Nat. Hazards Earth Syst. Sci.

, 12, 403–413, 2012

www.nat-hazards-earth-syst-sci.net/12/403/2012/ Natural Hazards
doi:10.5194/nhess-12-403-2012 and Earth
© Author(s) 2012. CC Attribution 3.0 License. System Sciences

Instability indices and forecasting thunderstorms: the case of

30 April 2009
S. Tajbakhsh1 , P. Ghafarian2 , and F. Sahraian3
1 Islamic
Republic of Iran Meteorological Organization, Tehran, Iran
2 Department of Meteorology, Science and Research Branch, Islamic Azad University, Tehran, Iran
3 Atmospheric Science and Meteorological Research Center, Tehran, Iran

Correspondence to: P. Ghafarian (p [email protected])

Received: 15 September 2010 – Revised: 27 November 2011 – Accepted: 16 December 2011 – Published: 17 February 2012

Abstract. In this paper, one meteorological case study for 30 ◦ C for KI, −2 ◦ C for LI and −3 ◦ C for SI suggests the oc-
two Iranian airports are presented. Attempts have been made currence and non-occurrence of thunderstorms at the Tehran
to study the predefined threshold amounts of some instability and Tabriz stations, respectively. The WRF model output
indices such as vertical velocity and relative humidity. Two of vertical velocity and relative humidity are the two most
important output variables from a numerical weather predic- important indices for examining storm occurrence, and they
tion model have been used to survey thunderstorms. The have a numerical threshold of 1 m s−1 and 80 %, respectively.
climatological state of thunder days in Iran has been deter- These results are comparable to other studies that have exam-
mined to aid in choosing the airports for the case studies. ined thunderstorm occurrence.
The synoptic pattern, atmospheric thermodynamics and out-
put from a numerical weather prediction model have been
studied to evaluate the occurrence of storms and to verify the
threshold instability indices that are based on Gordon and 1 Introduction
Albert (2000) and Miller (1972).
Using data from the Statistics and Data Center of the Iran Since 1951, several techniques for predicting severe thun-
Meteorological Organization, 195 synoptic stations were derstorms have been presented. Most of these methods are
used to study the climatological pattern of thunderstorm days applicable for large-scale systems that are dominant in a
in Iran during a 15-yr period (1991–2005). Synoptic weather given region (Fawbush et al., 1953; Miller, 1972). Doswell
maps and thermodynamic diagrams have been drawn using et al. (1981) and Maddox and Doswell (1982) have shown
data from synoptic stations and radiosonde data. A 15-km that these previously recommended methods were not suit-
resolution version of the WRF numerical model has been able for thunderstorm forecasting because all severe thun-
implemented for the Middle East region with the assistance derstorms develop due to mesoscale factors in an appropriate
of global data from University Corporation for Atmospheric synoptic context. It is, therefore, necessary to study the be-
Research (UCAR). haviour of thunderstorms in mesoscale conditions. Moore
The Tabriz airport weather station has been selected for et al. (2003) have investigated the environment of warm-
further study due to its high frequency of thunderstorms season elevated thunderstorms associated with heavy rain-
(more than 35 thunderstorm days per year) and the existence fall over the central United States. Jacobs and Maat (2005)
of an upper air station. Despite the fact that storms occur less have presented numerical guidance methods for aiding the
often at the Tehran weather station, the station has been cho- decision making related to aviation meteorological forecasts.
sen as the second case study site due to its large amount of air Ramis et al. (2009) analysed a severe thunderstorm that oc-
traffic. Using these two case studies (Tehran at 00:00 UTC, curred on 4 October 2007 in Mallorca, Spain from an obser-
31 April 2009 and Tabriz at 12:00 UTC, 31 April 2009), the vational viewpoint. It was found that the synoptic scenario
results of this research show that the threshold amounts of was characterised by the advection of warm, moist air at low

Published by Copernicus Publications on behalf of the European Geosciences Union.

404 S. Tajbakhsh et al.: Instability indices and forecasting thunderstorms

levels over the Balearic Islands and an upper-level trough 300 J kg−1 show little or no convective potential, and CIN
over mainland Spain. values greater than 200 J kg−1 are sufficient to prevent con-
Vertical velocity, relative humidity and wind shear, which vection (Knutsvig, 2009). Haklander et al. (2003) examined
can be retrieved from Numerical Weather Prediction (NWP) 32 different indices and parameters, especially in the Nether-
model output, are very useful products for diagnosing the lands, to gain statistical information on these parameters and
structure of thunderstorms. Colquhoun (1987, 1998) intro- compare their ability to forecast thunderstorms. Haklander
duced the threshold values of these variables that are needed and Delton (2003) have estimated the probability of a thun-
for thunderstorm occurrence, and he explained thunderstorm derstorm as a function of various thunderstorm predictors.
occurrence in terms of a decision tree and has shown that a They found that the thunderstorm probability depends on the
vertical velocity larger than 0.5 m s−1 at 800 hPa and a rela- following parameters (in order of importance): latent insta-
tive humidity larger than 75 % are suitable threshold values bility (especially near the surface), potential instability and
for thunderstorm occurrence. conditional instability.
Systematic methods to diagnose the atmosphere’s poten- In the remaining sections of this paper, the synoptic con-
tial to produce severe convective weather can be of great ditions associated with thunderstorm events are reviewed for
help to forecasters. The most notable of these systematic the two case studies. In addition, NWP model output and
approaches, the methods developed by Gordon and Al- the threshold values of important thermodynamic instability
bert (2000) and Miller (1972), applied in this paper. The indices for these two case studies are presented. Section 2 de-
checklists and reference guides developed by Gordon and scribes the synoptic and upper-air data that are necessary to
Albert, and Miller are specifically designed to aid in the di- draw the synoptic, thermodynamic and climatological charts.
agnosis of severe convective conditions. Also in this section, the WRF model is described in more
Instability is a critical factor in the development of severe detail. The climatology of thunderstorm days is used to de-
weather, and severe weather instability indices can be a use- scribe the level of thunderstorm occurrence in Iran. This cli-
ful tool when applied correctly to a given convective weather matology has been analysed for a period of 15 yr, and the
situation. Severe weather indices only indicate the potential stations with the most and least thunderstorm days have been
for convection. A lifted index with a value of 0 ◦ C is suffi- noted. The climatology of thunderstorm days in Iran is pre-
cient for severe weather development only if the dynamics sented in Sect. 3. In Sect. 4, two case studies are analysed
are very strong. When the lifting index is −8 ◦ C or less, using threshold values from NWP model output and thermo-
severe weather can occur with very weak upper-air support dynamic indices for thunderstorm forecasting, and the simi-
(Hales, 1996). Some of the most common indices that are larity with other studies is shown. The verification results are
applied in the study of thunderstorms are used in this paper: discussed in Sect. 5. Finally, a discussion of the potential for
the K Index (George, 1960), the Lifted Index (Galway, 1956), these indices to improve forecasts is given in Sect. 6, along
the Showalter Stability Index (Showalter, 1953), Total To- with possibilities for future improvements.
tals (Miller, 1972), the Severe Weather Threat Index (Miller,
1972), Convective Available Potential Energy (CAPE) and
Convective Inhibition (CIN). A brief history of the applica- 2 Data
tion of these indices is described below.
In 1960, George introduced the relationship between KI To identify the most appropriate indices for thunderstorm
and thunderstorm probability. According to his research, development at the two chosen airports in Iran, we needed
thunderstorms will occur with a 50 % probability if the KI to obtain more information about the climatology of thun-
value is 26 ◦ C or more. derstorm days in the country. To better understand the
In 1956, Galway showed the critical values of LI for insta- thunderstorm climatology, we investigated the annual aver-
bility and argued that a negative LI indicates that the bound- age number of thunder days that occurred in a 15-yr pe-
ary layer is unstable with respect to the middle troposphere. riod. The primary data for this study were obtained from
This instability represents an environment in which convec- the archives of the I.R. of the Iran Meteorological Orga-
tion can occur. Galway declared that LI values between −3 nization for the years 1990–2005 at 195 synoptic stations.
and −5 ◦ C represent marginal instability. Showalter (1953) The Surfer software programme and the Kriging interpo-
presented the threshold values of SI for thunderstorms, indi- lation method were used to analyse the data. The synop-
cating that SI values from 0 to −4 ◦ C are suitable for mod- tic conditions were analysed using Global Forecast System
erate thunderstorms. The Total Totals and SWEAT indices (GFS) data with 1◦ resolution in both latitude and longi-
were described by Miller in 1972. A SWEAT index be- tude (http://nomads.ncdc.noaa.gov/data.php). The thermo-
tween 300 and 400 indicates moderate thunderstorm activity, dynamic diagrams were completed with upper-air sounding
while a TT index between 50 and 55 ◦ C shows the possi- data from the University of Wyoming (http://weather.uwyo.
bility of severe thunderstorm activity. Generally, tornadoes edu/upperair/sounding.html). Because of their importance
are possible with thunderstorms when the SWEAT index in forecasting thunderstorms on the mesoscale, the vertical
reaches more than 400 (Henz, 2009). CAPE values less than velocity, relative humidity and temperature fields from the

Nat. Hazards Earth Syst. Sci., 12, 403–413, 2012 www.nat-hazards-earth-syst-sci.net/12/403/2012/

S. Tajbakhsh et al.: Instability indices and forecasting thunderstorms 405

Table 1. Parameterization of WRF model configuration.

Fully compressible no hydrostatic dynamics

15-km horizontal grid increment over Iran and 28 vertical levels
Radiative upper-boundary condition that mitigates noise result-
ing from reflecting gravity waves
Kain–Fritsch version 2 (Kain, 2004) cumulus parameterisation:
month-long sensitivity tests showed this scheme produced the
most realistic precipitation patterns and amounts compared to
satellite-derived estimates
Lin et al. (1983) microphysics, which predicts the mixing ratio
of four hydrometeor species (cloud droplets, cloud ice, rain, and
snow)
Mellor-Yamada-Janjic PBL scheme (Janjic, 2002a,b). A set
of multiday sensitivity experiments showed that this scheme is
suitable over Iran.
Noah land surface model with four soil layers (Chen and Dud-
hia, 2001a, b)
Cloud effects on radiative transfer, Goddard scheme for short- Fig. 1. Geographical position of Iran.
wave and Rapid Radiative Transfer Model (Mlawer et al., 1997)
for longwave the westerlies are dominant over Iran. The westerlies steer
Mediterranean disturbances toward Iran, causing instability
and precipitation. Gheibi et al. (2003) presented temporal
WRF regional model were used in this analysis. Table 1 and spatial distributions of thunderstorms in Iran and showed
shows the parameterisations that were chosen for producing that the annual distribution of thunderstorms is concentrated
the best behaviour of the atmosphere within the boundary over the northwestern part of Iran. Thunderstorms usually
layer and for maximising the numerical stability. Azadi et occur during the spring and summer in Iran (Alijani, 2009),
al. (2009) show that this parameterisation configuration pro- and for this reason, we selected our case studies in April.
vides good results over Iran. As noted earlier, the two airports (Tehran and Tabriz) are se-
lected for further study due to their relatively large number
of thunderstorm days and air traffic. Tehran is located on the
3 The climatology of thunderstorm days in Iran southern slopes of the Alborz Mountains (35.7◦ N, 51.4◦ E).
The city has a dry climate with warm summers and relatively
A thunderstorm occurs in response to synoptic-scale and cold winters. Although the Alborz Mountains are far from
mesoscale forcing. To develop a thunderstorm it usually the Tehran station (approximately 70 km), the lifting effect
needs warm and moist air at low levels, cool and dry air at up- of the mountains may intensify the vertical motion due to
per levels and a synoptic-scale disturbance. Under these con- frontal convergence (Fig. 1). The Tabriz station (38.6◦ N,
ditions, thunderstorm formation is probable. In fact, thunder- 46.3◦ E) is located in northwestern Iran. The Sabalan Moun-
storm development requires large instability, vertical forcing, tains are located east of the city of Tabriz, while the Sahand
water vapour convergence at low levels and a high convec- Mountains are located south of the city (Fig. 1). Due to the
tive available energy; the synoptic-scale and mesoscale fea- orographic conditions, Tabriz has a cold and moist climate.
tures described above favour these conditions. Several cli- Its topography provides favourable conditions for thunder-
matological studies have examined thunderstorms and their storm formation. Warm air rising and cold air sinking (con-
associated phenomena such as hail, heavy precipitation and vection) play a key role in the formation of severe thunder-
severe winds. A study by Whiteman (2003) that considered storms. If the warm air at the surface is forced to rise, it will
the influence of orography on thunderstorms describes some continue to rise because it is less dense than the surrounding
of these climatological issues very well. Iran is located in a air. In addition, it will transfer heat from the land surface
dry to semi-dry climate zone; therefore, moisture is gener- to the upper levels of the atmosphere through the convection
ally not plentiful, and a large number of thunder days are not process. The orographic conditions cause thunderstorm for-
expected. During the warm season, the Azores high pressure mation by forcing air upward along the side of a mountain or
dominates over Iran via the Mediterranean Sea. The Azores large hill.
high covers the area south of the Alborz Mountains, and sta- Figure 2 shows the monthly distribution of thunderstorm
ble conditions cover all of Iran. During the cold season, days for the stations at Tehran and Tabriz. Both of the

www.nat-hazards-earth-syst-sci.net/12/403/2012/ Nat. Hazards Earth Syst. Sci., 12, 403–413, 2012

406 S. Tajbakhsh et al.: Instability indices and forecasting thunderstorms

10
9
TEHRAN
8
TABRIZ
No. of TS days

7
6
5
4
3
2
1
0
Jan. Feb. Mar. Apr. May Jun. Jul. Agu. Sep. Oct. Nov Dec.

Month

Fig. 2. Monthly average of 15 yr period (1991–2005) thunder storm

days for Tabriz (dark gray line) and Tehran (light gray line).

stations have similar distributions that confirm the study per- Fig. 3. Climatology of thunderstorm days (annual) in Iran during
formed by Alijani (2009). Figure 2 shows that there is a max- the period 1991 to 2005.
imum of thunderstorm days in May. Thunderstorm days de-
crease during the summer and reach a small peak in October.
The number of thunderstorm days peaks at approximately 9 this reason, we study some of the most important indicators
and 4.5 days for Tabriz and Tehran, respectively. for identifying and predicting storms at the Tabriz Airport
A climatological study of thunderstorm events can help to because of its large number of days with thunderstorms and
identify airport stations with more thunder days. The pat- at the Tehran Airport because of its large number of flights.
tern of thunderstorm days in Iran is shown in Fig. 3, which
shows that thunderstorms are not common in Iran. Addition-
ally, Fig. 3 shows that the minimum number of stormy days 4 Case study of a thunderstorm event on 30 April 2009
occurs at the Jask station (one of Iran’s southern islands in
the Oman sea) with an average of 1.7 days annually and that In considering the relationship between synoptic features and
the maximum number of stormy days occurs at the Makoo thunderstorm probability, the most important empirical in-
station in northwestern Iran with an average of 52 days an- dices to consider for severe weather conditions (instability
nually. Examining Fig. 3 in the context of the climatological, indices) are KI, LI, SI, TT and SWEAT. These instability in-
meteorological and orographic conditions described above, a dices were calculated for the Tabriz and Tehran stations at
more detailed picture of Iran’s thunderstorm climatology can 00:00 UTC, 30 April 2009 and 12:00 UTC, 30 April 2009,
be seen. For example, the maximum in thunderstorm days respectively. The Tabriz thunderstorm event was observed to
in northwestern Iran can be attributed to the orographic fea- be less than 8 km from the station. WRF model output of ver-
tures found in that region. In contrast, the central, eastern tical velocity, relative humidity and temperature were used to
and southeastern parts of Iran have the minimum number of predict the thunderstorms. Gordon and Albert (2000) present
thunderstorm days due to the presence of the Loot and Kavir two checklists for identifying and forecasting thunderstorms
deserts, which have very little moisture. The highest num- using NWP model output and instability indices. These crite-
ber of thunderstorm days occurs in the western half of the ria are based on studies by various researchers that took place
country, particularly in the northwestern part of the country during and after the 1970s.
where storms occur more than 20 days of the year. Stations
with more than 30 thunder days, including Makoo, Safi-Abad 4.1 Synoptic conditions
Dezfool, Jolfa, Tabriz, Ahar, Sarab, Zanjan and Khoy, are
mostly located in northwestern Iran. Stations with more than On 30 April 2009, a cold front passed over the northern
45 stormy days are marked in Fig. 3. Additionally, more than half of Iran and produced a moderate thunderstorm with rain
70 stations have less than 10 thunderstorm days annually. showers in parts of Iran including the Tabriz airport station
These stations are mostly located in the central and south- at 00:00 UTC and the Tehran airport station at 12:00 UTC.
eastern portions of the country. There are approximately The synoptic patterns and thermodynamic diagrams for this
16 days per year with thunderstorms at the Tehran station system were reviewed in detail.
in Tehran. It is clear that the prediction of thunderstorm days The surface map at 00:00 UTC shows cold air from Siberia
is not a priority at airports with either a relatively small num- intruding into the eastern part of Iran via a high pressure
ber of thunderstorm days or a small number of flights. For system and the gradual intensification of a trough of polar

Nat. Hazards Earth Syst. Sci., 12, 403–413, 2012 www.nat-hazards-earth-syst-sci.net/12/403/2012/

S. Tajbakhsh et al.: Instability indices and forecasting thunderstorms 407

Fig. 4. (a) Surface chart include isobars with 2.5 mb intervals on Fig. 5. (a) 850 hPa chart include contours with 40 m and isotherms
30 April 2009 at 00:00 UTC. (b) Surface chart include isobars with with 2.5 ◦ C intervals, respectively, on 30 April 2009 at 00:00 UTC.
2.5 mb intervals on 30 April 2009 at 12:00 UTC. (b) 850 hPa chart include contours with 40 m and isotherms with
2.5 ◦ C intervals, respectively, on 30 April 2009 at 12:00 UTC.

low pressure centre over the northwestern portion of Iran. troughs intermittently passed across the northwestern and
A pressure gradient was present on the west coast of the central Alborz Mountains at 00:00 and 12:00 UTC, respec-
Caspian Sea and on the southern slopes of the central Al- tively (Fig. 6a and b). One of these troughs, while passing
borz Mountains while there was a centre of high pres- over the central Alborz Mountains, produced a thunderstorm
sure (1015 hPa) in the western part of Iran (Fig. 4a). By event in the Tehran area, which was reported at 12:00 UTC.
12:00 UTC, the pattern changed as the polar low deepened in In addition, the temperatures at 500 hPa showed weak ad-
the north and the Siberian high weakened (Fig. 4b). These vection of cold air in the western half of Iran. At 300 hPa,
conditions produced a pressure gradient in the southwest- there was no discernible strong jet streak at either 00:00 or
ern portion of the Alborz Mountains. The surface pres- 12:00 UTC. However, a 60-knot wind speed (30 m s−1 ) was
sure was 1011 hPa (1009 hPa) at 00:00 UTC and 1009 hPa reported at the Rasht upper air station at 00:00 UTC (Fig. 7a),
(1005 hPa) at 12:00 UTC, while the dewpoint temperature while only a 20 knot (10 m s−1 ) wind speed was recorded at
was ∼10◦ C (6 ◦ C) at the Tehran (Tabriz) station at both the Tabriz station at this time. At the Tehran station, the
times. At 00:00 UTC, the Tabriz station was affected by a wind speed values at 00:00 UTC and 12:00 UTC were 45
ridge of high pressure, but the Tehran station was not yet and 50 knots (22 and 25 m s−1 ), respectively. These val-
affected. The 850-hPa chart shows an approaching low pres- ues support the fact that no considerable velocity was re-
sure trough in the northwestern strip of Iran at 00:00 UTC ported for the jet stream that passed over Tehran. The Rasht
(Fig. 5a). Twelve hours later (Fig. 5b), the region was charac- and Tabriz stations only had sounding data at 12:00 UTC
terised by a thermal trough that extended from the northwest (Fig. 7b); therefore, the unstable conditions can only be qual-
to the southern slopes of the central Alborz Mountains. This itatively derived from synoptic conditions at the Tehran sta-
pattern provided appropriate conditions for warm advection tion at 12:00 UTC.
in the mid-western parts of Iran and cold advection along
the slopes of the Alborz Mountains. At 500 hPa, shallow

www.nat-hazards-earth-syst-sci.net/12/403/2012/ Nat. Hazards Earth Syst. Sci., 12, 403–413, 2012

408 S. Tajbakhsh et al.: Instability indices and forecasting thunderstorms

Fig. 6. (a) 500 hPa chart include contours with 40 m and isotherms Fig. 7. (a) 300 hPa chart on 30 April 2009 at 00:00 UTC.
with 2.5 ◦ C intervals, respectively, on 30 April 2009 at 00:00 UTC. (b) 300 hPa chart on 30 April 2009 at 12:00 UTC.
(b) 500 hPa chart include contours with 40 m and isotherms with
2.5 ◦ C intervals, respectively, on 30 April 2009 at 12:00 UTC.
SI, LI, KI, SWEAT, TT, CAPE and CIN were calculated
(Table 2). Because the Tabriz event did not occur exactly
Table 2. Instability indices for Tehran and Tabriz stations on
30 April 2009.
over the station at 00:00 UTC, the thermodynamic diagram
shown in Fig. 8a does not contain considerable instability.
There is no significant observed jet stream in the 200 to
Tehran station Tabriz station
30 April 2009 at 12:00 UTC 30 April 2009 at 00:00 UTC
300 hPa layer, and the values of the instability indices indi-
cate a non-storm environment, which is shown in Table 2.
value Index value Index Due to the lack of significant instability, we did not calculate
−2.95 LI 2.29 LI CAPE or CIN at this station and time. By 12:00 UTC,
−2.3 SI 2.17 SI Tehran had experienced a relatively significant storm; hence,
36.8 KI 26.4 KI the dewpoint and temperature profiles indicate unstable con-
54.6 TT 49 TT ditions (strong convergence of the temperature and humidity
256 SWEAT 94.6 SWEAT profiles) and a 60 knot wind velocity in the 200 to 300 hPa
313.6 CAPE – CAPE layer (Fig. 8b). Based on the criteria for thunderstorm
−17 CIN – CIN occurrence from Galway (1956) and George (1960), the LI,
KI and SI indices in Tehran at 12:00 UTC are acceptable for
thunderstorm formation.
4.2 Thermodynamic diagrams
4.3 NWP model output
To study the status of thunderstorm development
more precisely, thermodynamic diagrams from the In this study, the WRF model was run and important model
Tehran (12:00 UTC) and Tabriz (00:00 UTC) stations results were investigated. The most important of these re-
were reviewed. Atmospheric instability indices including sults are the vertical velocity cross-section, temperature and

Nat. Hazards Earth Syst. Sci., 12, 403–413, 2012 www.nat-hazards-earth-syst-sci.net/12/403/2012/

S. Tajbakhsh et al.: Instability indices and forecasting thunderstorms 409

Table 3. Gordon severe weather check list for Tehran and Tabriz stations on 30 March 2009.

Severe weather parameters Indicators Answer Y/N Answer

(Tabriz) Y/N
(Tehran)
√ √
Depth of low level moisture now (or expected) to be greater than 3000 ft?
Surface dew point >14◦ C? – –
Low level moisture and temperature Distinct low level surface temperature axis present? Location/Time?
Will the 850 hPa max temperature ridge be over or west of the 850 hPa moisture over over
axis?
Is a low level jet present or expected to develop? Location/Time? – –
Low level jet What is the Highest 850 hPa jet speed 14 knots 11 knots
850 hPa moisture convergences expected? Location/Time? – –
Upper Level Support Will there be a 300/250 hPa jet > 65 knots? – –
√ √
Are any lifting mechanisms such as fronts or outflow boundaries present? List
Lifting mechanism them along with their location.
√ √
Will lifting mechanisms be able to overcome capping inversion?
√
CIN < 50 J kg−1 ? –

Winds show significant veering (0–3 km shear values > 35 kts)? – –

Vertical wind shear √
Is there (or will there be) speed shear > 25 kts and/or directional shear > 30 –
degrees between 850 and 500 hPa?
√
Is/will the Lifted Index be <0? –
√
Is/will the K Index be >30? –
C Will CAPE > 800 J kg−1 ? – –
Instability
Will there be high mid level lapse rates (700–500 hPa) 6.5 ◦ C km−1 ? (6.8 ◦ C Km−1 )
√
√
Will there be warm advection at 850 hPa? –
√
Will a significant capping inversion remain in place? –

700 mb Dry Intrusion Is there or will there be a dry intrusion of air at or near the 700 mb level? (Dew – –
point depression >6◦ C)
√ √
Upper Vertical Motion Is large scale forcing indicated by model Q vector and omega fields?
√ √
Satellite Imagery/ Cloud Indicators Are there lines of cumulus or mid clouds (altocumulus castellanus–ACCAS) on
the morning satellite imagery? Is there significant mid level drying present on
water vapor imagery?
√ √
Is there or will there be strong surface pressure falls?
Surface Pressure Falls
Will there be a corresponding pressure rise moving toward the fall area? (The – –
larger the absolute value of this rise-fall couplet, the larger the potential for
severe weather in the pressure fall area)

relative humidity patterns at the 700 hPa level. These pat- between 0.5 and 1 m s−1 , which are in agreement with the
terns were reviewed to identify characteristics of the thun- results from Colquhoun (1987, 1998). Figure 10 shows val-
derstorm events at 00:00 UTC on 30 April 2009 at the Tabriz ues of humidity and temperature at 700 hPa. Figure 10a
station and 12:00 UTC at the Tehran station. Figure 9a rep- (Fig. 10b) shows that the temperature at the Tabriz (Tehran)
resents the vertical velocity profile (VVP) at 00:00 UTC, and station is ∼2 ◦ C (4 ◦ C) with a relative humidity of less than
it shows a weak downdraft at 700 hPa over the Tabriz sta- 60 % (90 %).
tion, while above this level, there is a weak updraft with
VVP values of 0.3 m s−1 . There is nearly zero VVP in the 4.4 The Gordon and Miller checklists
700- to 800-hPa layer. Figure 9a closely corresponds with
the observed values by showing that at 00:00 UTC the thun- The severity of the thunderstorm events in these two case
derstorm was not active in Tabriz. Figure 9b shows a ver- studies was investigated using the data described above and
tical cross-section of VVP. Regions of upward motion were the checklists provided by Gordon and Albert (2000) and
observed in the 700 to 800 hPa layer, and the values were Miller (1972). Table 3 is a checklist that can be used to

www.nat-hazards-earth-syst-sci.net/12/403/2012/ Nat. Hazards Earth Syst. Sci., 12, 403–413, 2012

410 S. Tajbakhsh et al.: Instability indices and forecasting thunderstorms

Fig. 9. (a) Vertical cross-section of vertical velocity (m s−1 ), on

30 April 2009 for Tabriz station. (b) Vertical cross-section of verti-
Fig. 8. (a) Thermodynamic diagrams on 30 April 2009 for
cal velocity (m s−1 ), on 30 April 2009 for Tehran.
Tabriz station at 00:00 UTC. (b) Thermodynamic diagrams on
30 April 2009 for Tehran at 12:00 UTC.
12 h (00:00 UTC to 12:00 UTC) is between 0 and −3 hPa.
In addition, the temperature axis at 850 hPa is located to
the east of the moisture axis, and the dewpoint temperature
predict and identify thunderstorms and their severity. As the is less than 8 ◦ C. The maximum jet stream velocity is less
table shows, of the 10 important parameters used to forecast than 66 kt (33 m s−1 ) in the 200- to 300-hPa layer, and the
thunderstorms and severe weather, approximately 6 of these mean relative humidity is ∼77 %. Using Table 2, we can
parameters show a positive response for the Tehran station. determine the instability indices: TT = 54.6 ◦ C, LI = −3 ◦ C,
In fact, answers to the questions related to parameters 4, 5, 6, CAPE = 314 J Kg−1 , SWEAT = 256 and SI = −2 ◦ C. Based
8, 9 and 10 are positive. Hence, based on the Gordon criteria, on Table 3, most of the Miller checklist parameters indicate
the event can be categorised into the organised thunderstorm a moderate to severe thunderstorm event. The same analysis
group due to the number of positive responses. To evaluate was performed for the Tabriz station. The results are shown
the severity of the event at the Tehran station at 12:00 UTC in Table 3. Using radiosonde data from the Tabriz station
on 30 April 2009, we used the Miller checklist. The sur- (Fig. 7a), the data show that only 2 of the parameters (1 and
face pressure map (Fig. 4b) and the thermodynamic diagrams 3) were positive, while the remaining 8 parameters were neg-
(Fig. 8b) show that the parameters in the Miller checklist (Ta- ative. This result means that the thunderstorm is not active
ble 3) have the following values at the Tehran station: the sur- and, according to Gordon, it will not be necessary to proceed
face pressure is more than 1010 hPa, the dewpoint is greater to the Miller checklist because there are less than 6 positive
than 13 ◦ C and the change in the surface pressure within the answers in the Gordon checklist.

Nat. Hazards Earth Syst. Sci., 12, 403–413, 2012 www.nat-hazards-earth-syst-sci.net/12/403/2012/

S. Tajbakhsh et al.: Instability indices and forecasting thunderstorms 411

The climatology of thunderstorm days in Iran shows that

statistically only a few stations have more than 50 days with
thunderstorms and that a slightly larger number of stations
have more than 30 days annually with thunderstorms. Be-
cause of this climatology, the country can be considered as
a region with a limited number of thunderstorm events. De-
spite the fact that most of these stations are not airport sta-
tions, the thunderstorm events are still considered important
because of the risks to flight operations associated with thun-
derstorms. Therefore, we chose to investigate two impor-
tant regions for our case study: the Tehran station due to
its high number of flights (more than 400 aircraft landing
and taking off daily) and the Tabriz station due to its rela-
tively high number of thunder days. The synoptic patterns
in this case study indicate relatively favourable conditions
for instability in Tehran. A surface pressure gradient, tem-
perature advection at 850 hPa and the presence of cold air
in the upper levels provided suitable conditions for instabil-
ity in the lower layers of the atmosphere. The thermody-
namic diagrams for the Tabriz station show a shallow radia-
tion inversion at 00:00 UTC that is indicative of stability at
very low levels. There was also no considerable moisture at
the station. In contrast, at 12:00 UTC in the Tehran region,
no lower-level temperature inversion was observed, and the
moisture content was higher than at the Tabriz station. Be-
cause the moisture content that is observed at Tehran is still
not significantly large, the thunderstorm intensity was mod-
erate to light. Two of the most important causes of severe
weather conditions are high humidity in the lower levels and
rising air; hence, these two parameters are the most impor-
tant NWP model output variables for thunderstorm forecast-
ing. In our study, humidity of more than 90 % and a ver-
tical velocity equal to 1 m s−1 at 700 hPa produced a rela-
tively severe thunderstorm in Tehran. Atmospheric instabil-
ity indices including KI and SI confirmed a weak to mod-
Fig. 10. (a) Temperature and relative humidity at 700 hPa on erate thunderstorm in Tehran, while no thunderstorm was
30 April 2009 for Tabriz station. (b) Temperature and relative hu- forecast for Tabriz. These results are verified with observa-
midity at 700 hPa on 30 April 2009 for Tehran. tions. Haklander and Delden (2003) determined the thresh-
old values for important instability indices for the Nether-
5 Discussion lands using statistical methods and 32 different parameters
to determine the ability of those threshold values to forecast
In our case studies, the synoptic charts present a weak pres- thunderstorms. They found the following threshold values:
sure gradient over Tehran and Tabriz. Pressure gradients play KI ≥ 21◦ C, LIsfc ≤ 1.6 ◦ C, SHOW ≤ 4.2 ◦ C, TT ≥ 46.7 ◦ C,
a main role in providing convergence and updrafts. Conver- CAPE ≥ 173 J Kg−1 and SWEAT ≥ 134. A comparison of
gence is one of the trigger mechanisms needed for providing the values of these indices with the two case studies in Iran
upward motions, which are part of the foundation for thun- (Table 3) confirms the occurrence and non-occurrence of
derstorm formation. Even a very moist atmosphere cannot thunderstorms for the Tehran and Tabriz stations, respec-
produce strong thunderstorm development without a conver- tively.
gence mechanism. In our cases, the role of the orographic Applying the Gordon checklist to the Tehran case study,
features clearly aids in preparing conditions for ascent. For it was found that the answers to 6 of the 10 questions were
example, despite the weak pressure gradient, the topographic positive. This result indicates that the event is categorised in
conditions helped to form stormy weather at the Tehran and the organised thunderstorm class. To determine the severity
Tabriz airports. The Sahand and Sabalan Mountains around of the thunderstorm event, the Miller checklist was used to
the city of Tabriz and the Alborz Mountains north of Tehran examine certain atmospheric parameters associated with the
(Fig. 1) have produced lifting for thunderstorm formation. event (Table 4). The table shows that the Tehran case may

www.nat-hazards-earth-syst-sci.net/12/403/2012/ Nat. Hazards Earth Syst. Sci., 12, 403–413, 2012

412 S. Tajbakhsh et al.: Instability indices and forecasting thunderstorms

Table 4. Miller severe weather checklist.

No. Parameter Weak Moderate Strong

1 Surface Pressure >1010 mb 1010 to 1005 mb <1005 mb
2 Surface Dew Point <55◦ F 55–64◦ F 65◦
3 12 h Surface Pres Change 0 to −3 −4 to −7 −8
4 850 mb Temp Axis East of Moist Axis Over Moist Axis West of Moist Axis
5 850 mb Jet <25 kts 25–35 kts >35 kts
6 850 mb Dew Point 8 ◦C 8–12 ◦ C >12 ◦ C
7 700 mb Dry Intrusion N/A or Weak 700 mb winds Winds from dry to moist intrude Winds intrude at an 40◦ and are
at <40◦ and are 15 kts 25 kts
8 700 mb Temp No Change Winds cross line 20◦ Winds cross line >20◦ and 40◦ Winds cross line >40◦
Line
9 500 mb Height Change <30 m 30 and 59 m 60 m
10 500 mb Wind Speed 35 kts 36–49 kts 50 kts
11 500 mb Vorticity Advection Neutral or NVA PVA-Contours Cross Vorticity PVA-Contours Cross Vorticity
Pattern 30◦ Pattern >30◦
12 850–500 mb Wind Shear
(a) Speed Shear 15–25 kts 26–35 kts >35 kts
(b) Directional Shear 20–30◦ 30–60◦ >60◦
13 300–200 mb Jet 65 kts 66–85 kts >85 kts
14 Mean R.H. 70–80 % or 40–50 % 50–70 % 50–70 %
15 TT <50 50–55 >55
16 LI > −2 −3 to −5 −6
17 CAPE 800–1500 J kg−1 1500–2500 J kg−1 >2500 J kg−1
18 SWEAT <300 300–500 >500
19 WBZ >11 000 ft 9000–11 000 ft 7000–9000 ft
<5000 ft 5000–7000 ft
20 Helicity (0–3 km) 150–300 m2 s−2 300–450 m2 s−2 >450 m2 s−2
21 SI −1 to +2 −1 to −3 < −3

be classified as a weak thunderstorm. In contrast, the event instabilities that may lead to thunderstorm development. In
that occurred at the Tabriz station was not placed into the addition, climatological information concerning storm oc-
organised class of thunderstorms found in the Gordon check- currence can aid in statistically determining the important
list. Therefore, according to the Miller checklist, there was indicators and phenomena that create these storms. The cli-
no need to review the severity of the thunderstorm because matology of thunderstorm days in Iran shows that some me-
it was very weak. In fact, the Gordon and Miller check- teorological stations in the western portion of the country
lists provide synoptic and mesoscale methods, respectively, have thunderstorms on more than 30 days a year. This study
to study thunderstorm events using specific criteria that con- showed that KI, LI and SI are appropriate indicators of insta-
sider all related parameters. If NWP models were config- bility. We determined suitable threshold values for thunder-
ured to answer the questions on the checklists, they could be storms that occurred at the Tehran and Tabriz airport stations:
considered a good guide for forecasting thunderstorms and 30 ◦ C for KI, −2 ◦ C for LI and −3 ◦ C for SI are indicators of
determining their severity. moderate to severe thunderstorms. Humidity and vertical ve-
locity from NWP models are the most important variables
for identifying atmospheric instability and can be consid-
6 Conclusions ered proper tools to support thunderstorm forecasting. The
threshold values of relative humidity and vertical velocity are
Atmospheric instability is a critical factor in determining 80 % and 1 m s−1 , respectively. Additional tools, such as the
if thunderstorm development will commence. Certain pa- Gordon and Miller checklists, can help and guide weather
rameters, including vertical motion and low-level moisture forecasters to predict thunderstorms. Although the threshold
content, are used to review and evaluate the atmospheric

Nat. Hazards Earth Syst. Sci., 12, 403–413, 2012 www.nat-hazards-earth-syst-sci.net/12/403/2012/

S. Tajbakhsh et al.: Instability indices and forecasting thunderstorms 413

values in these checklists have already been determined for Gheiby, A., Sen, N., Puranik, D., and Karekar, R.: Thunderstorm
North America, we found that using these checklists can also identification from AMSU-B data using an artificial neural net-
be a suitable approach for forecasting thunderstorms in Iran. work, Meteorol. Appl., 10, 329–336, 2003.
Certainly, for these checklists to be used in an operational Gordon, J. and Albert, D.: A comprehensive severe weather fore-
forecasting environment, additional statistical analyses will cast checklist and reference guide, Noaa technical service publi-
cation, TSp 10, NWS central region, 2000.
need to be performed that include local and regional factors
Hales, E.: Severe weather forecasting, Personal SPC notes, Severe
such as terrain and elevation for the particular location of in- Weather Forecast., 1996.
terest. Haklander, A. J. and Delton, A. Van.:Thunderstorm predictors and
their forecast skill for the Netherlands, Atmos. Res., 67–68, 273–
Acknowledgements. The authors are grateful to the Editor and
299, 2003.
the anonymous reviewers for the precious remarks and comments
Henz, D.: Soundings and Adiabatic Diagrams for Severe Weather
which led to improve the initial version of this paper.
Prediction and Analysis, AOS, 453, 2009.
Jacobs, A. J. M. and Maat, N.: Numerical guidance methods for
Edited by: M.-C. Llasat
decision support in aviation meteorological forecasting, Weather
Reviewed by: three anonymous referees
Forecast., 20, 82–100, 2005.
Janjic, Z. I.: A Nonhydrostatic Model Based on a New Approach,
EGS XVIII, Nice France, 21–26 April 2002, 2002a.
References
Janjic, Z. I.: Nonsingular Implementation of the Mellor–Yamada
Alijani, B.: Iran Climate, Payam nour pres., Iran, 2009. Level 2.5 Scheme in the NCEP Meso model, NCEP Office Note,
Azadi, M., Kaffash, N., and Abdollahzadeh, I.: Probabilistic pre- 437, 61 pp., 2002b.
cipitation forecasting using ensemble forecasting system over Kain, J. S.: The Kain–Fritsch convective parameterization: An up-
Iran., 8th Conference on Numerical Weather Prediction, At- date, J. Appl. Meteor., 43, 170–181, 2004.
mospheric Science and Meteorological Research Center (AS- Knutsvig, R.: “Severe Weather Indices”, University of North
MERC), Tehran, Iran, 2009. Dakota, available at: http://www.geocities.com/weatherguyry/
Chen, F. and Dudhia, J.: Coupling an advanced land-surface/ hy- swx2.html, 2009.
drology model with the Penn State/NCAR MM5 modeling sys- Lin, Y.-L., Farley, R. D., and Orville, H. D.: Bulk parameterization
tem, Part I: Model implementation and sensitivity. Mon. Weather of the snow field in a cloud model, J. Clim. Appl. Meteor., 22,
Rev., 129, 569–585, 2001a. 1065–1092, 1983.
Chen, F. and Dudhia, J.: Coupling an advanced land- Maddox, R. A. and Doswell III, C. A.: An examination of jet
surface/hydrology model with the Penn State/NCAR MM5 mod- stream configurations, 500 mb vorticity advection patterns, and
eling system, Part II: Model validation, Mon. Weather Rev., 129, low-level thermal advection patterns during periods of intense
587–604, 2001b. convection, Mon. Weather Rev., 110, 184–197, 1982.
Colquhoun, J. R.: A decision tree method of forecasting thunder- Miller, R. C.: Notes on analysis and severe storm forecasting proce-
storms, severe thunderstorms and tornadoes, Weather Forecast., dures of the Air Force Global Weather Center, AWS Tech. Report
2, 337–345, 1987. 200 (Rev.), Headquarters Air Weather Service, Scott AFB, 106
Colquhoun, J. R.: A decision tree method of forecasting thun- pp., 1972.
derstorms, severe thunderstorms and tornadoes revisited, Proc. Mlawer, E. J., Taubman, S. J., Brown, P. D., Iacono, M. J., and
Fifth Australian Severe Thunderstorm Conf., Avoca Beach, New Clough, S. A.: Radiative transfer for inhomogeneous atmo-
South Wales, Australia, Australian Bureau of Meteorology, 135– spheres: RRTM, a validated k-correlated model for the long-
141, 1998. wave, J. Geophys. Res., 102, 16663–16682, 1997.
Doswell III, C. A., Lemon, L. R., and Maddox, R. A.: Forecaster Moore, J. T., Glass, F. H., Graves, C. E., Rochette, S. M., and
training – a review and analysis, Bull. Amer. Meteor. Soc., 62, Singer, M. J.: The Environment of Warm-Season Elevated
983–988, 1981. Thunderstorms Associated with Heavy Rainfall over the Central
Easterling, D. R.: Climatological patterns of thunderstorm activity United States, Weather Forecast., 18, 861–878, 2003.
in south-eastern USA, Int. J. Climatol., 11, 213–221, 2007. Ramis, C., Romero, R., and Homar, V.: The severe thunderstorm of
Fawbush, E. J. and Miller, R. C.: A method of forecasting hailstone 4 October 2007 in Mallorca: an observational study, Nat. Haz-
size at the earth’s surface, Bull. Amer. Meteor. Soc., 34, 235– ards Earth Syst. Sci., 9, 1237–1245, doi:10.5194/nhess-9-1237-
244, 1953. 2009, 2009.
Galway, J. G.: The lifted index as a predictor of latent instability, Showalter, A. K.: A stability index for thunderstorm forecasting,
Bull. Amer. Meteor. Soc., 37, 518–529, 1956. Bull. Amer. Meteor. Soc., 34, 240–252, 1953.
George, J. J.: Weather forecasting for aeronautics, Academic press, Whitman, C. D.: Mountain Meteorology, Oxford University Press,
673 pp., 1960. 2003.

www.nat-hazards-earth-syst-sci.net/12/403/2012/ Nat. Hazards Earth Syst. Sci., 12, 403–413, 2012