Statistics and Possible Sources of Aviation Turbulence over South...

Statistics and Possible Sources of Aviation Turbulence over South Korea

JUNG-HOON KIM AND HYE-YEONG CHUN

Department of Atmospheric Sciences, Yonsei University, Seoul, South Korea

(Manuscript received 11 February 2010, in final form 7 September 2010)

ABSTRACT

The characteristics of aviation turbulence over South Korea during the recent five years (2003–08, ex-

cluding 2005) are investigated using pilot reports (PIREPs) accumulated by the Korea Aviation Meteoro-

logical Agency (KAMA). Among the total of 8449 PIREPs, 4607 (54.53%), 1646 (19.48%), 248 (2.94%),

7 (0.08%), and 1941 (22.97%) correspond to the turbulence categories of null, light, moderate, severe, and

missing, respectively. In terms of temporal variations, the annual total number of turbulence events increased

from 2003 to 2008, and the seasonal frequency is the highest in the spring. With regard to spatial distributions,

reported turbulence encounters are dominant along the prevailing flight routes, but are locally higher over the

west coast, Jeju Island, and the Sobaek and Taebaek mountains. The turbulence events in these regions vary

by season. To examine the regional differences and possible sources of the observed turbulence, lightning

flash data, Regional Data Assimilation and Prediction System (RDAPS) analysis data with a 30-km hori-

zontal grid spacing provided by the Korean Meteorological Administration (KMA), and a digital elevation

model (DEM) dataset with a 30-s resolution, are additionally used. Convectively induced turbulence (CIT)

and clear-air turbulence (CAT) events comprised 11% and 89% of the total 255 moderate or greater (MOG)-

level turbulence events, respectively. CAT events are classified as tropopause/jet stream–induced CAT

(TJCAT) and mountain-wave-induced CAT (MWCAT) events. The MOG-level TJCAT and MWCAT

events are responsible for 41.2% and 19.6% of the total MOG-level turbulence events, respectively. The CIT

events in summer and the TRCAT and MWCAT events in spring occur most frequently over the previously

mentioned regions of South Korea, associated with specific generation mechanisms.

1. Introduction

Turbulence at aircraft scale (10–1000 m) or that di-

rectly affects aircraft is commonly referred to as aviation

turbulence (Lester 1994). Aviation turbulence in the free

atmosphere is a serious concern in the general aviation

industry because it frequently causes occupant injuries,

flight delays, fuel losses, and structural damage. It is more

dangerous when it occurs unexpectedly at cruising levels,

where most of the passengers and crew are unbuckled.

According to the 2009 annual report of the National

Transportation Safety Board (NTSB 2009), from 1996 to

2005 turbulence was the leading cause of weather-related

aircraft accidents in the United States. Over South Korea,

from 1957 to the present turbulence has accounted for

about 24% of the aircraft accidents caused by weather,

making it the largest contributor to weather-related

accidents [statistics from the Aviation and Railway Ac-

cident Investigation Board (ARAIB), information online

at http://www.araib.go.kr].

Possible sources of aviation turbulence include con-

vective systems (Pantley and Lester 1990; Lane et al.

2003), jet streams along with upper-level fronts (Dutton

and Panofsky 1970; Ellrod and Knapp 1992), complex

terrain (Clark et al. 2000; Doyle et al. 2005), and inertial

instabilities (Koch et al. 2005; Knox et al. 2008). Even

though the relationship between source and turbulence

is not understood completely, temporal and spatial dis-

tributions of turbulence events can provide insights into

possible turbulence sources (Lane et al. 2009). In addi-

tion, the relative frequency of the turbulence events in

different areas can provide information useful for long-

range aviation route planning (Wolff and Sharman 2008).

Therefore, many studies involving statistical analyses of

turbulence events have been carried out using diagnostic

indices derived from numerical weather prediction (NWP)

models and turbulence observations, including pilot re-

ports (PIREPs) and in situ measurements. The former has

the advantage of uniform coverage in all areas, but

Corresponding author address: Prof. Hye-Yeong Chun, Dept.

of Atmospheric Sciences, Yonsei University, 262 Seongsanno,

Seodaemun-gu, Seoul 120-749, Korea.

E-mail: [email protected]

FEBRUARY 2011 K I M A N D C H U N 311

DOI: 10.1175/2010JAMC2492.1

� 2011 American Meteorological Society

includes uncertainties such that predicted turbulence po-

tential cannot represent the observed turbulence perfectly.

For example, overall performances of the graphic tur-

bulence guidance (GTG) in the United States (Sharman

et al. 2006) and in Korea (Kim et al. 2009) are only about

75% and 70%, respectively, and are highly dependent

upon the model configurations (e.g., resolution, param-

eterization, etc). On the other hand, the latter has the

advantage in that it reports the locations and timings of

aviation turbulence, although the observations are in-

herently subjective (Wolff and Sharman 2008).

The characteristics of turbulence events based on PIREPs

have been reported for various regions in the world.

Wolff and Sharman (2008) examined the climatology of

upper-level turbulence over the contiguous United States

(CONUS) using many years of PIREPs. They showed that

possible sources of upper-level (18 000–60 000 ft) turbu-

lence include mountain waves in the western region and

winter-season jet streams in the northeastern regions.

The relatively higher numbers of incidents over Florida,

eastern Texas, and along the Gulf coast are presumably

associated with convective clouds. In Greenland, turbu-

lence events reported in PIREPs were relatively higher

in the winter season from 2000 to 2006, caused mainly by

mountain-wave breaking when surface cyclones produce

prevailing easterlies in the lower troposphere (Lane et al.

2009). Lee and Choi (2003) investigated the spatial and

temporal distributions of turbulence events over South

Korea, reported in PIREPs from 1996 to 2000. They

showed that these events were more frequent in the spring

season, with regional differences in the seasonal and spa-

tial distributions of the turbulence events.

Geographically, Korea and eastern Asia have signifi-

cant potential for turbulence events because the stron-

gest jet stream in the world exists there (Jaeger and

Sprenger 2007; Koch et al. 2006). Moreover, since

complex terrain covers more than 70% of the Korean

Peninsula, the prevailing westerlies can generate a wide

spectrum of mountain waves. Given the significant in-

crease in air transportation over Korea and eastern Asia

during the past decade, the characteristics of aviation tur-

bulence in this region should be investigated. This would

provide invaluable information to pilots, dispatchers, and

forecasters to maintain air flight safety and to reduce the

amount of unexpected damage from aviation turbulence.

The remainder of this paper is organized as follows. In

section 2, the PIREP data used in this study are described.

In section 3, temporal and spatial distributions of the

PIREP data are examined. In section 4, possible sources

of the observed turbulence events are investigated using

lightning flash data, Regional Data Assimilation and

Prediction System (RDAPS) analysis data with a 30-km

horizontal grid resolution, and a digital elevation model

(DEM) dataset with a 30-s resolution. A summary and

discussion of the results are provided in section 5.

2. Pilot reports (PIREPs)

Currently, PIREPs are the only routinely available

observations of turbulence from aircraft in South Korea.

The PIREPs used in this study were collected using the

following steps. First, pilots in commercial aircraft flying

over the Incheon Flight Information Region (FIR; Fig. 1)

routinely and occasionally report the existence of turbu-

lence to the Korea Air Traffic Control Center (KATCC)

through regular voice radio contact. These PIREPs in-

clude several pieces of information such as location, flight

level, type of aircraft, wind speed and direction, and the

turbulence (and/or icing) level and intensity [null (NIL),

light (LGT), moderate (MOD), severe (SEV), or extreme

(EXT)]. Some pilots additionally report weather condi-

tions such as visibility and cloud cover as well as the extent

of the turbulence level. Second, the air traffic controller

documents the information in the PIREPs. Finally, the

raw text of these PIREPs is transmitted to the Korea

Aviation Meteorological Agency (KAMA) and accumu-

lated in their database. In this study, a total of 8449 PIREPs

during the 5 yr from 2003 to 2008 (excluding 2005 due to

the lack of raw data) are used after applying quality

FIG. 1. The FIR (dashed) and standard flight routes (solid) over

South Korea. The locations and names of major cities and moun-

tain ranges are depicted as points and shadings, respectively.

312 J O U R N A L O F A P P L I E D M E T E O R O L O G Y A N D C L I M A T O L O G Y VOLUME 50

controls such as removing duplicates and discarding

PIREPs that did not include both location and flight

level.

As pointed out by Schwartz (1996), potential uncer-

tainties in the PIREP data, such as turbulence intensity

and location, should be examined before investigating

turbulence statistics. The turbulence intensity reported

in a PIREP tends to be determined by the pilot’s expe-

rience, and the locations recorded sometimes differ by

more than 10s or 100s of kilometers from the actual loca-

tions of the turbulence observed by in situ measurements

(Cornman et al. 2004). To evaluate these uncertainties,

we compared reported intensities with the intensities

observed in PIREPs located within 50 km horizontally,

500 ft vertically, and 10 min temporally, following the

method proposed by Sharman et al. (2006). It was found

that the percentages of total agreement for the NIL- and

LGT-level events during the present research period were

about 98% (55 cases) and 61% (23 cases), respectively. No

MOD events and only one SEV event could be selected

due to the small sample size. However, when the ranges

and time windows are extended to 1000 ft vertically and

20 min, respectively, the percentage of total agreement

for five MOD-level events was 80%. In general, only a few

examples of coexisting turbulence were detected, because

most pilots tend to avoid precautioned locations where

significant turbulence events have been reported within

a certain range and time window. This may cause an un-

derestimation of the actual number of turbulence events

in all statistical research of turbulence that use PIREPs

(Wolff and Sharman 2008). Nevertheless, total agreement

for the NIL-, LGT-, and MOD-level events in the present

study is likely sufficient to ignore the uncertainties in

turbulence intensity and location, especially when com-

pared with those in CONUS (Wolff and Sharman 2008) of

68%, 40%, and 88% for NIL-, LGT-, and MOD-level

events, respectively. It is worth noting that the number of

cases considered in this study is much smaller than those in

the previous work by Wolff and Sharman (2008). Hence,

direct comparison of the agreement between South Korea

and CONUS PIREP data may not be reliable. Also, tur-

bulence intensities reported in PIREPs within the same

range and time windows can differ depending on the air-

craft type and size. The aircrafts providing PIREP data in

the present study are heavy and large types of commercial

airliners, such as the Boeing and Airbus series.

3. Turbulence statistics

Table 1 shows the numbers and percentages of the

PIREP data over South Korea from 2003 to 2008 cor-

responding to the NIL, LGT, MOD, SEV, and missing

categories. Missing PIREPs correspond to PIREPs that

lack turbulence information. Of 8449 PIREPs, 1646, 248,

and 7 were LGT-, MOD-, and SEV-level events, respec-

tively. In the previous statistical analysis of PIREPs over

South Korea, 4709 PIREPs obtained during the 5 yr from

1996 to 2000 yielded 477, 100, and 3 LGT-, MOD-, and

SEV-level events, respectively (Lee and Choi 2003).

Compared with the previous research by Lee and Choi

(2003), the absolute numbers of all types of turbulence

events analyzed in this study are higher and the relative

portions of all severity levels are also higher (19.48% versus

10.13%, 2.94% versus 2.12%, and 0.08% versus 0.07% for

LGT-, MOD-, and SEV-level events, respectively).

a. Yearly and seasonal distributions

To investigate annual variations in the turbulence fre-

quencies from 2003 to 2008, the annual total of PIREPs

(including missing PIREPs), and light or greater (LOG)-

and moderate or greater (MOG)-level events, are pre-

sented in Fig. 2. Relative percentages normalized by the

total number of PIREPs for each year also are presented,

as this can reduce air traffic bias in the PIREP data (Wolff

and Sharman 2008; Lane et al. 2009). Although annual

counts of the total and LOG- and MOG-level events

differ from year to year, the numbers of LOG- and MOG-

level events show a definitely increasing trend during the

5-yr period. In particular, the 68 MOG-level events re-

ported in 2008 are more than twice the number reported

in 2003 (32 MOG-level events). This increasing pattern is

also shown in the relative percentages of both the LOG-

and MOG-level events, except for the LOG-level events

in 2007. The relative percentages of the LOG- (MOG)-

level events increased from 28.92% (2%) in 2003 to

40.68% (6.51%) in 2008.

To investigate a possible reason for this increasing

pattern, the jet stream frequencies among the LOG- and

MOG-level turbulence events are examined using the

RDAPS analysis data with a 30-km horizontal grid spac-

ing. If horizontal wind speeds greater than 40 m s21 are

found near a LOG- or MOG-level turbulence event, then

the event is regarded as being related to the jet stream.

TABLE 1. Numbers and percentages corresponding to the NIL,

LGT, MOD, SEV, and missing categories of aviation turbulence

obtained from PIREPs over South Korea from 2003 to 2008, ex-

cluding 2005.

Turbulence intensity No. Percentage

NIL 4607 54.53

LGT 1646 19.48

MOD 248 2.94

SEV 7 0.08

Missing 1941 22.97

Total 8449 100.0


The frequencies and relative percentages normalized by

total PIREPs within a given year generally increased from

2003 [LOG (104/1603; 6.5%), MOG (9/1603; 0.6%)] to

2008 [LOG (159/1597; 10%), MOG (27/1597; 1.7%)],

implying that atmospheric conditions over South Korea

likely were more conducive for generating turbulence in

2008 than they were in 2003. Although unavoidable un-

certainties in the PIREP data still remain, this suggests

that forecasting of aviation turbulence over South Korea

becomes more important.

Figure 3 shows seasonal variations in the LOG- and

MOG-level events. In Fig. 3, the first and second maxi-

mum frequencies of the LOG-level events appear in the

spring and winter seasons, respectively, consistent with

the results obtained by Lee and Choi (2003). For relative

percentages normalized by the total number of PIREPs

within each season (solid line), the LOG-level events

also occur most frequently in the spring and winter sea-

sons. Although more detailed synoptic analyses are re-

quired to explain this seasonal variation, enhanced vertical

wind shear below or above the jet stream causes aviation

turbulence more frequently during the winter and spring

seasons (e.g., Dutton and Panofsky 1970; Ellrod and Knapp

1992). This seasonal effect is likely due to the prevailing

westerlies in these seasons, which are climatologically

greater than those in the summer and fall.

For the MOG-level events (Fig. 3), the first maximum

frequency occurs in the spring months, while the second

maximum frequency occurs in the summer months, in

contrast to the LOG-level events shown in Fig. 3. Moreover,

the relative percentages of the MOG-level events in the

summer and fall months are higher than those in the

winter months. Given that convective clouds associated

with the monsoon system and typhoons frequently pass

through the Korean Peninsula in the summer and fall,

the higher frequencies of the MOG-level events in these

seasons likely are related to the presence of convective

systems (e.g., Pantley and Lester 1990; Lane et al. 2003).

The relationship between the observed turbulence and

possible sources will be examined in section 4.

b. Vertical and horizontal distributions

To understand the vertical distributions of the observed

turbulence over South Korea, all LOG- and MOG-level

events for the entire study period are sorted into 5000-ft

bins, from flight level (FL) 0 to FL500 (50 000 ft). Average

annual counts for each bin are shown in Fig. 4. To quantify

the height dependency of the turbulence intensity, the

relative percentages of the LOG- and MOG-level events

normalized by the total numbers of PIREPs within a given

altitude are also depicted in Fig. 4. When a turbulence

altitude is reported in a PIREP as a range (e.g., 30 000–

34 000 ft), a mean altitude (e.g., 32 000 ft) is used to

construct the vertical distribution shown in Fig. 4. Note

that the flight levels shown in Fig. 4 were converted from

pressure levels measured by a barometer in the commer-

cial aircraft to the heights using standard atmospheric as-

sumptions (Sharman et al. 2006). In Fig. 4, the maximum

frequencies of both the LOG- and MOG-level events

appear between the flight levels of 20 000 and 25 000 ft

(z 5 about 6.1–7.6 km), the predominant flight levels

FIG. 2. Annual counts of total PIREPs (white bars) and LOG-

(gray bars) and MOG- (black bars) level turbulence events over

South Korea from 2003 to 2008. The relative percentages of the

LOG- and MOG-level turbulence events normalized by the total

number of PIREPs within a given year are depicted as dashed and

solid lines, respectively. For better presentation, the numbers of

the LOG- and MOG-level turbulence events are multiplied by 2

and 4, respectively.

FIG. 3. The 5-yr mean values of seasonal frequencies for the

LOG- (black) and MOG- (gray) level turbulence events over South

Korea from 2003 to 2008 with their error bars represented by the

standard deviation. The relative percentages of the LOG- and

MOG-level events normalized by the total number of PIREPs

within a given season are depicted as solid and dashed lines, re-

spectively.


for domestic airliners over South Korea. The relative

percentages of the LOG- and MOG-level events above

20 000 ft are consistently about 4% and 30%, respectively,

implying that the height dependency of the turbulence

intensity over South Korea during this period is not

significant.

To identify the horizontal distributions of aviation

turbulence over South Korea, frequencies for the total

and LOG- and MOG-level events are accumulated in

horizontal grid boxes of 30 km 3 30 km, the same hori-

zontal resolution as RDAPS. Results are shown in Fig. 5.

Note that the domains in Fig. 5 and all subsequent figures

are the subdomains of the RDAPS focused on the Korean

Peninsula. As expected, turbulence events are dominant

along the prevailing flight routes over South Korea, al-

though local maxima occur along the routes. According to

the aviation statistics from the Korea Airport Corporation

(KAC; information online at http://kac.airport.co.kr), there

are three dominant domestic flight routes in the Korea FIR

(see Fig. 1): the routes from Seoul to Jeju, Seoul to Busan,

and Jeju to Busan. In addition to these major air routes,

a flight route from Seoul to Gangneung is additionally

important, because international airliners from Korea to

Japan or to the United States usually use this flight route.

Along these dominant flight routes, the relatively higher

frequencies of both the LOG- and MOG-level events oc-

cur near the west coast, Jeju Island, and the Sobaek and

Taebaek mountains.

To better understand the regional patterns of aviation

turbulence, horizontal distributions of the LOG-level

events for each season are presented in Fig. 6. As the ac-

cumulated turbulence events in each grid box are nor-

malized by the total number of PIREPs within a box for a

given season, air traffic changes over different seasons and

areas are also considered. In addition, any grid box with

less than 10 PIREPs was ignored to avoid artificially large

values of the relative percentages due to the lack of suf-

ficient samples (Wolff and Sharman 2008).

In Fig. 6, the overall percentages in the winter and spring

seasons are higher than those in the summer and fall

seasons, which is consistent with the results of the seasonal

variation shown in Fig. 3a. The aforementioned four re-

gions with relatively higher frequencies in Fig. 5 have dif-

ferent turbulence frequencies in each season. In the winter

(Fig. 6a), turbulence events occur more frequently along

the west coast region of South Korea, especially near

Gunsan, Incheon, and northeast of Jeju Island. In the

spring season (Fig. 6b), the maximum turbulence fre-

quency is located near Gangneung on the lee side of the

Taebaek mountains, with a relatively higher frequency

near Gwangju in the southwest coastal region of South

Korea. In the summer (Fig. 6c) and fall seasons (Fig. 6d),

areas of relatively higher turbulence frequencies are lo-

cated near the southwest coast, and the Sobaek and

Taebaek mountains, although their frequencies are not as

large as those in winter and spring. Possible sources of the

observed turbulence during these seasons include con-

vective systems, the jet stream, and mountain waves.

Details of each will be examined in section 4.

The MOG-level turbulence results in Fig. 7 are similar

to those in Fig. 6, except for the relatively higher fre-

quencies near Jeju Island in winter and spring, and the low

frequencies near the Taebaek mountains in summer and

near the Sobaek mountains in fall. As can be expected

from the seasonal variations in the MOG-level turbulence

shown in Fig. 3, the overall percentages of the MOG-level

turbulence frequencies in spring and summer are higher

than those in winter and fall.

4. Possible sources of turbulence

In this section, possible sources of the observed tur-

bulence events are investigated using lightning flash data,

the RDAPS analysis data with a 30-km horizontal grid

spacing provided by the KMA, and the DEM dataset

with a 30-s resolution.

a. Convective system

Convective systems are important sources of aviation

turbulence. Turbulence events related to a convective sys-

tem, referred to as convectively induced turbulence (CIT),

are classified into two categories, in-cloud and out-of-cloud

CIT events, depending on their location (Lane et al. 2003).

Generation mechanisms for these turbulence events in-

clude strong variations in upward and downward motions

within a small horizontal distance in a cloud boundary (in-

cloud convective instability), the enhancement of vertical

FIG. 4. Similar to Fig. 3, but for vertical frequencies.


wind shear due to flow deformation near a cloud boundary

(Grabowski and Clark 1991), and breaking of convectively

induced gravity waves out of a cloud boundary (Lane et al.

2003; Lane and Sharman 2008). According to a study of

severe turbulence events that caused accidents in the

United States, conducted by Kaplan et al. (2005a), 86% of

the turbulence events were located horizontally within

100 km of well-organized convective clouds. In the pres-

ent study, lightning flash data accumulated by the KMA

were used to isolate CIT events from aviation turbulence

reported in the current PIREPs. Since 2001, the KMA has

established 17 observation sites to detect cloud-to-ground

(CG) and cloud-to-cloud (CC) lightning over South Ko-

rea. In the present study, turbulence reported within

100 km spatially and 640 min temporally of any CG and

CC lightning activities was regarded as CIT. This classifi-

cation method is the same as that used by Sharman et al.

(2006) and Kim et al. (2009), who evaluated the clear-air

turbulence (CAT) forecasting system over the CONUS

and South Korea, respectively, and Wolff and Sharman

(2008), who investigated the correlation between turbu-

lence and lightning over the CONUS. After extracting CIT

events, the remaining turbulence events were regarded as

CAT events.

The numbers and percentages of CIT and CAT events

corresponding to individual intensities over South Korea

from 2003 to 2008 are given in Table 2, which shows that

152 (1494) and 28 (227) are LGT- and MOG-level CIT

(CAT) events, respectively. Table 1 shows 1646 (255)

LGT- (MOG)-level turbulence events, so the relative

amounts of the LGT- (MOG)-level CIT and CAT events

among the LGT- (MOG)-level turbulence events are

about 9.2% (11%) and 90.8% (89%), respectively. The

relative portions of the LGT- (MOG)-level CIT events

among the total number of CIT events is 20.74%

(3.82%), which is slightly higher than that of the LGT-

(19.36%) [MOG (2.94%)]-level CAT events among the

total CAT events.

Figure 8 shows the monthly variations in the LOG-

and MOG-level CIT events. As expected, CIT events

occur more frequently in the summer than in any other

season. The maximum occurrence is in August, when well-

developed convective systems are dominant over Korea,

driven either by thermally forced or large-scale conver-

gent flow. To account for monthly air traffic changes, the

relative percentages of the LOG- and MOG-level CIT

events normalized by the total number of PIREPs within a

given month are shown as dashed and solid lines, re-

spectively, in Fig. 8. These relative percentages of both the

LOG- and MOG-level CIT events are also highest in

August, at about 9.44% and 2.04%, respectively.

Other noticeable results are seen in Fig. 8. First, a

relatively high number of turbulence events also appear

in September, when typhoon activity directly or indirectly

affects the Korean Peninsula. Second, relatively large

frequencies of CIT events occur in the spring season when

well-organized convective systems pass through the Ko-

rean Peninsula along with occasionally developed low

pressure systems. Although actual frequencies of CIT

events may be underestimated because pilots usually try

to avoid previously identified convective systems, the

overall features of the seasonal variations are similar to

FIG. 5. Horizontal distributions (numbers) of the (a) total and (b) LOG- and (c) MOG-level turbulence events within a 30 km 3 30 km

horizontal grid box using 5 yr of PIREPs over South Korea from 2003 to 2008. Terrain heights (solid lines) are contoured in all plots with

a 200-m interval.


those obtained by Wolff and Sharman (2008, Fig. 19). In

the spatial distribution of CIT events in South Korea (not

shown), the highest frequencies of CIT events occur in the

areas of the southwest coast, Jeju Island, and the So-Baek

mountain region.

b. Tropopause/jet stream

A large portion of CAT events can be explained by the

Kelvin–Helmholtz instability, which is generated when the

dimensionless local Richardson number (Ri 5 N2/VWS2,

FIG. 6. Relative percentages of the LOG-level turbulence events normalized by the total number of PIREPs within

the (a) winter (DJF), (b) spring (MAM), (c) summer (JJA), and (d) fall (SON) seasons derived using 5 yr of PIREPs

over South Korea from 2003 to 2008. Terrain heights (solid lines) are contoured in all plots with a 200-m interval.


where N and VWS are the Brunt–Vaisala frequency and

vertical wind shear, respectively) is below 0.25 (Dutton

and Panofsky 1970). This instability frequently occurs

above or below a strong jet stream accompanied by an

upper-level front (Ellrod and Knapp 1992). Due to this

upper-level frontogenesis, the dynamic tropopause is

folded down into the midtroposphere. According to

previous studies, CAT events frequently occur near this

tropopause region (Koch et al. 2005) and are important

not only for aviation safety but also for the stratosphere–

troposphere exchange (STE) of chemical constituents

(Shapiro 1980). To isolate CAT events occurring near

FIG. 7. As in Fig. 6, but for MOG-level turbulence.


the tropopause region that is associated with the jet

stream, RDAPS analysis data with 30-km horizontal

grid spacing produced at 0000 and 1200 UTC were used

to calculate the dynamic tropopause height [2 potential

vorticity units (PVU)]. CAT events located near the

2-PVU level, within 61 km vertically and 66 h tem-

porally, were regarded as being related to the tropo-

pause/jet stream region. Hereafter, these CAT events are

referred to as TJCAT (tropopause/jet stream–induced

CAT) events. This method is the same as that used by

Wolff and Sharman (2008).

In the tropopause/jet stream along with the upper-

level front structures, CAT encounters are also associ-

ated with inertial instability and geostrophic adjustment

(Knox 1997). In rotating flow, any existence of an abso-

lute rotation opposite to the earth’s rotation is immedi-

ately corrected and stabilized to new balance through the

inertial instability, as the flows spontaneously emit the

gravity waves (Holton 2004). Kaplan et al. (2005b) pro-

posed the North Carolina State University turbulence

index 2 (NCSU2), which is the cross product between the

vertical vorticity and Montgomery streamfunction on an

isentropic surface, to diagnose severe turbulence in a

strong ageostrophic flow. Kaplan et al. (2006) showed that

the NCSU2 index is maximized in a region where the

ageostrophically induced strong gradient of the stream-

wise vertical vorticity and the pressure gradient force is

orthogonal in a supergradient flow regime, in contrast to

the geostrophic flow where the pressure gradient force

and vorticity gradient vectors are parallel. This finally

transits to the horizontal vortex tubes (HVTs; Clark et al.

2000) and causes localized turbulence. Recently, Knox

et al. (2008) proposed a turbulence index based on the

Lighthill–Ford theory (LHF) of spontaneous imbalance

and emission of the inertia–gravity waves that locally

modify the environmental Richardson number.

The numbers and percentages of TJCAT events cor-

responding to individual intensities over South Korea

from 2003 to 2008 are shown in Table 3, where the num-

bers of the LGT- and MOG-level TJCAT events are 603

and 105, respectively. Considering that the number of the

LGT- (MOG)-level CAT events in Table 2 is 1494 (227),

the relative portion of the LGT- (MOG)-level TJCAT

events among the total number of LGT- (MOG)-level

CAT events is about 40.4% (46.3%). The relative portions

of the LGT- and MOG-level TJCAT events among the

total number of TJCAT events are 17.97% and 3.13%,

respectively.


and MOG-level TJCAT events. In Fig. 9, TJCAT events

are shown to occur the most frequently in winter and

spring seasons, and the maximum relative percentages of

the LOG- and MOG-level TJCAT events among the total

number of PIREPs within a given month are about 15.6%

in February and 3.9% in March, respectively. Considering

that TJCAT events most frequently occur at the flight

levels of 25 000–30 000 ft (z 5 7.6–9.1 km), these seasonal

patterns are likely due to the seasonal variations in the

dynamic tropopause height over the Korean Peninsula,

where the mean dynamic tropopause heights are higher in

the summer and fall (z 5 12–13 km) than in the winter and

spring (z 5 9–10 km) (Lee 2008). Moreover, the monthly

variation shown in Fig. 9 is similar to that of the relative

frequency of the secondary tropopause measured by ozo-

nesonde sounding at Pohang and the satellite observations

over eastern Asia shown by Hwang et al. (2007, Fig. 8).

To examine the synoptic–dynamic structure for the

TJCAT events, we calculate and show in Fig. 10 the dif-

ferences in the composite averages between the MOG-level

TJCAT and other MOG-level CAT events (TJCAT-CAT)

for the horizontal wind speed (WS) at 200 hPa and verti-

cally averaged turbulence index 1 [TI1; Eq. (9) in Ellrod

and Knapp (1992)], NCSU2 [Eq. (4) in Kaplan et al.


LGT, MOD, and missing categories of CIT and CAT events ob-

tained from PIREPs over South Korea from 2003 to 2008. The

detailed classification methods are described in the text.

Turbulence intensity

CIT CAT

No. Percentage No. Percentage

NIL 365 49.8 4242 54.98

LGT 152 20.74 1494 19.36

MOG 28 3.82 227 2.94

Missing 188 25.65 1753 22.72

Total 733 100 7716 100

FIG. 8. Monthly counts of the LOG- and MOG-level (gray and

black bars, respectively) CIT events over South Korea from 2003 to

2008. MOG-level data are multiplied by 2 for clearer presentation.

The relative percentages of the LOG- and MOG-level CIT events

normalized by the total number of PIREPs within a given month

are depicted as dashed and solid lines, respectively.


(2006)], and Lighthill–Ford radiation [LHF; Eq. (23) in

Knox et al. 2008] between 20 000 and 30 000 ft using

RDAPS analysis data with 30-km horizontal grid spacing.

The 95% significant confidence level is shaded in all plots

in Fig. 10. Note that the seasonal frequencies in the 105

MOG-level TJCAT events [December–February (DJF),

24 (23%); March–May (MAM), 51 (48%); June–August

(JJA), 13 (13%); and September–November (SON) 17

(16%)] and the other 122 MOG-level CAT events [DJF,

24 (20%); MAM, 38 (32%); JJA, 33 (26%), and SON,

27 (22%)] are not significantly different, so that the dif-

ferences in the four parameters shown in Fig. 10 are not

significantly dependent upon seasonal changes in atmo-

spheric conditions. In Fig. 10a, the jet stream magnitudes

over South Korea are higher during the TJCAT events

than those during the other CAT events within a 95%

confidence level. Of 105 TJCAT events, 86 (82%) oc-

curred near a horizontal wind speed greater than 40 m s21,

which implies that the strong jet stream is dominant over

South Korea during the TJCAT events. In Figs. 10b–d,

the relatively larger values of the TI1, NCSU2, and LHF

indices are located over South Korea for the TJCAT

events within a 95% confidence level. And of the total 105

TJCAT events, 84 (80%), 76 (72%), and 79 (75%) oc-

curred near the grid point greater than the MOG-level

thresholds of TI1 [1.7 3 1026 s22; Sharman et al. (2006)],

NCSU2 [10 3 10212 s23; Kaplan et al. (2006)], and LHF

[1 3 10212 s23; Knox et al. (2008)] indices, respectively, in

contrast to the 50 (41%), 50 (41%), and 73 (59%) of the

total of 122 other CAT events. This result confirms the

idea that the selected TJCAT events are related to the

upper-level front/jet stream and associated with shear

instability, inertial instability, and geostrophic adjustment.

c. Mountain waves

According to several previous studies, mountain waves

are one of the important sources of CAT events. The

amplitude of vertically propagating mountain waves in-

creases with height due to decreasing air density (Hines

1960; Lindzen 1981), which leads to wave steepening,

overturning, and subsequent breaking at higher altitudes

(Doyle et al. 2005). Stationary mountain waves can break

when they approach a critical level at which the back-

ground wind speed equals zero (Lilly 1978). To isolate

CAT events related to mountain waves, possible areas

of turbulence induced by mountain-wave breaking are

diagnosed by mountain-induced gravity wave drag pa-

rameterization. The parameterization used in this study is

based on Palmer et al. (1986) and Chun et al. (1996) using

the RDAPS analysis data at 0000 and 1200 UTC. A brief

description of the parameterization is as follows. First,

mountain-wave stress along with the surface-level hori-

zontal wind vector (denoted by the subscript s) is calculated

by ts 5 2K0rsNsUsjh9j2, where K0 is the horizontal wave-

number (1.0 3 1023 is used in this study, corresponding to

a 6.28-km horizontal wavelength), rs is the surface air

density, Ns is the Brunt–Vaisala frequency at the surface, Us

is the basic-state wind projected on the surface wind vector,

and jh9j2 is the subgrid-scale topography height variance.

The variances in the subgrid-scale mountain height (jh9j2)at the grids of the RDAPS domain are derived using the

DEM dataset with a 30-s resolution. Second, regions of

wave breaking induced by mountain waves are determined

by using the minimum Richardson number including the

wave effects [Rim 5 Rik(1� «)/(1 1 «Ri1/2k )2], where Ri

is the local Richardson number at a given level (denoted

by the subscript k) and « is an inverse Froude number [« 5

(dh)kNk/Uk], where Uk is the horizontal wind projected

onto the surface wind and (dh)k 5 [jtk21j/(K0rkNkUk)]1/2.

If Rim , 1/4, wave breaking is assumed to occur, and the

saturation wave stress is calculated by tk 5 «s2 K0rkUk

3/Nk,

where «s 5 Ri21/2 (1 1 2Ri1/2)[2Ri1/4 (1 1 2Ri1/2)21/2 2 1].

Finally, mountain-induced gravity wave drag is calculated

by 2(1/rk)(›tk/›z). If Rim $ 1/4, there is no wave breaking

and the wave stress is the same as that at the level below

(Eliassen and Palm 1960). When a critical level (Uk 5 0)


LGT, MOG, and missing categories of TJCAT and MWCAT

events obtained from PIREPs over South Korea from 2003 to 2008.

The detailed classification methods are described in the text.

Turbulence intensity

TJCAT MWCAT

No. Percentage No. Percentage

NIL 1865 55.57 609 51.65

LGT 603 17.97 273 23.16

MOG 105 3.13 50 4.24

Missing 783 23.33 247 20.95

Total 3356 100 1179 100

FIG. 9. As in Fig. 8, but for TJCAT.


is encountered in the calculation, we assumed that the

gravity waves are absorbed at the critical level and then

the wave stress is set to zero at and above the critical level.

When Nk 5 0, saturation stress becomes infinity, implying

no wave breaking occurs, and the wave stress is assumed to

be the same as that at the level below. The aforementioned

steps are repeated at higher levels until the wave stress is

zero or the model top is reached. When the reported CAT

events are located near the grid point of the nonzero

mountain-induced gravity wave drag within 30 km hori-

zontally, 1000 ft vertically, and 66 h temporally, the CAT

events were regarded as being related to mountain waves.

Hereinafter, CAT events related to mountain waves are

referred to as mountain-wave-induced CAT (MWCAT)

events. It is noted that the frequency of MWCAT events

categorized in this study would tend to underestimate the

percentage of MWCAT encounters, because the classifi-

cation method used in this study is based on a hydrostatic

assumption and, therefore, may miss some events that

occurred due to the effects of wave breaking in non-

hydrostatic (e.g., trapped) waves.

The numbers and percentages of MWCAT events

corresponding to individual intensities over South Korea

from 2003 to 2008 are shown in Table 3, where the num-

bers of LGT- and MOG-level MWCAT events are 273

and 50, respectively. Since there are 1494 (227) LGT-

(MOG)-level CAT events in Table 2, the relative portions

of the LGT- (MOG)-level MWCAT events among the

total number of LGT- (MOG)-level CAT events is about

18.3% (22%). The relative portions of the LGT- and

MOG-level MWCAT events among the total number of

MWCAT events are 23.16% and 4.24%, respectively,

FIG. 10. Differences in the composite averages between the MOG-level TJCAT and other MOG-level CAT events

for the (a) horizontal wind speed (WS, m s21) at 200 hPa and the vertically averaged (b) TI1 [s22; Eq. (9) in Ellrod

and Knapp (1992)], (c) NCSU2 [s23; Eq. (4) in Kaplan et al. (2006)], and (d) LHF [s23; Eq. (23) in Knox et al. (2008)]

between 20 000 and 30 000 ft, derived using RDAPS analysis data with 30-km horizontal grid spacing. Contour

intervals in (a)–(d) are 2 m s21, 1 3 1027 s22, 1 3 10213 s23, and 1.5 3 10213 s23, respectively. Boldfaced and dashed

lines are zero and negative values, respectively, and shadings are the 95% significant confidence level.


which are slightly higher than those of the LGT- (17.97%)

and MOG-level (3.13%) TJCAT events among the total

number of TJCAT events, respectively.


and MOG-level MWCAT events. MWCAT events occur

most frequently in the spring season, and the maxi-

mum relative percentages of the LOG- and MOG-level

MWCAT events among the total number of PIREPs

within a given month are about 8.7% in May and 1.6% in

April, respectively. This result is consistent with the rel-

atively high percentages of turbulence frequencies over

the lee side of the Taebaek mountains during the spring

season, as shown in Figs. 6 and 7. The highest frequency of

downslope windstorms over the lee side of the Taebaek

mountains occurs in the springtime (Kim and Chung 2006;

Jang and Chun 2008), lending further support to the idea

that the maximum frequency of MWCAT events in this

season is likely related to mountain-wave breaking.

Another noticeable result in Fig. 11 is that relatively

higher frequencies appear during the summer season.

The relative percentages of the LOG- and MOG-level

MWCAT events among the total number of PIREPs in

June and August rise to about 7.1% and 1.53%, respec-

tively. According to the seasonal variations in the zonal

mean wind in the Northern Hemisphere, the westerlies

in the troposphere change to the easterlies in the lower

stratosphere during the summer season. Because of this

variation, the critical level for stationary mountain waves

can be located near the tropopause during the sum-

mer. In addition, expansion of the northwestern Pa-

cific high toward the Korean Peninsula occasionally

cause easterlies and southeasterlies in the lower tropo-

sphere and midtroposphere, which can induce flow re-

versal in the troposphere and increase the potential for

mountain-wave breaking over the mountainous regions

of South Korea. Under these conditions, the relatively

higher frequencies of MWCAT events can occur during

the summer season.

Figure 12 is a Venn diagram of three generation mech-

anisms (convective system, tropopause–jet stream, and

mountain waves) for the MOG-level turbulence events

that occurred over South Korea from 2003 to 2008. The

largest portion (41.2%; 105/255) of the total MOG-level

turbulence events is associated with the tropopause/jet

stream, while 19.6% (50/255) and 11% (28/255) of the total

MOG-level turbulence events are associated with the

mountain waves and convective systems, respectively.

Of the total MOG-level CAT events, 19 cases (7.4%) oc-

cur simultaneously due to the tropopause/jet stream and

mountain waves. In the present study, the RDAPS analysis

data used are interpolated equally into a 1000-ft vertical

interval, since the observed turbulence events are reported

at every 1000 ft. Some variables used to categorize the

CAT events are sensitive to the vertical resolution of the

data, such as the Richardson number, which is determined

by the vertical gradients of the wind and temperature.

When the Richardson number is calculated using the

coarse-resolution RDAPS data with a 2000-ft vertical in-

terval, slightly fewer (46) MWCAT events are selected

compared with those (50) based on the RDAPS data with

a 1000-ft vertical resolution. This is somewhat expected,

given that the higher vertical resolutions of the back-

ground wind and stability can provide the better estima-

tion of the wave-breaking heights.

5. Summary and discussion

Characteristics of aviation turbulence over South

Korea are investigated using the PIREP data accumulated

by the KAMA from 2003 to 2008. Before investigating

turbulence statistics, uncertainties in the PIREPs were

FIG. 11. As in Fig. 8, but for MWCAT events.

FIG. 12. A Venn diagram of the three mechanisms for the MOG-

level turbulence events over South Korea from 2003 to 2008. The

numbers in each category denote the frequency of occurrence as

a number and a percentage. Detailed definitions of CIT, TJCAT,

and MWCAT are given in the text.


examined by evaluating the agreement of turbulence in-

tensity reports with those observed in adjacent PIREPs

within a certain range and time period. The uncertainties in

the turbulence intensity and location in the current PIREPs

were found to be negligible.

The frequencies of the LGT-, MOD-, and SEV-level

turbulence events in the present study are higher than

those in the previous statistical study by Lee and Choi

(2003), which used PIREP data over South Korea from

1996 to 2000. In addition, increasing trends of the relative

percentages of the LOG- and MOG-level turbulence

frequencies are evident in the present study (2003–08),

implying that turbulence forecasting over South Korea

is becoming more important as air transportation has

increased over Korea as well as over eastern Asia. The

frequencies and relative percentages of the LOG- and

MOG-level turbulence events are the highest during the

spring season. In terms of spatial distributions, reported

turbulence events occur most frequently along the dom-

inant flight levels and routes, as expected, with relatively

higher frequencies along those routes near the west coast,

Jeju Island, and the Sobaek and Taebaek mountains. The

turbulence frequencies over these regions vary by season,

likely due to different source mechanisms.

To understand the generation mechanisms of the ob-

served turbulence events, three potential sources (the

convective system, tropopause/jet stream, and mountain

waves) are investigated using lightning flash data, RDAPS

analysis data, and the DEM dataset. The total aviation

turbulence events are classified into convectively induced

turbulence (CIT) and clear-air turbulence (CAT) events

depending on the existence of lightning activity. The rel-

ative portions of the MOG-level CIT and CAT events

among the total number of the MOG-level aviation tur-

bulence events are 11% and 89%, respectively. As ex-

pected, CIT events are dominant in the summer due

mainly to convective systems. The CAT events are classi-

fied as tropopause/jet stream–induced CAT (TJCAT) and

mountain-wave-induced CAT (MWCAT) events. The

relative portions of the MOG-level TJCAT and MWCAT

events among the total number of the MOG-level turbu-

lence events are 41.2% and 19.6%, respectively. TJCAT

events are dominant in the winter and spring seasons due

to the seasonal variations in the jet stream strength. Based

on the differences in composite averages for the synoptic–

dynamic structures between the TJCAT and other CAT

events, the shear instability, inertial instability, and geo-

strophic adjustment associated with the strong jet stream

are likely the generation mechanisms for the TJCAT

events over South Korea. MWCAT events are dominant

during the spring and summer seasons over the mountain

regions of South Korea, and are related to breaking and

a critical level of mountain waves, respectively.

This study extends the previous study by Lee and Choi

(2003) that analyzed the spatial and temporal distribu-

tions of aviation turbulence recorded in PIREP data over

South Korea from 1996 to 2000. In the present study,

normalization of the turbulence frequencies is applied in

all statistical results to take into account air-traffic

changes over South Korea. In addition, possible sources

of the turbulence events are examined using the available

observations (e.g., lightning data), CAT diagnostics, and

a mountain-wave drag parameterization.

Even though unavoidable uncertainties still remain in

the PIREP data, these results of the turbulence events

over South Korea can provide useful and invaluable in-

formation to pilots, dispatchers, and forecasters to help

reduce unexpected damage from the turbulence en-

counters over South Korea. Constructing a more reliable

climatology of the turbulence events over South Korea

requires collecting PIREP data over longer periods of

time. To understand more precisely the mechanisms of

the turbulence events that occurred over South Korea,

numerical modeling studies on the observed turbulence

events under various conditions are required (e.g., Kim

and Chun 2010). The current results based on observa-

tions in the Korean Peninsula under the synoptic envi-

ronments over eastern Asia and local weather conditions

associated with terrain and convective systems can pro-

vide useful information on research into aviation turbu-

lence in general over other areas of the world.

Acknowledgments. This work was supported by the

Korea Research Foundation Grant funded by the Ko-

rean Government (MOEHRD, Basic Research Pro-

motion Fund) (KRF-2007-313-C00778).

REFERENCES

Chun, H.-Y., J.-H. Jung, J.-H. Oh, and J.-W. Kim, 1996: Effects of

mountain-induced gravity wave drag on atmospheric general

circulation (in Korean with English abstract). J. Kor. Meteor.

Soc., 32, 581–591.

Clark, T. L., W. D. Hall, R. M. Kerr, D. Middleton, L. Radke,

F. M. Ralph, P. J. Nieman, and D. Levinson, 2000: Origins of

aircraft-damaging clear-air turbulence during the 9 December

1992 Colorado downslope windstorm. J. Atmos. Sci., 57, 1105–

1131.

Cornman, L. B., G. Meymaris, and M. Limber, 2004: An update on

the FAA Aviation Weather Research Program’s in situ tur-

bulence measurement and report system. Preprints, 11th Conf.

on Aviation, Range, and Aerospace Meteorology, Hyannis,

MA, Amer. Meteor. Soc., P4.3. [Available online at http://

ams.confex.com/ams/pdfpapers/81622.pdf.]

Doyle, J. D., M. A. Shapiro, Q. Jiang, and D. L. Bartels, 2005:

Large-amplitude mountain wave breaking over Greenland.

J. Atmos. Sci., 62, 3106–3126.

Dutton, J. A., and H. A. Panofsky, 1970: Clear air turbulence: A

mystery may be unfolding. Science, 167, 937–944.


Eliassen, A., and E. Palm, 1960: On the transfer of energy in sta-

tionary mountain waves. Geophys. Publ., 22 (3), 1–23.

Ellrod, G., and D. Knapp, 1992: An objective clear-air turbulence

forecasting technique: Verification and operational use. Wea.

Forecasting, 7, 150–165.

Grabowski, W. W., and T. L. Clark, 1991: Cloud–environment in-

terface instability: Rising thermal calculations in two spatial

dimensions. J. Atmos. Sci., 48, 527–546.

Hines, C. O., 1960: Internal atmospheric gravity waves at iono-

spheric heights. Can. J. Phys., 38, 1441–1481.

Holton, J. R., 2004: An Introduction to Dynamic Meteorology. 4th

ed. Academic Press, 535 pp.

Hwang, S.-H., J. Kim, and G.-R. Cho, 2007: Observation of sec-

ondary ozone peaks near the tropopause over the Korean

Peninsula associated with stratosphere–troposphere exchange.

J. Geophys. Res., 112, D16305, doi:10.1029/2006JD007978.

Jaeger, E. B., and M. Sprenger, 2007: A Northern Hemispheric

climatology of indices for clear air turbulence in the tropo-

pause region derived from ERA40 reanalysis data. J. Geo-

phys. Res., 112, D20106, doi:10.1029/2006JD008189.

Jang, W., and H.-Y. Chun, 2008: Severe downslope windstorms

of Gangneung in the springtime (in Korean with English ab-

stract). Atmosphere, 18, 207–224.

Kaplan, M. L., A. W. Huffman, K. M. Lux, J. J. Charney,

A. J. Riordan, and Y.-L. Lin, 2005a: Characterizing the severe

turbulence environments associated with commercial aviation

accidents. Part 1: A 44-case study synoptic observational

analyses. Meteor. Atmos. Phys., 88, 129–152.

——, ——, ——, J. D. Cetola, J. J. Charney, A. J. Riordan, Y.-L. Lin,

and K. T. Waight III, 2005b: Characterizing the severe turbu-

lence environments associated with commercial aviation acci-

dents. Part 2: Hydrostatic mesoscale numerical simulations of

supergradient wind flow and streamwise ageostrophic fronto-

genesis. Meteor. Atmos. Phys., 88, 153–173.

——, and Coauthors, 2006: Characterizing the severe turbulence

environments associated with commercial aviation accidents.

A real-time turbulence model (RTTM) designed for the op-

erational prediction of hazardous aviation turbulence envi-

ronments. Meteor. Atmos. Phys., 94, 235–270.

Kim, J.-H., and I.-U. Chung, 2006: Study on mechanisms and

orographic effect for the springtime downslope windstorm

over the Yeongdong region (in Korean with English abstract).

Atmosphere, 16, 67–83.

——, and H.-Y. Chun, 2010: A numerical study of clear-air tur-

bulence (CAT) encounters over South Korea on 2 April 2007.

J. Appl. Meteor. Climatol., 49, 2381–2403.

——, ——, W. Jang, and R. D. Sharman, 2009: A study of forecast

system for clear-air turbulence in Korea. Part II: Graphical

turbulence guidance (GTG) system (in Korean with English

abstract). Atmosphere, 19, 269–287.

Knox, J. A., 1997: Possible mechanism of clear-air turbulence in

strongly anticyclonic flow. Mon. Wea. Rev., 125, 1251–1259.

——, D. W. McCann, and P. D. Williams, 2008: Application of the

Lighthill–Ford theory of spontaneous imbalance to clear-air

turbulence forecasting. J. Atmos. Sci., 65, 3292–3304.

Koch, P., H. Wernli, and H. W. Davies, 2006: An event-based

jet-stream climatology and typology. Int. J. Climatol., 26,

283–301.

Koch, S. E., and Coauthors, 2005: Turbulence and gravity waves

within an upper-level front. J. Atmos. Sci., 62, 3885–3908.

Lane, T. P., and R. D. Sharman, 2008: Some influences of back-

ground flow conditions on the generation of turbulence due to

gravity wave breaking above deep convection. J. Appl. Me-

teor. Climatol., 47, 2777–2796.

——, ——, T. L. Clark, and H.-M. Hsu, 2003: An investigation of

turbulence generation mechanisms above deep convection.

J. Atmos. Sci., 60, 1297–1321.

——, J. D. Doyle, R. D. Sharman, M. A. Shapiro, and

C. D. Watson, 2009: Statistics and dynamics of aircraft en-

counters of turbulence over Greenland. Mon. Wea. Rev., 137,2687–2702.

Lee, Y.-G., and B.-C. Choi, 2003: Analysis of quantitative and

qualitative characteristics of aircraft turbulence occurred

in South Korea using PIREP data. J. Kor. Meteor. Soc., 39,321–335.

Lee, Y.-J., 2008: Regional and seasonal variations of different

tropopauses observed by radiosonde and ozonesonde. M.S.

thesis, Dept. of the Atmospheric Sciences, Yonsei University,

112 pp.

Lester, P. F., 1994: Turbulence: A New Perspective for Pilots. Jep-

pesen Sanderson, 212 pp.

Lilly, D. K., 1978: A severe downslope windstorm and aircraft

turbulence event induced by a mountain wave. J. Atmos. Sci.,

35, 59–77.

Lindzen, R. S., 1981: Turbulence and stress owing to gravity wave

and tidal breakdown. J. Geophys. Res., 86, 9707–9714.

NTSB, 2009: U.S. air carrier operations, calendar year 2005: An-

nual review of aircraft accident data. National Transportation

Safety Board NTSB/ARC-09/01, Washington, DC, 66 pp.

Palmer, T. N., G. J. Shutts, and R. Swinbank, 1986: Alleviation of

a systematic westerly bias in general circulation and numerical

weather prediction models through an orographic gravity

wave drag parameterization. Quart. J. Roy. Meteor. Soc., 112,

1001–1039.

Pantley, K. C., and P. F. Lester, 1990: Observations of severe tur-

bulence near thunderstorm tops. J. Appl. Meteor., 29, 1171–

1179.

Schwartz, B., 1996: The quantitative use of PIREPs in developing

aviation weather guidance products. Wea. Forecasting, 11,

372–384.

Shapiro, M. A., 1980: Turbulent mixing within tropopause folds as

a mechanism for the exchange of chemical constituents be-

tween the stratosphere and troposphere. J. Atmos. Sci., 37,

994–1004.

Sharman, R., C. Tebaldi, G. Wiener, and J. Wolff, 2006: An in-

tegrated approach to mid- and upper-level turbulence fore-

casting. Wea. Forecasting, 21, 268–287.

Wolff, J. K., and R. D. Sharman, 2008: Climatology of upper-level

turbulence over the contiguous United States. J. Appl. Meteor.

Climatol., 47, 2198–2214.


Statistics and Possible Sources of Aviation Turbulence over South...

Documents

Transcript of Statistics and Possible Sources of Aviation Turbulence over South...