 290° azimuth angle (i.e., perpendicular to the runway), which was found to be ~16 m s−1 (cf. Fig. 8). This magnitude was comparable to that of the FDR measurements described earlier. The westerly flow inside the rainband was also evident at the ground level, as observed by the PH and AWOS stations (Figs. 7 and 9), although its intensity decreased to ~6–8 m s−1, presumably due to the significant influence of surface friction.
(a) Vertical cross section of radar reflectivity (dBZ, color shading) and ground-relative radial velocities (contours with an interval of 3 m s−1) from the CG radar along A–B in Fig. 6 valid at 1101 UTC 23 Jul 2014. Negative values of radial velocities are hatched, and zero radial velocities are highlighted with thick solid curves. The location of the runway (R20) is also indicated. (b) As in (a), but for band-relative radial velocities.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
(a) Vertical cross section of radar reflectivity (dBZ, color shading) and ground-relative radial velocities (contours with an interval of 3 m s−1) from the CG radar along A–B in Fig. 6 valid at 1101 UTC 23 Jul 2014. Negative values of radial velocities are hatched, and zero radial velocities are highlighted with thick solid curves. The location of the runway (R20) is also indicated. (b) As in (a), but for band-relative radial velocities.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
(a) Vertical cross section of radar reflectivity (dBZ, color shading) and ground-relative radial velocities (contours with an interval of 3 m s−1) from the CG radar along A–B in Fig. 6 valid at 1101 UTC 23 Jul 2014. Negative values of radial velocities are hatched, and zero radial velocities are highlighted with thick solid curves. The location of the runway (R20) is also indicated. (b) As in (a), but for band-relative radial velocities.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
A range height indicator (RHI) vertical cross section of the CG radar perpendicular to the orientation of R3 and passing through the location of the runway at 1101 UTC (i.e., 5 min prior to the flight accident) was used to illustrate the vertical distributions of rainband precipitation and airflow (line A–B, Fig. 6). To obtain a better approximation of horizontal winds by radial Doppler velocities in these RHI analyses, the velocity component of the particle terminal velocity projected onto the radar beam was removed from the raw radial velocities (Yu and Jou 2005; Yu and Hsieh 2009). In addition to the ground-relative radial velocities, band-relative radial velocities are also calculated and shown in Fig. 9b for reference because of the propagating nature of the rainband. The band-relative field was obtained by subtracting a band-normal propagation speed of 5.5 m s−1 from the observed radial velocities based on the propagation speed and direction of R3 at 18.1 m s−1 and 65°, respectively. In the context of the observed propagating characteristics, the eastern (western) flank of R3 may refer to the front (rear) side of the propagating convective system.
The RHI vertical cross section indicates the highly convective characteristic of R3 precipitation, with the 40-dBZ contour exceeding 5 km MSL. A narrow zone of heaviest precipitation coincided with enhanced convergence between positive and negative radial velocities at the leading edge of the rainband, with the strongest reflectivities (>50 dBZ) confined to the lowest 2.5 km MSL (Fig. 9a). The area of the enhanced westerly flow, namely, strongly negative radial velocities (>12–15 m s−1), extended vertically up to 4.5 km MSL within the primary region of the leading heavy precipitation, and relatively weaker intensities of negative radial velocities (i.e., weaker westerly flow) were observed behind. As indicated in the band-relative field (Fig. 9b), a deep layer of south-southwesterly inflow (i.e., positive band-relative radial velocities) was present in the prerainband region. The inflow had stronger intensities (~6–12 m s−1) below 3 km MSL and was characterized by a frontward vertical shear of the horizontal wind component perpendicular to the rainband. The magnitude of the calculated low-level cross-band vertical shear was approximately 3.6 m s−1 km−1. This strength of the cross-band vertical shear may be considered moderate compared to those of previously documented OTCRs (Yu and Tsai 2013; Yu et al. 2019) but was comparable to that of the prerainband environment of tropical squall lines (Barnes and Sieckman 1984). The low-level inflow fed the rainband and extended rearward (westward) and upward to higher altitudes (above 6 km MSL) immediately ahead of the leading edge of the low-level rear-to-front flow, consistent with the upward transport of horizontal momentum by leading convective updrafts (cf. Fig. 9b). The maximum intensities of the low-level rear-to-front flow inside the rainband could reach 6–9 m s−1, which was somewhat stronger than those of the radar-derived rear-to-front flow documented previously within OTCRs (3–6 m s−1) (Yu and Tsai 2013). The band-relative airflow structures shown in Fig. 9b were quite similar to the airflow patterns of squall-line-like OTCRs described in YU18. In addition, the RHI analyses, together with the results from Figs. 6–8, strongly suggest that the occurrence of hazardous crosswind (i.e., the westerly flow) encountered by GE222 was exactly a manifestation of band-relative rear-to-front flow generated locally inside the rainband.
Similar radar analyses performed for R1 and R2 indicate that these two OTCRs also exhibited squall-line-like structures, such as the highly convective nature of precipitation, a deep layer of prerainband inflow and a low-level convergence zone between the band-relative rear-to-front flow and front-to-rear flow near the leading edge of the rainbands (not shown). Despite some differences in the details of these structures among the three rainbands (R1–3), they share common characteristics of precipitation and airflow patterns. Accordingly, the time–distance section of low-level radial velocities from the CG radar over a longer time window, corresponding to that of Fig. 4, shows the repetitive appearance of negative radial velocities, namely, the westerly flow or band-relative rear-to-front flow inside R1–3, near runway R20 of MG Airport (Fig. 10). Over the prerainband regions or in the precipitation gap regions between the rainbands, positive radial velocities generally prevailed, reflecting the presence of environmental south-southwesterly flow. Note that GE222 encountered the strongest precipitation and westerly flow (i.e., crosswind) observed inside R3 during its landing (Fig. 10). Fortunately, all other commercial aircraft landed on the runway in time either prior to the arrival of the heaviest precipitation and strong crosswind associated with R1–3 or just after the passage of R3. These analyses provide important evidence that squall-line-like OTCRs represent a potentially threatening weather phenomenon hazardous to aviation safety.
As in Fig. 4, but for the ground-relative radial velocities (m s−1, color shading) from the CG radar. Positive (negative) values represent velocities away from (toward) the radar.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
As in Fig. 4, but for the ground-relative radial velocities (m s−1, color shading) from the CG radar. Positive (negative) values represent velocities away from (toward) the radar.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
As in Fig. 4, but for the ground-relative radial velocities (m s−1, color shading) from the CG radar. Positive (negative) values represent velocities away from (toward) the radar.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
5. Surface characteristics of OTCRs
To further evaluate the potential similarity of the observed OTCRs with squall lines, observations from the surface stations located on Penghu Island as the rainbands passed by were analyzed. Figure 11 shows the time series analyses of 1-min-temporal-resolution measurements recorded from the PH station (location in Fig. 1b) within a time window corresponding to that of Fig. 4. Seven surface meteorological quantities, namely, temperature (T), dewpoint temperature (Td), equivalent potential temperature (θe), perturbation pressure (p′), wind speed (Ws), wind direction (Wd), and rainfall rate (RR), are presented. To better isolate pressure fluctuations caused by the passage of the rainbands, the perturbation pressure was calculated by subtracting the cyclone-scale pressure tendency (i.e., 1-h running mean of surface pressure) from the pressure values recorded at a given time (Yu and Tsai 2010; Yu and Chen 2011). The time–height cross section of elevated radar reflectivity observed at the position of the PH station by the CG radar is also shown in Fig. 11 to provide a better context of precipitation information for locating the rainbands.
(top) Time–height cross section of radar reflectivity (dBZ, color shading) at the PH station observed by the CG radar. (middle),(bottom) Time series of p′ (mb; 1 mb = 1 hPa), T (°C), Td (°C), θe (K), Wd (°), Ws (m s−1), and rainfall intensity (mm h−1) at 1-min temporal resolution observed from the PH station (location shown in Fig. 1) from 0700 to 1230 UTC 23 Jul 2014. Note that time increases to the left so the rainband geometry (i.e., rear/front side) is consistent with that of the vertical cross section shown in Fig. 9. Ground-relative winds are also indicated by wind flags with full wind barbs corresponding to 5 m s−1 and half barbs corresponding to 2.5 m s−1. The duration encompassing the passage of the observed OTCRs (R1, R2, and R3) is highlighted by gray shading. The accident time of GE222 is marked by the red solid star.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
(top) Time–height cross section of radar reflectivity (dBZ, color shading) at the PH station observed by the CG radar. (middle),(bottom) Time series of p′ (mb; 1 mb = 1 hPa), T (°C), Td (°C), θe (K), Wd (°), Ws (m s−1), and rainfall intensity (mm h−1) at 1-min temporal resolution observed from the PH station (location shown in Fig. 1) from 0700 to 1230 UTC 23 Jul 2014. Note that time increases to the left so the rainband geometry (i.e., rear/front side) is consistent with that of the vertical cross section shown in Fig. 9. Ground-relative winds are also indicated by wind flags with full wind barbs corresponding to 5 m s−1 and half barbs corresponding to 2.5 m s−1. The duration encompassing the passage of the observed OTCRs (R1, R2, and R3) is highlighted by gray shading. The accident time of GE222 is marked by the red solid star.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
(top) Time–height cross section of radar reflectivity (dBZ, color shading) at the PH station observed by the CG radar. (middle),(bottom) Time series of p′ (mb; 1 mb = 1 hPa), T (°C), Td (°C), θe (K), Wd (°), Ws (m s−1), and rainfall intensity (mm h−1) at 1-min temporal resolution observed from the PH station (location shown in Fig. 1) from 0700 to 1230 UTC 23 Jul 2014. Note that time increases to the left so the rainband geometry (i.e., rear/front side) is consistent with that of the vertical cross section shown in Fig. 9. Ground-relative winds are also indicated by wind flags with full wind barbs corresponding to 5 m s−1 and half barbs corresponding to 2.5 m s−1. The duration encompassing the passage of the observed OTCRs (R1, R2, and R3) is highlighted by gray shading. The accident time of GE222 is marked by the red solid star.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
The results indicate quite similar surface fluctuations during the passage of the three observed rainbands (R1–3). The leading edges of the rainbands were highlighted by sharp changes in wind direction from south-southwesterly flow to westerly flow, as we have seen from the FDR and radar observations (Figs. 6–8). The winds became stronger (8–11 m s−1) in the region of most intense rainfall and gradually returned to the environmental south-southwesterly flow toward the rear edges of the rainbands. These wind alternations strongly support that the westerly flow (i.e., crosswind or band-relative rear-to-front flow) was produced locally and convectively inside the rainbands.
A prominent drop in T (~2.5°–3°C) and θe (~15 K) across the leading edge of the rainbands was evident (Fig. 11). This cold pool signature remained within the rainbands, but the temperature tended to recover to some degree shortly after the passage of the primary precipitation of the rainbands. Much lower equivalent potential temperature could be attributed to the presence of colder and less-moist air (i.e., lower T and Td) inside the rainbands. The observed reductions in both T and Td cannot be explained simply by the evaporative effect of precipitation that is also expected to be relatively minor under nearly saturated ambient conditions (T ≈ Td) for the present case. This thermodynamic feature has been shown to commonly occur for TCRs or tropical deep convection due to the downward transport of low-θe air originating from higher altitudes by convectively induced downdrafts (Barnes et al. 1983; Skwira et al. 2005; Tompkins 2001; Yu and Chen 2011).
The perturbation pressure was observed to rise notably during and after the passage of the leading edge. The maximum pressure perturbations (~0.8–1 mb) tended to coincide with the region of the heaviest precipitation and prominent temperature deficits, suggesting that the observed positive perturbation pressure (i.e., mesohigh) was primarily caused by the negative buoyancy associated with the cold pool and water loading. The horizontal extent of the mesohigh associated with R3 in the cross-band direction was estimated to be ~13 km, based on the band-normal propagation speed of this rainband (5.5 m s−1) and the duration of its associated positive pressure perturbations (~40 min, cf. Figure 11). The important role played by the mesohigh observed on the front side of R3 will be further elaborated in the next section. On the rear side of or outside the rainbands, the perturbation pressure generally had near-zero or slightly negative values. These thermodynamic characteristics observed for the present rainbands are fundamentally similar to those of squall-line-like OTCRs documented previously (Yu and Tsai 2013; YU18; Yu et al. 2019).
The precipitation associated with R1–3 was quite intense, and their maximum rainfall rates could reach above 80 mm h−1, as expected from the highly convective nature of radar echoes with a pronounced horizontal gradient and a significant vertical extent of reflectivities shown in Figs. 11 and 9. As described in section 3, rapidly decreasing visibility at MG Airport due to the arrival of R3 heavy precipitation was one of the striking and hazardous weather conditions contributing to the landing failure of GE222. Time series analyses of the visibility data recorded by the AWOS station located close to the runway (location in Fig. 1b) indicate an extremely sharp and dramatic change in visibility from values well above 2 km in the prerainband region to only 200–400 m inside the rainbands (Fig. 12). In addition to R3, both R1 and R2 similarly produced poor visibility due to their intense rainfall, which was well beyond the visibility limitation of 1.6 km for landing on runway R20, as described in section 3. GE222 encountered the most hazardous timing of the heaviest precipitation (>100 mm h−1) and lowest visibility (~200 m) caused by the arrival of R3 just after 1100 UTC.
(top) Time–height cross section of radar reflectivity (dBZ, color shading) at the AWOS station observed by the CG radar. (middle),(bottom) Time series of rainfall intensity (mm h−1) and visibility (km) at 1-min temporal resolution observed from the AWOS station (location shown in Fig. 1) from 0700 to 1230 UTC 23 Jul 2014. The duration encompassing the passage of the observed OTCRs (R1, R2, and R3) is highlighted by gray shading. The accident time of GE222 is marked by the red solid star.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
(top) Time–height cross section of radar reflectivity (dBZ, color shading) at the AWOS station observed by the CG radar. (middle),(bottom) Time series of rainfall intensity (mm h−1) and visibility (km) at 1-min temporal resolution observed from the AWOS station (location shown in Fig. 1) from 0700 to 1230 UTC 23 Jul 2014. The duration encompassing the passage of the observed OTCRs (R1, R2, and R3) is highlighted by gray shading. The accident time of GE222 is marked by the red solid star.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
(top) Time–height cross section of radar reflectivity (dBZ, color shading) at the AWOS station observed by the CG radar. (middle),(bottom) Time series of rainfall intensity (mm h−1) and visibility (km) at 1-min temporal resolution observed from the AWOS station (location shown in Fig. 1) from 0700 to 1230 UTC 23 Jul 2014. The duration encompassing the passage of the observed OTCRs (R1, R2, and R3) is highlighted by gray shading. The accident time of GE222 is marked by the red solid star.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
6. Forcing mechanism for the rear-to-front flow
The low-level rear-to-front flow (i.e., ground-relative westerly flow) generated locally within R3 not only is responsible for strong crosswinds but is also one of the primary kinematic signatures for OTCRs documented previously (Yu and Tsai 2013; YU18). This kind of low-level airflow pattern has been frequently observed in the convective region of a mature squall line (e.g., Roux et al. 1984; Smull and Houze 1985; Roux 1988; Houze et al. 1989; Wang et al. 1990; Jorgensen et al. 1997; Houze 2004). In fact, the low-level band-relative rear-to-front flow, which implies that wind speeds within the layer of cold air are substantially faster than the propagating speed of the cold-air leading edge, also represents one of the key features of atmospheric gust fronts or laboratory density currents (e.g., Charba 1974; Goff 1976; Simpson and Britter 1980). Although the existence of the low-level rear-to-front flow has been commonly considered to be related to the convectively generated outflow from the existing convective cells (e.g., Wakimoto 1982), the forcing mechanism of the low-level rear-to-front flow for OTCRs and squall lines remains ambiguous in the literature. For squall-line systems, the elevated rear-to-front flow, known as the so-called “rear inflow,” is usually present in the low- to midtroposphere (or above the layer of the cold pool) behind the leading convection or over the trailing stratiform region. The forcing mechanism for the elevated rear-to-front flow has been shown to be closely related to the buoyancy-induced low pressure underneath the warm updrafts that slopes over the low-level cold pool (LeMone et al. 1984; Lafore and Moncrieff 1989; Klimowski 1994). The resulting pressure difference between the back and leading portions of the convective system may act to accelerate the midlevel air from the rear to the front. However, whether this forcing mechanism can substantially contribute to the development of low-level rear-to-front flow is uncertain because the elevated rear inflow and the low-level rear-to-front flow would probably belong to a different circulation system within squall lines.
Another potential forcing mechanism for the low-level rear-to-front flow is related to mesoscale vortices, if any, developing along convective rainbands, which have been found to reinforce the rear-to-front flow locally to produce arc-shaped or bow-shaped radar echoes (Weisman 1993; Weisman and Davis 1998; Yu and Tsai 2013; Wakimoto et al. 2015). For the present case, the velocity signatures seen from radar observations did indicate evidence of mesoscale vortices located near the northern end of R3 and the southern end of R2 (figure not shown). Nevertheless, these line-end vortices were small in nature (~10 km in radius) and did not appear directly relevant to the overall development of the low-level rear-to-front flow along the rainbands, such as that evident in the middle segment of R3 (cf. Figs. 6–9).
Since the low-level rear-to-front flow observed in the present case was mostly westerly, its forcing mechanisms may be appropriately evaluated by the horizontal momentum equation in the east–west direction, which can be written as
where u, υ, and w are the east–west, north–south, and vertical wind components, respectively; ρ is the air density; p is the pressure; f is the Coriolis parameter; and
The PH and AWOS stations, separated by a horizontal distance of 7.3 km, are oriented roughly in the east–west direction (cf. Fig. 1b), so their measurements can provide reasonable estimates of the spatial gradient for terms B and C in (2). The MG sounding valid at 0000 UTC 23 July (not shown) reveals a slightly stable, shallow boundary layer (H ~ 100 m), a common vertical extent characterizing the coastal atmospheric boundary layer (Samelson and Lentz 1994). If all terms on the right-hand side of (2) are estimated properly, their summation may well predict the local acceleration (i.e., term A). On the other hand, the local acceleration term can also be calculated directly based on the temporal variation in winds measured at a fixed location such as surface stations. The degree of consistency between the predicted and calculated local acceleration may provide a relative sense of reliability for the diagnosis of the momentum equation.
Figure 13a shows the time series of the u component measured by the PH station from 1020 to 1220 UTC as R3 passed by. The near-surface westerly flow started to intensify prominently near and after the arrival of the leading edge of R3 and reached a maximum of 9 m s−1 at ~1105 UTC (i.e., close to the accident time of GE222). After this time, the westerly flow continued to weaken and decrease to a minimum intensity (1 m s−1) near the rear edge of the band, a magnitude similar to those observed in the prerainband region (Fig. 13a). The corresponding local acceleration calculated based on Fig. 13a is shown in Fig. 13b, whose temporal trend was generally consistent with that of term A predicted by the summation of all terms (B–E) in (2), as shown in Fig. 13c. The contribution of the Coriolis effect (term D) to flow acceleration was calculated to be quite minor. However, the pressure gradient force (term C) was found to be a dominant forcing contributing to the positive acceleration. This frontward pressure gradient force was provided by the convectively generated mesoscale high on the front side of the band as described in section 5 (cf. Fig. 11). The initial reduction in the u component inside the band at approximately 1108 UTC was due to the negative effect of both friction (term E) and advective acceleration (term B) (cf. Fig. 13c); however, the pressure gradient acceleration became negative after ~1120 UTC, which also contributed partly to the persistent decrease in the u component on the rear side of the band. The appearance of the relatively small, rearward pressure gradient force (i.e., toward the west) in these regions (or times), where convection-induced perturbation pressure was much weaker (Fig. 11), would mostly reflect the presence of the storm-scale low pressure system of Matmo located northwest of Penghu Island (cf. Fig. 1a).
(a) Time series of the east–west wind component (u) measured by the PH station from 1020 to 1220 UTC 23 Jul 2014. (b) Corresponding local acceleration calculated based on u components in (a). (c) Time series of magnitudes of various terms (A–E) in the horizontal momentum [Eq. (2); see details in the text]. The duration encompassing the passage of R3 is highlighted by gray shading and the accident time of GE222 is marked by vertical dashed line.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
(a) Time series of the east–west wind component (u) measured by the PH station from 1020 to 1220 UTC 23 Jul 2014. (b) Corresponding local acceleration calculated based on u components in (a). (c) Time series of magnitudes of various terms (A–E) in the horizontal momentum [Eq. (2); see details in the text]. The duration encompassing the passage of R3 is highlighted by gray shading and the accident time of GE222 is marked by vertical dashed line.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
(a) Time series of the east–west wind component (u) measured by the PH station from 1020 to 1220 UTC 23 Jul 2014. (b) Corresponding local acceleration calculated based on u components in (a). (c) Time series of magnitudes of various terms (A–E) in the horizontal momentum [Eq. (2); see details in the text]. The duration encompassing the passage of R3 is highlighted by gray shading and the accident time of GE222 is marked by vertical dashed line.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
It is noteworthy that the layer of the rear-to-front flow associated with R3 extended vertically up to ~4 km MSL (cf. Fig. 9b). In addition to the forcing mechanism identified above that played a key role in driving the low-level rear-to-front flow, we cannot completely rule out the possibility of other processes that would also operate and favor the development of the elevated rear-to-front flow observed within the rainband. Given that the airflow structure of R3 seen in Fig. 9b suggests a strong, rearward tilting updraft sloping over the low-level cold pool, the buoyancy-induced pressure gradient, as described earlier, would also be likely to occur in the present case to strengthen the upper-level rear-to-front flow. This process may help explain why an additional, local maximum of the rear-to-front flow existed between 3 and 4 km MSL, as evident in Fig. 9b. Future detailed kinematic and thermodynamic information collected within OTCRs and squall lines will be required to clarify the relative roles of the buoyancy-induced high pressure (at low levels) and low pressure (at upper levels) in contributing to the development of the rear-to-front flow inside the squall-line-like rainbands.
7. Operational implications
It is well known that the spatial coverage of TC warnings depends closely on the storm size of TCs, which is practically determined by the radius of a certain threshold of near-surface wind speed or by the radial extent of the outermost closed isobar (e.g., Frank and Gray 1980; Merrill 1984; Knaff et al. 2014). Global observations have indicated the most typical, median size of TCs at approximately 200 km based on the azimuthally averaged radius of 12 m s−1 (Chavas and Emanuel 2010). Nevertheless, in addition to a common characteristic of outward propagation, the OTCRs had been observed to be active over extensive outer regions of TCs from the inner-core boundary (~50–100 km) to the broad vicinity at large radii (~500–600 km) (Yu and Chen 2011; YU18). A considerable portion of OTCRs occurring in the real atmosphere may be located in or propagate into regions outside the operational alert area of TCs.
For the present case, according to the warning report of the typhoon issued by CWB, Matmo maintained a similar size of ~180–200 km (in radius) during the period of our primary interest, which was approximately identical to the average radius of 34 kt (1 kt ≈ 0.51 m s−1) (~199 km) recorded by the Joint Typhoon Warning Center (JTWC). The OTCRs observed ~250–300 km from the center of Matmo, as described in section 3 (cf. Fig. 2b), were apparently situated in outer regions well beyond the TC size. In fact, the TC warning for MG Airport was officially terminated at 0940 UTC, ~5 min prior to the take-off of GE222 (TASC 2016) as the center of Matmo was located ~250 km north-northwest of MG Airport, a radial distance obviously greater than the storm size of the typhoon (cf. Fig. 1). In this context, together with the frequent development of squall-line-like OTCRs (YU18), one may suggest a potential need to expand the area of routine TC alert, or alternatively, to reasonably postpone the termination of TC warnings. Such operational work may help pilots and/or ground controllers maintain a strong lookout for TC-related severe weather and prevent any risky decisions on the flight operations. When considering the threat of TCs on aviation, closely monitoring convective activities in the outer regions of TCs and their potential transformation into squall-line-like storms by real-time radar observations would also represent a critical task for operational forecasters to make a timely warning of rapidly changing weather conditions.
8. Conclusions
Analysis results presented in the previous sections have elaborated that the OTCRs of Typhoon Matmo exhibited squall-line-like convective characteristics, which in turn produced critical weather hazards impacting aviation safety. Specifically, both the heavy rainfall–induced low visibility and strong convectively generated crosswind associated with one of the observed OTCRs (i.e., R3), were primary weather hazards affecting the flight GE222. The structural features of R3 and their relationship with the hazardous weather conditions encountered by GE222 during its final approach to MG Airport are illustrated schematically in Fig. 14. R3 was observed to possess squall-line-like structures with a highly convective nature of precipitation (maximum rain rates > 100 mm h−1), obvious low-level convergence between the prerainband south-southwesterly inflow and band-relative rear-to-front flow and surface cold pool signatures. The low-level inflow fed the rainband and rose rearward immediately ahead of the leading edge of the low-level rear-to-front flow. As GE222 descended to approach the runway of MG Airport and flew immediately behind and roughly parallel to the leading edge of R3, the aircraft encountered the heaviest precipitation of the rainband and the prominent crosswind (i.e., westerly flow) that was exactly a manifestation of band-relative rear-to-front flow generated locally by the rainband. The momentum budget analyses presented in section 6 suggest that the frontward pressure gradient force caused by the near-surface, convectively generated mesohigh (marked by H in Fig. 14) played a major role in driving the low-level rear-to-front flow (crosswinds) inside the band.
Schematic diagram illustrating the structural characteristics of the target OTCR (i.e., R3) and their potential connection with the flight accident of GE222 that occurred near the MG Airport on Penghu Island on 23 Jul 2014. The band-relative rear-to-front flow (RTF), south-southwesterly (SSW) inflow, and rearward-tilting updraft are indicated by blue, red, and white arrows, respectively. The spatial extent of the surface cold pool is indicated qualitatively by blue solid and dashed lines. The approximate location of the mesohigh produced locally inside the rainband is marked by H. The flight track (altitude) of GE222 during its final approach to the airport is depicted by the dark solid arrow (light shading), and the position of the aircraft crash is marked by the red solid star. The runway of MG Airport is indicated by the green area. The leading edge of the rainband is indicated by the dashed line, and the front and rear sides of the rainband are also marked.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
Schematic diagram illustrating the structural characteristics of the target OTCR (i.e., R3) and their potential connection with the flight accident of GE222 that occurred near the MG Airport on Penghu Island on 23 Jul 2014. The band-relative rear-to-front flow (RTF), south-southwesterly (SSW) inflow, and rearward-tilting updraft are indicated by blue, red, and white arrows, respectively. The spatial extent of the surface cold pool is indicated qualitatively by blue solid and dashed lines. The approximate location of the mesohigh produced locally inside the rainband is marked by H. The flight track (altitude) of GE222 during its final approach to the airport is depicted by the dark solid arrow (light shading), and the position of the aircraft crash is marked by the red solid star. The runway of MG Airport is indicated by the green area. The leading edge of the rainband is indicated by the dashed line, and the front and rear sides of the rainband are also marked.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
Schematic diagram illustrating the structural characteristics of the target OTCR (i.e., R3) and their potential connection with the flight accident of GE222 that occurred near the MG Airport on Penghu Island on 23 Jul 2014. The band-relative rear-to-front flow (RTF), south-southwesterly (SSW) inflow, and rearward-tilting updraft are indicated by blue, red, and white arrows, respectively. The spatial extent of the surface cold pool is indicated qualitatively by blue solid and dashed lines. The approximate location of the mesohigh produced locally inside the rainband is marked by H. The flight track (altitude) of GE222 during its final approach to the airport is depicted by the dark solid arrow (light shading), and the position of the aircraft crash is marked by the red solid star. The runway of MG Airport is indicated by the green area. The leading edge of the rainband is indicated by the dashed line, and the front and rear sides of the rainband are also marked.
Citation: Monthly Weather Review 148, 7; 10.1175/MWR-D-20-0054.1
The results from the study not only support YU18’s finding, that is, the frequent development of squall-line-like rainbands in the outer region of TCs, but also demonstrate the hazardous weather conditions associated with OTCRs. As discussed in section 7, a considerable portion of OTCRs occurring in the real atmosphere are expected to be located in or propagate into regions outside the operational alert areas of TCs. When considering the threat of TCs on aviation, closely monitoring convective activities in the outer regions of TCs and their potential transformation into squall-line-like storms by real-time radar observations would thus also be critical for operational forecasters to make a timely warning of rapidly changing weather conditions. These works are important but much more challenging compared to the routine operational procedures to issue TC warnings of the inner-core hazardous region with a quasi-circular geometry that can be practically identified given a predicted or observed TC center.
Acknowledgments
The Doppler radar data and surface observations used in this study were provided by both the Taiwan Central Weather Bureau (CWB) and the Weather Wing of the Air Force of Taiwan. The authors thank three anonymous reviewers for providing constructive comments that improved the manuscript. This study was supported by the Ministry of Science and Technology of Taiwan under Research Grants MOST106-2111-M-002-002-MY3 and MOST106-2111-M-002-013-MY3.
REFERENCES
Anthes, R. A., 1982: Tropical Cyclones: Their Evolution, Structure and Effects. Meteor. Monogr ., No. 41, Amer. Meteor. Soc., 208 pp.
ATR, 2010: ATR72-500 supplementary operations manual. Rev. No. 1, 198 pp., https://www.scribd.com/document/255107069/ATR-72-500-Supplementary-Operations-Manual.
Barnes, G. M., and K. Sieckman, 1984: The environment of fast- and slow-moving tropical mesoscale convective cloud lines. Mon. Wea. Rev., 112, 1782–1794, https://doi.org/10.1175/1520-0493(1984)112<1782:TEOFAS>2.0.CO;2.
Barnes, G. M., E. J. Zipser, D. P. Jorgensen, and F. D. Marks, 1983: Mesoscale and convective structure of a hurricane rainband. J. Atmos. Sci., 40, 2125–2137, https://doi.org/10.1175/1520-0469(1983)040<2125:MACSOA>2.0.CO;2.
Bogner, P. B., G. M. Barnes, and J. L. Franklin, 2000: Conditional instability and shear for six hurricanes over the Atlantic Ocean. Wea. Forecasting, 15, 192–207, https://doi.org/10.1175/1520-0434(2000)015<0192:CIASFS>2.0.CO;2.
Breslin, S., 2016: 10 most weather-delayed U.S. major airports. Accessed 30 August 2018, https://weather.com/travel/commuter-conditions/news/most-weather-delayed-airports-us-2016.
Charba, J., 1974: Application of gravity current model to analysis of squall-line gust front. Mon. Wea. Rev., 102, 140–156, https://doi.org/10.1175/1520-0493(1974)102<0140:AOGCMT>2.0.CO;2.
Chavas, D. R., and K. A. Emanuel, 2010: A QuikSCAT climatology of tropical cyclone size. Geophys. Res. Lett., 37, L18816, https://doi.org/10.1029/2010GL044558.
Chun, H., and Coauthors, 2017: Research collaborations for better predictions of aviation weather hazards. Bull. Amer. Meteor. Soc., 98, ES103–ES107, https://doi.org/10.1175/BAMS-D-17-0010.1.
Eastin, M. D., T. L. Gardner, M. C. Link, and K. C. Smith, 2012: Surface cold pools in the outer rainbands of Tropical Storm Hanna (2008) near landfall. Mon. Wea. Rev., 140, 471–491, https://doi.org/10.1175/MWR-D-11-00099.1.
Fankhauser, J. C., G. M. Barnes, and M. A. LeMone, 1992: Structure of a midlatitude squall line formed in strong unidirectional shear. Mon. Wea. Rev., 120, 237–260, https://doi.org/10.1175/1520-0493(1992)120<0237:SOAMSL>2.0.CO;2.
Frank, W. M., 1977: The structure and energetics of the tropical cyclone. Part I: Storm structure. Mon. Wea. Rev., 105, 1119–1135, https://doi.org/10.1175/1520-0493(1977)105<1119:TSAEOT>2.0.CO;2.
Frank, W. M., and W. M. Gray, 1980: Radius and frequency of 15 m s−1 (30 kt) winds around tropical cyclones. J. Appl. Meteor., 19, 219–223, https://doi.org/10.1175/1520-0450(1980)019<0219:RAFOMS>2.0.CO;2.
Fujita, T. T., and H. R. Byers, 1977: Spearhead echo and downburst in the crash of an airliner. Mon. Wea. Rev., 105, 129–146, https://doi.org/10.1175/1520-0493(1977)105<0129:SEADIT>2.0.CO;2.
Goff, R. C., 1976: Vertical structure of thunderstorm outflows. Mon. Wea. Rev., 104, 1429–1440, https://doi.org/10.1175/1520-0493(1976)104<1429:VSOTO>2.0.CO;2.
Goodman, C. J., and J. D. Small Griswold, 2019: Meteorological impacts on commercial aviation delays and cancellations in the continental United States. J. Appl. Meteor. Climatol., 58, 479–494, https://doi.org/10.1175/JAMC-D-17-0277.1.
Haddad, Z. S., and K.-W. Park, 2010: Vertical profiling of tropical precipitation using passive microwave observations and its implications regarding the crash of Air France 447. J. Geophys. Res., 115, D12129, https://doi.org/10.1029/2009JD013380.
Helmreich, R. L., 1997: Managing human error in aviation. Sci. Amer., 276, 62–67, https://doi.org/10.1038/scientificamerican0597-62.
Houze, R. A., Jr., 2004: Mesoscale convective systems. Rev. Geophys., 42, RG4003, https://doi.org/10.1029/2004RG000150.
Houze, R. A., Jr., 2010: Clouds in tropical cyclones. Mon. Wea. Rev., 138, 293–344, https://doi.org/10.1175/2009MWR2989.1.
Houze, R. A., Jr., 2014: Cloud Dynamics. Academic Press, 496 pp.
Houze, R. A., Jr., S. A. Rutledge, M. I. Biggerstaff, and B. F. Smull, 1989: Interpretation of Doppler weather radar displays of midlatitude mesoscale convective systems. Bull. Amer. Meteor. Soc., 70, 608–619, https://doi.org/10.1175/1520-0477(1989)070<0608:IODWRD>2.0.CO;2.
Humphreys, W. J., 1930: Meteorology and its importance to aviation. Mon. Wea. Rev., 58, 196–197, https://doi.org/10.1175/1520-0493(1930)58<196:MAIITA>2.0.CO;2.
Jorgensen, D. P., M. A. LeMone, and S. B. Trier, 1997: Structure and evolution of the 22 February 1993 TOGA COARE squall line: Aircraft observations of precipitation, circulation, and surface energy fluxes. J. Atmos. Sci., 54, 1961–1985, https://doi.org/10.1175/1520-0469(1997)054<1961:SAEOTF>2.0.CO;2.
Keller, T. L., S. B. Trier, W. D. Hall, R. D. Sharman, M. Xu, and Y. Liu, 2015: Lee waves associated with a commercial jetliner accident at Denver International Airport. J. Appl. Meteor. Climatol., 54, 1373–1392, https://doi.org/10.1175/JAMC-D-14-0270.1.
Kessler, E., 1985: Wind shear and aviation safety. Nature, 315, 179–180, https://doi.org/10.1038/315179a0.
Klimowski, B. A., 1994: Initiation and development of rear inflow within the 28–29 June 1989 North Dakota mesoconvective system. Mon. Wea. Rev., 122, 765–779, https://doi.org/10.1175/1520-0493(1994)122<0765:IADORI>2.0.CO;2.
Knaff, J. A., S. P. Longmore, and D. A. Molenar, 2014: An objective satellite-based tropical cyclone size climatology. J. Climate, 27, 455–476, https://doi.org/10.1175/JCLI-D-13-00096.1.
Kulesa, G., 2003: Weather and aviation: How does weather affect the safety and operations of airports and aviation, and how does FAA work to manage weather-related effects? Conf. on the Potential Impacts of Climate Change on Transportation, Washington, DC, U.S. Department of Transportation, 10 pp., https://trid.trb.org/view/663829.
Lafore, J., and M. W. Moncrieff, 1989: A numerical investigation of the organization and interaction of the convective and stratiform regions of tropical squall lines. J. Atmos. Sci., 46, 521–544, https://doi.org/10.1175/1520-0469(1989)046<0521:ANIOTO>2.0.CO;2.
LeMone, M. A., G. M. Barnes, and E. J. Zipser, 1984: Momentum flux by lines of cumulonimbus over the tropical oceans. J. Atmos. Sci., 41, 1914–1932, https://doi.org/10.1175/1520-0469(1984)041<1914:MFBLOC>2.0.CO;2.
Li, Q., Y. Wang, and Y. Duan, 2017: A numerical study of outer rainband formation in a sheared tropical cyclone. J. Atmos. Sci., 74, 203–227, https://doi.org/10.1175/JAS-D-16-0123.1.
Li, Y., Y. Lin, and Y. Wang, 2019: A numerical study on the formation and maintenance of a long-lived rainband in Typhoon Longwang (2005). J. Geophys. Res. Atmos., 124, 10 401–10 426, https://doi.org/10.1029/2019JD030600.
Marks, F. D., 2003: State of the science: Radar view of tropical cyclones. Radar and Atmospheric Science: A Collection of Essays in Honor of David Atlas, Meteor. Monogr., No. 52, Amer. Meteor. Soc., 33–74.
Merrill, R. T., 1984: A comparison of large and small tropical cyclones. Mon. Wea. Rev., 112, 1408–1418, https://doi.org/10.1175/1520-0493(1984)112<1408:ACOLAS>2.0.CO;2.
Molinari, J., D. M. Romps, D. Vollaro, and L. Nguyen, 2012: CAPE in tropical cyclones. J. Atmos. Sci., 69, 2452–2463, https://doi.org/10.1175/JAS-D-11-0254.1.
Moon, Y., and D. S. Nolan, 2015: Spiral rainbands in a numerical simulation of Hurricane Bill (2009). Part I: Structures and comparisons to observations. J. Atmos. Sci., 72, 164–190, https://doi.org/10.1175/JAS-D-14-0058.1.
Pennell, W. T., and M. A. LeMone, 1974: An experimental study of turbulence structure in the fair-weather trade wind boundary layer. J. Atmos. Sci., 31, 1308–1323, https://doi.org/10.1175/1520-0469(1974)031<1308:AESOTS>2.0.CO;2.
Roux, F., 1988: The West African squall line observed on 23 June 1981 during COPT 81: Kinematics and thermodynamics of the convective region. J. Atmos. Sci., 45, 406–426, https://doi.org/10.1175/1520-0469(1988)045<0406:TWASLO>2.0.CO;2.
Roux, F., J. Testud, M. Payen, and B. Pinty, 1984: West African squall-line thermodynamic structure retrieved from dual-Doppler radar observations. J. Atmos. Sci., 41, 3104–3121, https://doi.org/10.1175/1520-0469(1984)041<3104:WASLTS>2.0.CO;2.
Rozoff, C. M., W. H. Schubert, B. D. McNoldy, and J. P. Kossin, 2006: Rapid filamentation zones in intense tropical cyclones. J. Atmos. Sci., 63, 325–340, https://doi.org/10.1175/JAS3595.1.
Saha, S., and Coauthors, 2010: The NCEP Climate Forecast System Reanalysis. Bull. Amer. Meteor. Soc., 91, 1015–1058, https://doi.org/10.1175/2010BAMS3001.1.
Samelson, R. M., and S. J. Lentz, 1994: The horizontal momentum balance in the marine atmospheric boundary layer during CODE-2. J. Atmos. Sci., 51, 3745–3757, https://doi.org/10.1175/1520-0469(1994)051<3745:THMBIT>2.0.CO;2.
Schaefer, J. T., and Coauthors, 1992: Summary of National Weather Service aviation workshop Kansas City, Missouri, 10–12 December 1991. Bull. Amer. Meteor. Soc., 73, 1835–1842, https://doi.org/10.1175/1520-0477-73.11.1835.
Senn, H. V., and H. W. Hiser, 1959: On the the origin of hurricane spiral rain bands. J. Meteor., 16, 419–426, https://doi.org/10.1175/1520-0469(1959)016<0419:OTOOHS>2.0.CO;2.
Simpson, J. E., and R. E. Britter, 1980: A laboratory model of an atmospheric mesofront. Quart. J. Roy. Meteor. Soc., 106, 485–500, https://doi.org/10.1002/qj.49710644907.
Skwira, G. D., J. L. Schroeder, and R. E. Peterson, 2005: Surface observations of landfalling hurricane rainbands. Mon. Wea. Rev., 133, 454–465, https://doi.org/10.1175/MWR-2866.1.
Smith, M., 1986: Visual observations of Kansas downbursts and their relation to aviation weather observations. Mon. Wea. Rev., 114, 1612–1616, https://doi.org/10.1175/1520-0493(1986)114<1612:VOOKDA>2.0.CO;2.
Smull, B. F., and R. A. Houze, 1985: A midlatitude squall line with a trailing region of stratiform rain: Radar and satellite observations. Mon. Wea. Rev., 113, 117–133, https://doi.org/10.1175/1520-0493(1985)113<0117:AMSLWA>2.0.CO;2.
Stull, R. B., 1988: An Introduction to Boundary Layer Meteorology. Kluwer Academic, 670 pp.
TASC, 2016: 23 July, 2014 TransAsia Airways flight GE222 ATR72-212A, B-22810 impacted terrain and collided with a residential area northeast of the threshold of runway 20 at Magong Airport. ASC-AOR-16-01-002, 382 pp., https://www.ttsb.gov.tw/media/3194/asc-aor-16-01-002-en.pdf.
Tang, X., W.-C. Lee, and B. Michael, 2014: A squall-line-like principal rainband in Typhoon Hagupit (2008) observed by airborne Doppler radar. J. Atmos. Sci., 71, 2733–2746, https://doi.org/10.1175/JAS-D-13-0307.1.
Tompkins, A. M., 2001: Organization of tropical convection in low vertical wind shears: The role of cold pools. J. Atmos. Sci., 58, 1650–1672, https://doi.org/10.1175/1520-0469(2001)058<1650:OOTCIL>2.0.CO;2.
Wakimoto, R. M., 1982: The life cycle of thunderstorm gust fronts as viewed with Doppler radar and rawinsonde data. Mon. Wea. Rev., 110, 1060–1082, https://doi.org/10.1175/1520-0493(1982)110<1060:TLCOTG>2.0.CO;2.
Wakimoto, R. M., P. Stauffer, and W.-C. Lee, 2015: The vertical vorticity structure within a squall line observed during BAMEX: Banded vorticity features and the evolution of a bowing segment. Mon. Wea. Rev., 143, 341–362, https://doi.org/10.1175/MWR-D-14-00246.1.
Wang, T. C., Y. Lin, H. Shen, and R. W. Pasken, 1990: Characteristics of a subtropical squall line determined from TAMEX dual-Doppler data. Part I: Kinematic structure. J. Atmos. Sci., 47, 2357–2381, https://doi.org/10.1175/1520-0469(1990)047<2357:COASSL>2.0.CO;2.
Wang, Y., 2009: How do outer spiral rainbands affect tropical cyclone structure and intensity? J. Atmos. Sci., 66, 1250–1273, https://doi.org/10.1175/2008JAS2737.1.
Weisman, M. L., 1993: The genesis of severe, long-lived bow echoes. J. Atmos. Sci., 50, 645–670, https://doi.org/10.1175/1520-0469(1993)050<0645:TGOSLL>2.0.CO;2.
Weisman, M. L., and C. A. Davis, 1998: Mechanisms for the generation of mesoscale vortices within quasi-linear convective systems. J. Atmos. Sci., 55, 2603–2622, https://doi.org/10.1175/1520-0469(1998)055<2603:MFTGOM>2.0.CO;2.
Willoughby, H. E., 1988: The dynamics of the tropical hurricane core. Aust. Meteor. Mag., 36, 183–191.
Willoughby, H. E., F. D. Marks, and R. J. Feinberg, 1984: Stationary and moving convective bands in hurricanes. J. Atmos. Sci., 41, 3189–3211, https://doi.org/10.1175/1520-0469(1984)041<3189:SAMCBI>2.0.CO;2.
Wilson, J. W., R. D. Roberts, C. Kessinger, and J. McCarthy, 1984: Microburst wind structure and evaluation of Doppler radar for airport wind shear detection. J. Climate Appl. Meteor., 23, 898–915, https://doi.org/10.1175/1520-0450(1984)023<0898:MWSAEO>2.0.CO;2.
WMO, 1989: 1989 World Meteorological Day celebrates meteorology in the service of aviation. Bull. Amer. Meteor. Soc., 70, 414–415, https://doi.org/10.1175/1520-0477-70.4.414.
Wurtele, M. G., 1970: Meteorological conditions surrounding the Paradise Airline crash of 1 March 1964. J. Appl. Meteor., 9, 787–795, https://doi.org/10.1175/1520-0450(1970)009<0787:MCSTPA>2.0.CO;2.
Yu, C.-K., and B. J.-D. Jou, 2005: Radar observations of the diurnally forced offshore convective lines along the southeastern coast of Taiwan. Mon. Wea. Rev., 133, 1613–1636, https://doi.org/10.1175/MWR2937.1.
Yu, C.-K., and Y.-H. Hsieh, 2009: Formation of the convective lines off the mountainous coast of southeastern Taiwan: A case study of 3 January 2004. Mon. Wea. Rev., 137, 3072–3091, https://doi.org/10.1175/2009MWR2867.1.
Yu, C.-K., and C.-L. Tsai, 2010: Surface pressure features of landfalling typhoon rainbands and their possible causes. J. Atmos. Sci., 67, 2893–2911, https://doi.org/10.1175/2010JAS3312.1.
Yu, C.-K., and Y. Chen, 2011: Surface fluctuations associated with tropical cyclone rainbands observed near Taiwan during 2000–08. J. Atmos. Sci., 68, 1568–1585, https://doi.org/10.1175/2011JAS3725.1.
Yu, C.-K., and L.-W. Cheng, 2013: Distribution and mechanisms of orographic precipitation associated with Typhoon Morakot (2009). J. Atmos. Sci., 70, 2894–2915, https://doi.org/10.1175/JAS-D-12-0340.1.
Yu, C.-K., and C.-L. Tsai, 2013: Structural and surface features of arc-shaped radar echoes along an outer tropical cyclone rainband. J. Atmos. Sci., 70, 56–72, https://doi.org/10.1175/JAS-D-12-090.1.
Yu, C.-K., and L.-W. Cheng, 2014: Dual-Doppler-derived profiles of the southwesterly flow associated with southwest and ordinary typhoons off the southwestern coast of Taiwan. J. Atmos. Sci., 71, 3202–3222, https://doi.org/10.1175/JAS-D-13-0379.1.
Yu, C.-K., C.-Y. Lin, L.-W. Cheng, J.-S. Luo, C.-C. Wu, and Y. Chen, 2018: The degree of prevalence of similarity between outer tropical cyclone rainbands and squall lines. Sci. Rep., 8, 8247, https://doi.org/10.1038/s41598-018-26553-8.
Yu, C.-K., C.-Y. Lin, and J.-S. Luo, 2019: Tracking a long-lasting outer tropical cyclone rainband: Origin and convective transformation. J. Atmos. Sci., 76, 3267–3283, https://doi.org/10.1175/JAS-D-19-0126.1.
The minimum descent altitude (MDA) is a specific altitude in a nonprecision approach, in which the descent of the aircraft must not be made without the required visual reference.