Tellus (2009), 61A, 250–263 C© 2009 The Authors
Journal compilation C© 2009 Blackwell Munksgaard
Printed in Singapore. All rights reserved
TELLUS
Convection forced by a descending dry layer
and low-level moist convergence
By ANDREW RUSSELL1∗, GERAINT VAUGHAN1, EM ILY G . NORTON1,
HUGO M . A . R ICKETTS1, CYRIL J. MORCRETTE2, TIM J. HEW ISON2,
KEITH . A . BROWNING3 and ALAN M . BLYTH3, 1Centre for Atmospheric Science, University of
Manchester, UK; 2Met Office, Exeter, UK; 3School of Earth and Environment, University of Leeds, UK
(Manuscript received 25 June 2008; in final form 12 November 2008)
ABSTRACT
A narrow line of convective showers was observed over southern England on 18 July 2005 during the Convective
Storm Initiation Project (CSIP). The showers formed behind a cold front (CF), beneath two apparently descending dry
layers (i.e. sloping so that they descended relative to the instruments observing them). The lowermost dry layer was
associated with a tropopause fold from a depression, which formed 2 d earlier from a breaking Rossby wave, located
northwest of the UK. The uppermost dry layer had fragmented from the original streamer due to rotation around the
depression (This rotation was also responsible for the observations of apparent descent—ascent would otherwise be
seen behind a CF). The lowermost dry layer descended over the UK and overran higher θw air beneath it, resulting in
potential instability. Combined with a surface convergence line (which triggered the convection but had less impact on
the convective available potential energy than the potential instability), convection was forced up to 5.5 km where the
uppermost dry layer capped it. The period when convection was possible was very short, thus explaining the narrowness
of the shower band. Convective Storm Initiation Project observations and model data are presented to illustrate the
unique processes in this case.
1. Introduction
1.1. Convective storms in the UK and CSIP
Convective storms are difficult to predict and can have severe
impacts. For example, the storm that caused the Boscastle flood,
which occurred in the UK during the summer of 2004, devel-
oped quickly and caused great damage and disruption (Golding
et al., 2005). The mechanisms that force the development of
such storms are well understood at the conceptual level (Bennett
et al., 2006). However, in practice, the observational network is
not dense enough to study the development of convection in de-
tail or to make use of this understanding and improve numerical
weather prediction (NWP) models using increasing model reso-
lution [This latter point is less relevant for mesoscale processes
that often develop within the domain of the highest resolution
models, such as surface convergence lines (e.g. Morcrette et al.,
2007), than for synoptic features that move into mesoscale do-
mains and this will be important in determining the focus of this
paper].
∗Corresponding author.
e-mail: andrew.russell-2@manchester.ac.uk
DOI: 10.1111/j.1600-0870.2008.00382.x
In response to the need to improve the representation of these
storms in numerical models, a greater understanding of the exact
conditions leading to their initiation is required. This knowledge
can be acquired by the execution of dedicated field campaigns
to observe these storms as they form. This paper focuses on data
collected from one case during such a campaign: the Convective
Storm Initiation Project (CSIP). The overall project is summa-
rized by Browning et al. (2007) and, to date, detailed analysis
from CSIP has shed light on the following subjects: secondary
initiation of convection forced by gravity waves (Marsham and
Parker, 2006; Morcrette et al., 2006); the varying ability of the
UK Met Office Unified Model (UM; Cullen, 1993) in predicting
the development of convective storms over the UK in different
cases (Clark and Lean, 2006); the initiation of an isolated storm
forced by a topographically induced convergence line and an
upper-level potential vorticity (PV) anomaly (Morcrette et al.,
2007); the role of cirrus shading in convective storm initiation
(Marsham et al., 2007a) and in boundary-layer development
(Marsham et al., 2007b) and how the origin of an atmospheric
lid was related to a breaking Rossby wave over the Atlantic
via the physical descent of upper-level air down the resultant
tropopause fold (Russell et al., 2008). In this paper, we will
describe a case where an apparently descending double layer
of stratospheric/upper-tropospheric air developed from a large
250 Tellus 61A (2009), 2
P U B L I S H E D B Y T H E I N T E R N A T I O N A L M E T E O R O L O G I C A L I N S T I T U T E I N S T O C K H O L M
SERIES A
DYNAMIC
METEOROLOGY
AND OCEANOGRAPHY

A CONVECTIVE STORM INITIATION PROJECT CASE STUDY 251
upper-level PV anomaly and show how it contributed to the ini-
tiation of convection over the UK. While neither of these layers
can strictly be called ‘tropopause folds’ as there is no evidence
of the dynamical tropopause (i.e. the 2 PVU contour) folding
back on itself, they are very similar in structure and identical in
origin to such folds.
It is important to understand this particular case as the UK
Met Office UM simulations of it failed to represent much of the
detail of the convective showers. This point is discussed further
in Section 7 of the paper but in the preceding sections, we will
show how the PV structure influencing the convection in this
case was particularly complex. As a result, the model struggled
to represent the more subtle aspects of this case, especially as
the main PV features developed outside the model’s mesoscale
domain. This is a crucial point with the goal of improving pre-
cipitation forecasts in mind. It should be noted at this point that
the convection in this case was triggered by a surface conver-
gence line but, as these features are very well understood in the
literature (Bennett et al., 2006; Morcrette et al., 2007, and ref-
erences therein) and are usually well modelled, including this
case, we will examine the upper-level features in much greater
detail in this paper.
1.2. Upper-level forcing of convection
Hoskins et al. (1985) reviewed the mechanisms by which upper-
level PV anomalies can reduce the convective stability of the
troposphere. Usually, of greatest relevance is how the upward
(downward) curvature of isentropes in the troposphere (strato-
sphere) associated with a moving PV anomaly is associated
with tropospheric ascent ahead of, and descent of air behind,
the depressed tropopause. It is this vertical displacement of the
isentropic surfaces that causes the reduction in static stability be-
neath the PV anomaly and, under certain conditions, can induce
convection (e.g. Griffiths et al., 2000; Morcrette et al., 2007).
Convection can also become more likely when upper-level
air descends into the troposphere: when low wet-bulb potential
temperature (θw) air from upper levels overruns higher θw air
nearer the surface, potential, or convective, instability is created.
Potential instability is important because an unsaturated atmo-
spheric column with θw decreasing with height will become
unstable if it is lifted to saturation. For example, Browning and
Roberts (1994, 1995) presented a case where an upper-level cold
front (UCF) of low θw ran ahead of the higher θw air in advance
of the surface cold front (SCF)—the region in between the two
features was characterized by potential instability and convec-
tive storms were observed in that region. In such cases, the UCF
often originates in the region behind the SCF where upper-level
(usually dry) air is drawn in from the tropopause depression as-
sociated with the low driving SCF. This flow of dry air is known
as the ‘dry intrusion’ (Browning, 1997) and it forms by the flow
of upper-level (stratospheric or upper tropospheric) air down
the disturbed isentropic surfaces of the tropopause fold associ-
ated with the PV anomaly (Danielsen, 1964, 1968). As a further
example of how such features influence convection, Vaughan
et al. (1994) have presented a case where the flow of air from
a tropopause depression started over the UK and extended into
a cut-off low (COL) over Spain. In that case, satellite images
showed arcs of convection to the west of Morocco, which was
at the leading edge of the descending and anticylonically curv-
ing flow initiated from the tropopause depression. Elsewhere in
that case, deep convection was observed in the eastern sector of
the COL (over eastern Spain) highlighting the impact of such
descending upper-level features on the wider region around the
tropopause depression. The investigation in the present paper is
particularly interesting as there are no obvious examples of the
role of multiple/concurrent stratospherically derived dry layers
in cases of the initiation of convection in the scientific literature.
The dynamics of such features can have a significant impact
on the development of convective storms and, therefore, require
much further understanding.
Morcrette et al. (2007) investigated another CSIP case
(IOP 1) where a small PV anomaly contributed to the triggering
of an isolated convective storm. They also showed how the UM
forecasts of that case were very good. Russell et al. (2008) also
investigated CSIP IOP 1 but from the perspective of a dry layer
of relatively high convective inhibition (CIN) located beneath
the small PV anomaly. This dry layer was found to have origi-
nated (3 d before the IOP) from the same breaking Rossby wave
which gave rise to the small PV anomaly. The case investigated
in the present paper (CSIP IOP 9) was very different to IOP 1.
First, the PV anomaly was much larger in IOP 9 and the main
body of it was not over the UK. Instead, the upper-level influ-
ence on the convection in the present investigation arose from
a pair of dry layers associated with the tropopause fold of the
main PV anomaly and the original streamer. The second sig-
nificant difference between the two cases was the poor predic-
tion/representation of the storms by the UM during IOP 9. We
argue that the complexity and fragmentation of the tropospheric
features, together with the precise alignment of the layers with
the boundary-layer convergence lines, was responsible for the
observed convection. Accurate forecasting of convection in this
case would therefore require the model to represent both these
processes accurately.
1.3. Aims and paper structure
The aim of this study is to identify the role of an apparently
descending dry feature in relation to the initiation and intensi-
fication of a band of convective precipitation on 18 July 2005.
Figure 1 shows where this precipitation broke out: there were
three convective arcs—two over eastern England and one over
southern England. This work will concentrate on the latter fea-
ture, which occurred in the CSIP experimental area, which was
centred on Chilbolton, Hampshire (51.2◦ N, 1.4◦ W). The ori-
gins of the dry feature in question will also be investigated to
Tellus 61A (2009), 2

252 A. RUSSELL ET AL.
Fig. 1. The convective showers being investigated in this paper as
represented by (a) the rainfall radar network, showing the precipitation
rate at 1615 UTC on 18 July 2005 and (b) high-resolution visible
imagery from the Meteosat-8 satellite at 1615 UTC showing the cloud
bands relating to the rainfall shown in Fig. 1a. The locations of places
referred to in the text are also shown in Fig. 1a.
place this event in a wider meteorological context and to help
understand similar events that occur in other cases.
The paper is structured as follows. Section 2 introduces the
data sources that will be used to meet the aims of this investiga-
tion. A synoptic and low-level analysis of the event in question
will be presented in Section 3. Section 4 presents specific ob-
servations of the apparently descending dry feature and explains
why we are referring to them as ‘apparently’ descending. Sec-
tion 5 identifies the role that the apparently descending dry lay-
ers played in forcing the convection and Section 6 examines the
way in which this feature arrived over the UK. The importance
of these findings will be discussed in Section 7 and conclusions
will be drawn in Section 8.
2. Data
The event investigated here occurred on 18 July 2005 dur-
ing CSIP. Observations were taken around Chilbolton (see
Fig. 1a for the location of places referred to in the text) by a
variety of instruments. For more information on the motivation
behind CSIP, a more detailed description of the CSIP obser-
vational network and brief examinations of some of the major
cases of interest, the reader is referred to the project summary by
Browning et al. (2007). Here, we will specifically present data
from: CSIP and UK Met Office radiosondes; CSIP automatic
weather stations (AWS); the UK Mesosphere Stratosphere Tro-
posphere (MST) radar (Vaughan, 2002), which is located near
Aberystwyth, Wales (52.4◦ N, 4.0◦ W); European Centre for
Medium Range Weather Forecasting (ECMWF) operational
analyses;UKMetOffice surface synoptic charts; aGPS station; a
Radiometrics TP/WVP-3000microwave radiometer (Ware et al.,
2003), which allows retrievals of temperature and humidity pro-
fileswith coarse vertical resolution aswell as the integratedwater
vapour (WV) and liquid water path; the Total Ozone Mapping
Spectrometer (TOMS; Heath et al., 1975); the Meteosat Second
Generation (MSG) satellite, specifically Meteosat-8 (Schmetz
et al., 2002); the 3 GHz (S-band) Chilbolton Advanced Mete-
orological Radar, which is known as CAMRa (Goddard et al.,
1994); the Universities’ Facility for Atmospheric Measurements
(UFAM) Aerosol and Ozone LIDAR and the UFAM UHF wind
profiling radar (Norton et al., 2006).
The latter of these instruments – the UHF wind profiling
radar – operates at 1290 MHz (23 cm wavelength) with three
beams: one pointing to the zenith and two at 17.5◦ off-vertical.
Echoes are obtained from refractive index inhomogenieties in
clear air and from raindrops. The turbulent convective boundary
layer is full of structure in refractive index and therefore gives
a strong echo; layers with sharp gradients in absolute humidity
and/or potential temperature (such as are found in atmospheric
lids) also show up as layers of enhanced echo power because
of the fractal nature (i.e. a cascade of scales) of the gradients.
Although the radar only detects sharp layers on a scale of 3 m or
less, high-resolution observations show that the stable layers and
sharp humidity gradients conventionally revealed by radiosonde
profiles (typically hundreds of metres deep) are built up from a
myriad of much smaller steps in temperature or humidity (Gage
and Green, 1981; Muschinski and Wode, 1998) which together
produce the radar echo.
The UK MST radar is a VHF radar, operating at a frequency
of 46.5 MHz (wavelength 6.41 m), and is able to observe higher
in the atmosphere than the UFAM wind profiler. It is also much
less sensitive to precipitation: echo power is due almost entirely
to quasi-specular reflection from gradients in static stability and
specific humidity (Gage and Green, 1981).
The combination of all these data sets will illustrate the
development of a breaking Rossby wave over the Atlantic,
the resultant COL that moved towards the UK and the as-
sociated apparently descending double ‘arm’ of relatively
high PV, dry air that helped initiate the convection over the
UK.
Tellus 61A (2009), 2

A CONVECTIVE STORM INITIATION PROJECT CASE STUDY 253
Fig. 2. Summary of the conditions over the UK at 1200 UTC on 18 July 2005: (a) high-resolution visible image from Meteosat-8 showing the cloud
bands associated with the two cold fronts shown in Fig. 2c; (b) contrast enhanced water vapour (WV) image from Meteosat-8 showing the dry
regions relating to the PV anomaly located to the northwest of the UK and the dry ‘arm’ over the southern UK; (c) ECMWF operational analysis of
potential vorticity (PV) in PV units (PVU; i.e. 1.0 × 10−6 m2 s−1 K kg−1) on the 315 K isentropic surface (see Fig. 11a for the height of this
surface) with the UK Met Office frontal analysis for 1200 UTC superimposed and (d) total ozone (in Dobson Units; DU) at ∼1130 UTC from
TOMS, contours are every 20 DU. The TOMS data are indicative of the tropopause depression associated with the PV anomaly as the total ozone
column detected by the spectrometer is enhanced by the intrusion of stratospheric, high-ozone air to lower levels than usual as the PV anomaly
results in the stretching of the vortex into the upper troposphere. Darker shading indicates higher levels of PV and O3 for Figs. 2c and d,
respectively. The dashed line plotted in Fig. 2c represents the line through which the vertical cross-sections presented in Figs. 10 and 11a are plotted.
3. Synoptic and low-level background
Figure 2 shows the synoptic situation for the UK area at 1200
UTC on 18 July 2005. The precipitation event of interest for this
investigation was related to the rearmost of two approximately
north-south orientated bands of cloud, parts of which were over
the UK, as seen in Fig. 2a. In particular, the two sharp lines of
precipitation (maximum rain rate of over 32 mm h−1 at 1645
UTC) that can be seen in Fig. 1a coincide with the location of
a band of rope cloud seen in the westernmost cloud band in
Fig. 2a. However, the convective elements of this band at 1200
UTC, were very shallow—they were being suppressed by the
dry air above. This can be inferred from Fig. 2b: there was a
swathe of dry air over Wales and some of southern England
that was covering the area in which the shallow cumulus was
observed in Fig. 2a. This event is similar to that observed by
Browning and Roberts (1994), who identified cumulonimbus
clouds breaking out at the northern end of rope clouds over the
Bay of Biscay under synoptic conditions similar to this case.
Indeed, the relatively deep cumulus clouds in our case can be
seen in a satellite image from later in the day (Fig. 1b) and
the convection was observed by CAMRa to reach as high as
5.5 km (Fig. 3). Figure 2a also shows the much broader band
of cloud associated with the leading SCF, which delivered the
light precipitation seen to the east of the narrow lines of intense
precipitation in Fig. 1a.
Figure 2b shows the north-south orientated moist (bright)
band associated with the leading SCF and, more interestingly,
the dry (dark) region over Wales and southern England behind
the moist band. As will be seen in Section 6, this feature was
rotating eastwards (anticlockwise) around the southern flank of
the main ‘dark eye’ (Roberts, 2000) to the northwest of the
Tellus 61A (2009), 2

254 A. RUSSELL ET AL.
Fig. 3. Range-height indicator (RHI) of reflectivity (dBZ) from the
3 GHz radar (CAMRa) located at Chilbolton (51.2◦ N, 1.4◦ W). This
RHI was taken at 1459 UTC and was at an azimuth of 270◦ to the north.
UK. The ‘dark eye’ is the most obvious feature of Fig. 2b and,
by comparison with Figs. 2c and d, was clearly related to the
tropopause depression (centred on 58◦ N, 11◦ W) associatedwith
the cyclone driving the atmospheric circulation in this region.
Figure 2c also shows an ‘arm’ of relatively high PV (centred
about 50◦ N), which Fig. 2b showed was also dry and moving
behind the split front (Browning and Monk, 1982) structure that
was evident from Fig. 2a. However, as we will discuss in the
remainder of the paper, the westernmost of the SCFs plotted
in Fig. 2c was related more to the influence of the upper-level
features being investigated here rather than any surface features.
Finally, Fig. 2d shows that the main PV anomaly and the ‘arm’
to the south were coincident with high total ozonemeasurements
and, therefore, correspond to depressions of the tropopause.
Similar to the CSIP case investigated by Morcrette et al.
(2007), the convective showers were initiated by a convergence
Fig. 4. Ten metre winds (scaled arrows) and
convergence (shading) from NIMROD, that
is, the Met Office nowcasting model
(Golding, 1998). The location of Chilbolton
is marked by a blue cross.
line that was over the southern UK. Evidence for one such con-
vergence line is shown in Fig. 4. Furthermore, data from the array
of CSIP AWS and the UFAM wind profiler (neither shown) con-
firm the view shown in Fig. 4. The wind over the CSIP region
was first influenced by the flow around the southern edge of the
peninsula, which moved over the CSIP region as southwester-
lies. This regime was followed by westerlies deflected to the
north of the peninsula—the line where these two flows met rep-
resented the convergence line, which moved over the CSIP area
at around 1600 UTC. This is typical of how convergence lines
form in this region, that is, downwind of the Devon and Cornwall
coasts because of these three methods: the varying effect of land
and sea on frictional forces (Hunt et al., 2004); surface temper-
atures differences between the land and sea (Simpson, 1997)
or orographic lifting (Bader et al., 1995). The convergence line
provided the boundary layer with the uplift required to over-
come the CIN that was present at low levels. Our focus in this
paper, however, is not on the boundary-layer forcing but on the
much less well-understood dynamics responsible for generating
potential instability and CIN, and it is to this that we now turn.
4. Observations of the descending dry layers
The observations presented in this case study were taken mostly
from stationary, vertically pointing instruments deployed in the
UK during the CSIP field campaign. In this section, we will
discuss the atmospheric features observed from this perspective.
Of particular interest in these observations was a pair of dry
layers in the troposphere that descended during the day. These
Tellus 61A (2009), 2

A CONVECTIVE STORM INITIATION PROJECT CASE STUDY 255
Fig. 5. The main line-contour plot shows: wet-bulb potential temperature (θw ; solid lines; contour interval is 1 K); the <10% relative humidity (RH)
area (grey shading, where there is no overlap with the radar data, darker = lower RH); the 10% RH contour (red lines) and the lifting condensation
level (LCL; thick black dashed line) all from the Swanage soundings for 18 July 2005. The vertical lines at the top of the profile indicate the
radiosonde release times and the contours have been interpolated from this data. The colour plot between 995 and 630 hPa shows signal-to-noise
ratio recorded, in dBZ, from the UFAM wind profiling radar that was located at Linkenholt. The smaller top plot shows atmospheric WV derived
from a GPS station at Linkenholt. The text in the box at the bottom of the figure describes the conditions as recorded by a web-camera located by the
wind profiler at Linkenholt (51.3◦ N, 1.5◦ W).
layers were observed by several of the CSIP instruments. In the
context of the observations, ‘descending’ means that there was
a consistent dry layer above the instrument, the height of which
decreased with time. This decrease in height occurred because
the atmospheric features rotated anticlockwise and advected hor-
izontally from west to east over the instrument and because of
their particular orientation/slope in the atmosphere. In this pa-
per, we will refer to this type of descent as ‘apparent descent’
or simply refer to the layer (i.e. not necessarily the air within
that layer) descending or sloping. This is in contrast to ‘physical
descent’, which will be used to refer to the vertical motion of
a distinct air parcel and has thermodynamic consequences. At
this stage of the study, we are not stating that the air within this
layer had previously descended from upper levels but that the
layer as a whole was descending during the day relative to the
instrument observing it, that is, it is the slope of the layer as a
whole that is of interest.
To give an example of some of these observations, we can turn
to the radiosondes launched from Swanage (50.4◦ N, 1.6◦ W),
which show the feature particularly well (Fig. 5). By examining
the red contours – denoting the 10% relative humidity (RH)
isoline – it can be seen that therewere two distinct dry layerswith
bases at around 650 and 400 hPa; this double layer structure of
the dry region was not evident from Fig. 2. As the two dry layers
descended during the day, Fig. 5 (top) shows that the atmospheric
column WV measurement also decreased, consistent with dry
stratospheric air having intruded into the troposphere.
Figure 5 also shows data from the UFAM UHF radar that was
located at Linkenholt (some 85 km to the northeast of Swanage).
Despite the relatively large distance between Linkenholt and
Swanage, the orientation of the leading edge of the descending
layer (Figs. 1 and 2c) means that it moved over the two locations
at approximately the same time. Indeed, the radar detects the
moisture gradient at the base of the feature as it moved into
its range at around 1615 UTC, coincident with the gradient
measured by the Swanage radiosondes. The radar signal-to-noise
ratio (SNR) was also strongly influenced by the precipitation
that started at around 1545 UTC. The presence of precipitation
is confirmed by the negative vertical wind speeds recorded by
the radar (not shown) and the relevant area of the SNR plot
appears white in Fig. 5. Furthermore, the radiosondes also show
a lid of dry, high θw air at around 850 hPa between 1430 and
1530 UTC, which appears to have had a role in inhibiting the
convection until after this time. As will be shown in Fig. 9, this
lid was weaker in the region where the convection broke out
and, therefore, partially explains why showers were not seen
at Swanage. The upper-level features, though, were much more
homogeneous over a wider region and, therefore, the Swanage
data provide a good illustration of the upper-level features over
a wider area.
Tellus 61A (2009), 2

256 A. RUSSELL ET AL.
Fig. 6. Data from the MST radar (located at Aberystwyth) for 18 July
2005 showing the period 0300–2100 UTC for (a) radar return signal
power (dBZ) below 10 km and (b) vertical shear (ms−1 km−1) below
7 km. Vertical shear is equal to du/dz calculated between radar gates
(300 m). The vertical extent of the vertical shear plot has been limited
to 7 km as there are very high values in the stratosphere that are not
relevant to our investigation. The dashed black and white lines indicate
the height of the bases of the two dry layers as identified by the
Swanage radiosondes (Fig. 5) with the distance between Aberystwyth
and Swanage accounted for.
Figure 6 presents observations from the MST radar and pro-
vides a second view of the sloping dry layers over a longer
period and up to greater altitudes. The lowermost dry layer
can be seen descending from around 8 km (5 UTC) to 3 km
(12 UTC) in the MST power return signal (Fig. 6a) and the
lowest region of this layer also has relatively high vertical shear
(Fig. 6b). These heights are consistent with Fig. 5 once the dis-
tance between Aberystwyth and Linkenholt has been accounted
for; the heights of the base of these two layers asmeasured by the
radiosondes have been plotted over the MST data but it should
be noted that these layers can be traced beyond the radiosonde
measurements in the MST data. The uppermost dry layer is less
clearly defined in the power signal but appears to start from ap-
proximately the same point as the lower layer and only descends
to 4 km by 1530 UTC. The region of low return signal power
between 0700 and 0800 UTC and 2–3 km is an indicator of the
frontal rain that passed over Aberystwyth (see Fig. 1a) at this
time and then moved on to the CSIP area.
The impact of the lowermost dry layer can also be seen in
observations made by the UFAM Ozone and Aerosol LIDAR
(located at Chilbolton) as the layer reached its lowest point –
down to as lowas 1.8 km– after around 1800UTC (Fig. 7),which
is consistent with radiosondes launched from Chilbolton at this
time (not shown). The low aerosol backscatter coefficient region
above around 1.75 km shows where the apparently descending
dry layer met the high aerosol backscatter coefficient region,
Fig. 7. Aerosol backscatter coefficient (β) from the 355 nm beam of
the UFAM Ozone and Aerosol LIDAR. The darkest reds correspond to
areas of cloud whereas the light blue, green and yellow areas show
regions of aerosol in the boundary layer. The dark blue and white
regions show the clean air that was apparently descending overhead
during the day and suppressing the boundary-layer growth.
Fig. 8. Vertical RH (%) profile from a UK Met Office Radiometrics
TP/WVP-3000 microwave radiometer that was located at Linkenholt.
The dashed black and white line indicates the height of the base of the
lowermost dry layer as identified by the Swanage radiosondes (Fig. 5).
that is, the boundary layer. There was no signal from the LIDAR
before this time due to the extensive cloud cover at Chilbolton
but even this observation from later in the day helps to identify
the role of the feature.
Figure 8 also shows the apparent descent of dry air over
the CSIP region, this time from radiometer data. The passing
of the storm and the associated precipitation can also be seen
as the vertical peak in RH at around 1600 UTC. What is also
interesting here is the reduced depth of the surface moist layer
beneath the descending dry air. This will be discussed further in
the next section as it was another key factor in the development
of the convection. This reduction had the important effect of
amplifying the changes in surface moisture, which altered from
dry (0900–1300 UTC) to moist (1300–1800 UTC). This change
Tellus 61A (2009), 2

A CONVECTIVE STORM INITIATION PROJECT CASE STUDY 257
Fig. 9. Dry-bulb temperature (T; solid line) and dew-point temperature
(T d; dot–dashed line) measured by radiosondes launched from (a)
Larkhill at 1158 UTC and (b) Reading at 1605 UTC. The lifting
condensation level (LCL) has been plotted onto these soundings (i.e.
where the solid grey lines starting from the surface T and T d meet
was also observed by the Advanced Clear-air Radar for Observ-
ing the Boundary layer And Troposphere (ACROBAT; data not
shown), which is a UHF radar mounted on the same Chilbolton
dish as CAMRa.
All the observations of the dry layers presented in this sec-
tion show them apparently descending. This is not what would
be expected from the synoptic data presented in Fig. 2, which
showed a cold front moving from west to east over the UK.
This would normally be observed as a layer ascending with time
over a fixed point; descent with time normally characterizes a
warm front (Bjerknes and Solberg, 1922). The explanation for
this contradiction lies in the way that the rotation around the
depression to the northwest of the UK altered how the CSIP in-
struments observed the features. While this will be investigated
further in Section 6, it has been noted here to avoid confusion
that could otherwise arise from these observations.
In this section, we have identified the dry feature in ques-
tion, shown that it had two distinct layers emanating from a
tropopause depression and that the origin/nature of the feature
requires some more investigation to understand it fully. It is now
necessary to analyse the impact of these apparently descending
layers on the convective stability of the troposphere during this
case.
5. Convective stability of the troposphere
The potential temperature (θ ) profiles from the Swanage sound-
ings (not shown) showed no evidence of a depressed tropopause
or an upward curvature of the isentropes in the troposphere. This
was typical of all the locations where radiosondes were launched
during this case [i.e. Reading (51.4◦ N, 1.0◦ W),Chilbolton, Bath
(51.4◦ N, 2.4◦ W) and Larkhill (51.2◦ N, 1.8◦ W)]. Therefore,
it can be assumed that the method by which upper-level PV
anomalies often influence convection, as discussed in the Sec-
tion 1, was not at work in this case. Indeed, the main body of
the PV anomaly (Fig. 2c) was too far north to influence the tro-
pospheric stability in this way over southern UK—this case was
driven more by what occurred at the fringes of the PV anomaly.
Figure 9, however, shows how the change in the surface and
mid-tropospheric conditions above Larkhill and Reading be-
tween 1200 and 1600 UTC helped create the conditions that led
to the convective showers. The relative dryness in the surface
when, respectively, the dry adiabat and the line of constant mixing ratio
are followed upwards) and, in the case of Fig. 9b, the vertical parcel
trajectory (dashed grey line) has been continued from the LCL up the
moist adiabat to show that the surface parcel would have reached about
450 hPa. It should also be noted that as a result of the irregular
radiosonde launch times during this IOP, we are not able to illustrate
the two phases of this case using soundings from the same location.
However, Larkhill and Reading are relatively close (see Fig. 1) and
would be influenced by the same surface and upper-level conditions
due to the eastward flow of the major features being investigated here.
Tellus 61A (2009), 2

258 A. RUSSELL ET AL.
layer at 1200 UTC (Fig. 9a) resulted in the lifted surface parcel
being limited by the base of a strong inversion, or lid, at around
850 hPa. However, by 1600 UTC at Reading the surface layer
had moistened due to the moisture convergence forced by the
low-level flow examined in Fig. 4, which was initially capped
by the apparent descent of dry air from above (Fig. 8). The
dry layer had also reached Reading by 1600 UTC and Fig. 5
shows a decrease in θw with height above 850 hPa (i.e. potential
instability—dry, low θw air above moist, high θw air) coincident
with the arrival of the leading edge of the dry layer. This, as will
be shown later in this section, increased the convective available
potential energy (CAPE) in the mid-troposphere.
However, and perhaps more importantly, Fig. 9b shows that
the temperature profile between about 925 and 450 hPa mostly
stays to the left-hand side of the moist adiabat associated with
the lifted surface parcel, indicating that it is unstable (i.e. con-
vective instability). The fact that the lifted surface parcel does
not clear the lid at 650 hPa highlights the intricate interplay of
the features forcing the convection during this case and that the
window of opportunity for convection to develop was short (The
aforementioned lid at 650 hPa can be seen in Fig. 3 as the ap-
proximately horizontal layer at around 3 km extending from the
convective plumes).
The key point is that the convection could only develop at
the point where the surface moisture convergence coincided
with weak CIN combined with potential instability—this is very
nearly the case in Fig. 9b. We interpret this finding as follows:
at the leading edge of a dry layer the air is of upper-tropospheric
origin with low RH but not much static stability. In the main
body of the layer, air from the tropopause region is found, whose
elevated PV manifests itself as enhanced static stability rather
than absolute vorticity due to the vertical extent of the anomaly.
This increase in static stability provides CIN beneath the main
body of the descending layer; the cooling of the boundary layer
also reduces the mid-tropospheric CIN. There was only a short
period of time when all these conditions were met and, thus, the
band of showers was narrow.
The lowermost descending dry layer appears to have played
a further role in initiating the convection by introducing colder
air directly below it, which resulted in the weakening of the
lid that was previously seen at 725 hPa (Fig. 9a). Had this lid
persisted until the 1600 UTC Reading sounding (Fig. 9b) it
would have very likely capped the convection at that height.
Hill and Browning (1987) have previously described how this
cooling mechanism affects atmospheric stability.
These profiles (Fig. 9) can also be used to assess the rel-
ative importance of the surface changes (i.e. moistening and
cooling) and mid-tropospheric changes (i.e. the descent of the
dry layers leading to the potential instability) throughout the
day. To do this, the changes in CAPE from the Larkhill 1158
UTC to the Reading 1605 UTC sounding have been attributed
to one or other of the two processes. Convective available poten-
tial energy is calculated using the method described by Emanuel
(1994). First, the surface parcel of the Larkhill 1158UTC sound-
ing was moistened and cooled to the level of the Reading 1605
UTC sounding. Calculating the CAPE of this new surface par-
cel increased the CAPE of the Larkhill 1158 UTC profile by
117 J kg−1. This essentially quantifies the impact on CAPE that
the moistening and cooling that occurred between the two pro-
files had. TheCAPEvalue of the surface parcel of the unmodified
Reading 1605 UTC sounding was 325 J kg−1. The difference be-
tween this value and the CAPE of the modified Larkhill 1158
UTC is 153 J kg−1. This value represents the impact on CAPE
that the potential instability had as it is the only significant dif-
ference between the modified Larkhill 1158 UTC profile and the
unmodified Reading 1605 UTC sounding. From this analysis, it
can be stated that the changes in the upper and mid-levels during
the day had the greater impact on CAPE than the boundary-layer
changes (117 vs. 153 J kg−1). However, it must not be forgotten
that it was the convergence line that triggered the convection so
both were essential in the development of the storms.
As discussed in the previous section, in relation to Fig. 8, it can
also be seen that the apparent descent of dry stable air to such low
levels played a role in amplifying the impact of the convergence
line; in limiting the vertical extent of the boundary layer, the
apparently descending dry air allowed the RH to increase to such
an extent as to promote the likelihood of deep convection. This
increase in low-level RH can be inferred from the radiosonde
ascents in Fig. 9 and seen directly in the radiometer data in
Fig. 8, which also shows the lower portion of the dry layer (with
the same shape as seen in Fig. 5 from 6 km at 1200 UTC to 4 km
at 1500 UTC) limiting the vertical extent of the moist region.
There was also a dry region near the surface which can be seen
in Fig. 8 between 0900 and 1300 UTC and up to 1300 UTC
in Fig. 5—low radar echo power signifies dry air. This initial
dryness would have delayed the onset of convection.
Figure 9b also indicates that, once the lids had been breached,
convection was free to reach around 450 hPa, which was around
the same height that the convection was seen to reach by the
Chilbolton radar (Fig. 3). Interestingly, Fig. 9b also shows that it
was the uppermost dry layer that was responsible for halting the
convection at 450 hPa, meaning that the dry feature, as a whole,
contributed both to the convective initiation and, subsequently,
to limiting its vertical extent. This is a similar role to that played
by the resultant parts of a breaking Rossby wave investigated
by Russell et al. (2008) from CSIP IOP 1 (15 June 2005). In
that case, however, the lid was located beneath an upper-level
PV anomaly (not between the tropopause and a lower level PV
anomaly as seen here), and was responsible for the widespread
CIN seen in that case.
To extend the CSIP observations and place them in context,
we now present diagnostics from the ECMWF operational anal-
yses for this day. Figure 10 shows a cross-section of RH and
temperature along the dashed line shown in Fig. 2c, NW-SE
from Iceland to Brest. This, like the observations, shows a moist
boundary layer beneath a sloping dry region (which consists of
Tellus 61A (2009), 2

A CONVECTIVE STORM INITIATION PROJECT CASE STUDY 259
Fig. 10. Relative humidity (RH; shading) and temperature (◦ C; white
contours with an interval of 4◦ C) from the ECMWF operational
analyses at 1200 UTC on 18 July 2005. This plot is a cross-section
through the dashed line seen in Fig. 2c, that is, from 20◦W, 64◦N to
2◦W, 46◦N.
two dry layers, see Fig. 5) southeastward of 10◦W. Figure 11a
shows the corresponding PV cross-section, seemingly identify-
ing the dry layers as extrusions of tropopause level air from
the tropopause depression centred around 12◦W. In Section 6,
we will show that this view is too simplistic but it is useful to
continue with this interpretation at this point while noting the
complexity of the PV anomaly.
Browning (1997) showed how the driest region of an intrusion
coincides with the region of maximum overrunning; this is also
the case here. Similarly, we can see, by considering both Figs. 10
and 11a, that the sloping region of high PV (if viewed as a single
feature rather than two distinct layers) is slicing into a region
of relatively high RH, that is, the moist surface layer beneath
the intrusion and the relatively high RH column at around 5◦W
that reaches up to around 200 hPa. This also agrees with the
overrunning θw surfaces mechanism summarized by Browning
(1997) and we can, therefore, be more confident in one of our
hypotheses regarding the mechanisms by which the convection
was induced in the present case.
In itself the reporting of two sloping dry layers of this sort is
rare, especially as they were observed, to some extent, by the
radiosondes, MST radar, the UFAM wind profiling radar and
the UFAM LIDAR. The features were also well captured by the
ECMWF operational model which is in good agreement with
our observations. These layers were also partly responsible for
inducing deep convection and then inhibiting the convection at
a higher level, which makes this a very interesting case. How
these features developed in the wider synoptic context remains
to be seen. It is to this that we now turn.
6. Development of the cut-off low
and the sloping dry layers
So far in this study we have concentrated on the role of the
sloping dry layers in forcing the convection observed on 18 July
2005. To understand the importance of this event, we need to
Fig. 11. Potential vorticity (in PVU; solid contours and shading;
contour interval is 0.25 PVU below 2 PVU, and 1 PVU above 2 PVU;
and darker shading relates to higher PV) and potential temperature (θ ;
dashed contours; contour interval is 15 K) from the ECMWF
operational analyses at 1200 UTC for Fig. 11a and 0600 UTC for Fig.
11b on 18 July 2005. These data are plotted as cross-sections through
(a) the dashed line seen in Fig. 2c, that is, from 20◦W, 64◦N to 2◦W,
46◦N and (b) along the curved path shown by the circular points
plotted in Fig. 12f.
understand how the situation arose. With this in mind, Fig. 12
shows a PV perspective of the build up to the convective showers
over the UK. Similar to the situation from CSIP IOP 1 (Russell
et al., 2008), the event was preceded by a breaking Rossby wave
[second life-cycle category (LC2) of Thorncroft et al., 1993]
over the Atlantic in the present case. This case differs from IOP
1 in that the wave broke much further north over the Atlantic. By
0000UTC on 17 July 2005, a cut-off low had formed at around
60◦N, 25◦Wandwas rotating and slowlymoving southeastwards
towards the UK. As a fold started to develop in the baroclinic
region to the west of the PV anomaly, upper-level air was drawn
around the fringes of the main body of the feature (best seen in
Figs. 12 and 2c). Subsequently, this flow apparently descended
over the UK due to the influence of the PV anomaly on the
isentropes (see Fig. 11) and was also drawn eastwards as a result
of the cyclonic rotation of the PV anomaly.
Tellus 61A (2009), 2

260 A. RUSSELL ET AL.
Fig. 12. A selection of ECMWF operational analyses of PV (thick contours and shading) on the 315 K isentropic surface for the build up to the
descending layers of upper-level air moving over the UK on the 18 July 2005. The contour interval is 1 PVU (i.e. 1.0 × 10−6 m2 s−1 K kg−1) and
darker regions of shading show areas of higher PV. The ECMWF operational analyses mean sea level pressure (MSLP) is also plotted here (dotted
contours; contour interval is 5 hPa; dash–dotted contour, for reference, is the 1020 hPa isobar). The positions of relevant fronts (there were other
fronts present for each time shown) from the Met Office surface analyses are plotted as well. The dots plotted in Fig. 12f relate to the gridpoints used
to plot the vertical PV cross-section shown in Fig. 11b. Point number (as used on the x-axis of Fig. 11d) increases from 0 in the northwest to 36 in
the southeast. Points 10, 20 and 30 can be identified by being bigger than the other points and they have a cross plotted on them.
Figure 11 gives a vertical perspective of the PV anomaly.
While Fig. 11a shows the two layers verywell, it does not explain
why they were observed to slope in the way that they were.
How did this occur? In short, it was the rotation around the PV
anomaly that was critical to the development and morphology
of the two dry layers. By plotting a further vertical cross-section
of PV around the depression (Fig. 11b; see Fig. 12f for the path
of this cross-section), we can see how regions of high PV were
pulled anticlockwise (cyclonically) around the depression from
what remained of the original streamer to the west of the COL.
This caused the streamer to fragment and where these fragments
reached the eastern side of the depression (where there was a
small fold, which was observed as the lowermost dry layer) they
were observed as the higher of the dry layers. Examples of these
fragments can be identified in Fig. 11b at 700 hPa between points
17 and 20 and, more significantly, at 400 hPa between points 20
and 30. The latter of these eventually formed the uppermost dry
layer.
As well as the twisting, the rotation also partially explains
why the frontal surfaces were observed to descend with time in
the CSIP area, when a cold frontal surface observed at a fixed
location would normally ascend with time as a weather system
moved across that point (Bjerknes and Solberg, 1922). The fact
that the structure as a whole sloped in the north-south direction
– it was higher in the north than the south – provides the rest of
the explanation. Initially, the MST radar and radiosondes would
have been on the southeastern fringe of the folded structure as-
sociated with the PV anomaly, where the fold was at a greater
elevation than at the southern edge. As the day progressed, the
PV anomaly rotated cyclonically, which moved the lower re-
gions of the fold (and the fragment) over the southern UK. This
rotation gave the impression in the observations that the layers
were descending. However, the layers were ascending radially
towards the centre of the PV anomaly, as Fig. 11a shows. If
the feature were observed along its radius, as in Fig. 11a, then
apparent ascent would have been observed. This explanation is
Tellus 61A (2009), 2

A CONVECTIVE STORM INITIATION PROJECT CASE STUDY 261
Fig. 13. Schematic of a PV isosurface (approximately 1 PVU) based on the evidence presented in this paper. The schematic represents the
tropospheric PV structure at approximately 0000 UTC on 18 July 2005—a sketch of mainland UK has been included to indicate the scale and
location the PV isosurface. Elevated features are shaded dark grey and the ground is shaded light grey. The sloping round/annular shaped feature in
the centre of the drawing shows the downward flow in the upper troposphere from the base of the main PV anomaly. The base of the round feature
was higher in the north than in the south; this is indicated by the dotted lines (vertical and horizontal) that show how the features would translate
onto the ground. The arrows indicate that the round feature was moving eastwards and rotating anticlockwise but, because of the north to south
slope, it was precessing around a moving axis rather than around a fixed vertical axis. Under these circumstances, observations taken from a fixed
point beneath the round feature would show that it’s base reduced in height over time. The points A and B plotted in Fig. 13 represent two such
observations taken in the frame of reference stationary with respect to the rotating feature; this shows how an observer would see the base coming
down in height, as was seen in this case. The streamer to the west of the main PV ‘skirt’ has been plotted to show the source of the uppermost PV
fragment that was observed during the case.
relatively complicated, so we have simplified this description in
a schematic sketch in Fig. 13 that summarizes this mechanism.
7. Discussion and representation of convection
in the Met Office mesoscale Unified Model
As introduced in Section 1, the UK Met Office UM was run for
this case at various resolutions, all of which fell short in captur-
ing the details of the showers. As well as the then operational
12 km resolution mesoscale model, the UM was also run quasi-
operationally at 4 km resolution and in a post-event trial mode
at 1 km resolution in a limited domain over the CSIP area (see
Clark and Lean, 2006, for more details). While the UM at these
various resolutions captures the intensity and orientation of the
frontal precipitation quite well, none of these models adequately
capture the extent, consistency or orientation of the line of show-
ers behind the front, as seen in fig. 50 of Clark and Lean (2006).
The differences between these precipitation forecasts and the
observations can be summarized thus:
(1) The angle of the line appears good in the 12 km resolution
model but the intensity is too low and neither of the regions of
precipitation associated with the front or the line of showers are
continuous enough.
(2) The 4 km model produces a very intense oval of precip-
itation approximately perpendicular (rather than parallel) to the
front, which is also represented as more intense than the rainfall
radar observed.
(3) Finally, the 1 km model produces the most continuous
line of showers when compared with the other two runs but it is
too intense and at the wrong angle when compared to the radar
observations.
This implies that the UM has not represented the complex
structure of the angle of the leading edge of the dry intrusion
adequately or its interaction with the surface convergence line.
The overestimation of the intensity in the model was due to poor
representation of the uppermost dry layer in the model (Peter
Clark, personal communication, 2005). We have shown in Fig. 9
that this layer was ultimately responsible for capping the convec-
tion in this case but, in the model, it was not dry or warm enough
relative to its environment to cap the convection at around
450 hPa as observed.
As far as we are aware, they are no other observations of dou-
ble apparently descending layers in themeteorological literature.
From this analysis, it appears to be derived from a PV anomaly
with a strong cyclonic rotation. The climatological occurrence
of such events is unknown and requires further investigation.
This would be a potentially significant study to undertake as the
simultaneous passing of two dry layers influences the intensity
of the convection observed and, as discussed above, is not well
represented by mesoscale models such as the UM.
More generally, this case highlights the importance of sloping
layers in both promoting convection via the reduction in tropo-
spheric stability and the weakening of lids as well as their role
in capping convection once it has been initiated.
8. Conclusions
On 18 July 2005, two sloping dry layers passed over the UK at
around 800 and 500 hPa while, nearer the surface, a narrow band
Tellus 61A (2009), 2

262 A. RUSSELL ET AL.
of convective showers developed behind a pair of cold frontswith
the aid of a surface convergence line. The lowermost of these
layers (a tropopause fold on the eastern side of the depression)
helped to promote these showers by reducing the tropospheric
stability by introducing potential instability and by weakening a
lid that had previously been capping the convective development.
Furthermore, once the convection was initiated, the uppermost
dry layer (a fragment of the streamer on the western side of
the depression, drawn around the depression in the cyclonic ro-
tation) capped the convection at around 450 hPa. The period
when convection was possible was very short as it required all
the factors at work – the convergence line/surface moisture, ad-
equate surface temperature, increase in mid-tropospheric CAPE
and decrease in mid-tropospheric CIN – to all coincide, this ex-
plains the narrowness of the shower band. It was also shown that
while the convergence line triggered the convection it was the
potential instability that had the greater impact on the CAPE.
This case was intensively observed during the CSIP and some of
these measurements have been used to investigate the role and
origin of these features.
9. Acknowledgments
We wish to express our gratitude to the following organizations
and individuals: the British Atmospheric Data Centre (BADC)
for their provision of ECMWF data, the web trajectory service
andMSG images; the UKMet Office for the frontal analyses that
were reproduced in Figs. 2 and 12 and for theNimrod data shown
in Fig. 4; NASA for providing their TOMS ozone data on their
website (http://toms.gsfc.nasa.gov); CarolineRussell for launch-
ing radiosondes from Swanage during this IOP; Ben Bowerman
from Godlingston Manor Farm in Swanage for kindly allowing
us to launch radiosondes from his land; Dr. Markus Ramatschi
(GeoForschungsZentrum Potsdam) for collecting and providing
the GPS data that appear in Fig. 5; Dr. Grant Allen (University of
Manchester) for help with calculating PV as plotted in Fig. 11;
the two anonymous reviewers whose comments helped to con-
siderably improve the paper; the many participants of CSIP who
helped the project run successfully and the Natural Environment
Research Council (NERC) for supporting the MST radar as a
national facility and for funding CSIP.
References
Bader, M. J., Forbes, G. S., Grant, J. R., Lilley, R. B. E. and Waters,
A. J. 1995. Images in Weather Forecasting: A Practical Guide for In-
terpreting Satellite and Radar Imagery. Cambridge University Press,
Cambridge, 499 pp.
Bennett, L. J., Browning, K. A., Blyth, A. M., Parker, D. J. and Clark,
P. A. 2006. A review of the initiation of precipitating convection in
the United Kingdon. Quart. J. R. Meteorol. Soc. 132, 1001–1020.
Bjerknes, J. and Solberg, H. 1922. Life cycle of cyclones and the polar
front theory of atmospheric circulation. Geophys. Publ. 3, 1–18.
Browning, K. A. 1997. The dry intrusion perspective of extra-tropical
cyclone development. Meteorol. Appl. 4, 317–324.
Browning, K. A. andMonk, G. A. 1982. A simple model for the synoptic
analysis of cold fronts. Quart. J. R. Meteorol. Soc. 108, 435–452.
Browning, K. A. and Roberts, N. M. 1994. Use of satellite imagery
to diagnose events leading to frontal thunderstorms: part I of a case
study. Meteorol. Appl. 1, 303–310.
Browning, K. A. and Roberts, N. M. 1995. Use of satellite imagery to
diagnose events leading to frontal thunderstorms: part II of a case
study. Meteorol. Appl. 2, 3–9.
Browning, K. A., Blyth, A. M., Clark, P. A., Corsmeier, U., Morcrette,
C. J. and co-authors. 2007. The Convective Storm Initiation Project.
Bull. Am. Meteorol. Soc. 88, 1939–1955.
Clark, P. A. and Lean, H. 2006. An overview of high resolution UM
performance for CSIP cases. JCMM Internal Report, No. 155, 42pp.
Cullen, M. J. P. 1993. The unified forecast/climate model. Meteorol.
Mag. 122, 81–94.
Danielsen, E. F. 1964. Project Springfield Report. Defense Atomic Sup-
port Agency, 20301, DASA 1517(NTIS#AD-607980), Washington
D. C., 99pp.
Danielsen, E. F. 1968. Stratospheric-troposheric exchange based on ra-
dioactivity, ozone and potential vorticity. J. Atmos. Sci. 25, 502–
518.
Emanuel, K.A. 1994.Atmospheric Convection. OxfordUniversity Press,
Oxford, 580 pp.
Gage, K. S. and Green, J. A. 1981. Evidence for specular reflection from
monostatic VHF radar observations of the stratosphere. Radio Sci. 13,
991–1001.
Goddard, J. W. F., Eastment, J. D. and Thurai, M. 1994. The Chilbolton
Advanced Meteorological Radar: a tool for multidisciplinary atmo-
spheric research. Electron. Commun. Eng. J. 6, 77–86.
Golding, B. 1998. Nimrod: a system for generating automated very short
range forecasts. Meteorol. Appl. 5, 1–16.
Golding, B., Clark, P. A. and May, B. 2005. The Boscastle flood: meteo-
rological analysis of the conditions leading to flooding on 16 August
2004. Weather 60, 230–235.
Griffiths, M., Thorpe, A. J. and Browning, K. A. 2000. Convective desta-
bilization by a tropopause fold diagnosed using potential-vorticity
inversion. Quart. J. R. Meteorol. Soc. 126, 125–144.
Heath, D. F., Krueger, A. J., Roeder, H. A. and Henderson, B. D. 1975.
The solar back-scatter ultraviolet and total ozone mapping spectrom-
eter (SBUV/TOMS) for Nimbus G. Opt. Eng. 14, 323–331.
Hill, F. F. andBrowning,K.A. 1987. Case study of a persistentmesoscale
cold pool. Meteorol. Mag. 116, 297–309.
Hoskins, B. J., McIntyre, E. M. and Robertson, A. W. 1985. On the use
and significance of isentropic potential vorticity maps. Quart. J. R.
Meteorol. Soc. 111, 877–946.
Hunt, J. C. R., Orr, A., Rottman, J. W. and Capon, R. 2004. Coriolis
effects in mesoscale flows with sharp changes in surface conditions.
Quart. J. R. Meteorol. Soc. 130, 2703–2731.
Marsham, J. H. and Parker, D. J. 2006. Secondary iniation of multiple
bands of cumulonimbus over southern Britain, part II: dynamics of
secondary initiation. Quart. J. R. Meteorol. Soc. 132, 1053–1072.
Marsham, J. H., Morcrette, C. J., Browning, K. A., Blyth, A. M., Parker,
D. J. and co-authors. 2007a. Variable cirrus shading during CSIP IOP
5. I: effects on the initiation of convection. Quart. J. R. Meteorol. Soc.
133, 1643–1660.
Tellus 61A (2009), 2

A CONVECTIVE STORM INITIATION PROJECT CASE STUDY 263
Marsham, J. H., Blyth, A. M., Parker, D. J., Beswick, K., Browning, K.
A. and co-authors. 2007b. Variable cirrus shading during CSIP IOP
5. II: effects on the convective boundary layer. Quart. J. R. Meteorol.
Soc. 133, 1661–1675.
Morcrette, C. J., Browning, K. A., Blyth, A. M., Bozier, K. E., Clark,
P. A. and co-authors. 2006. Secondary iniation of multiple bands of
cumulonimbus over southern Britain, part I: an observational case
study. Quart. J. R. Meteorol. Soc. 132, 1021–1051.
Morcrette, C. J., Lean, H., Browning, K. A., Roberts, N., Clark,
P. A. and co-authors. 2007. Combination of mesoscale and synop-
tic mechanisms for triggering of an isolated thunderstorm: a case
study of CSIP IOP 1. Mon. Wea. Rev. 135, 3728–3749.
Muschinski, A. and Wode, C. 1998. First in situ evidence for coex-
isting submeter temperature and humidity sheets in the lower free
troposphere. J. Atmos. Sci. 55, 2893–2906.
Norton, E. G., Vaughan, G., Methven, J., Coe, H., Brooks, B. and co-
authors. 2006. Boundary layer structure and decoupling from synoptic
scale flow during NAMBLEX. Atmos. Chem. Phys. 6, 433–445.
Roberts, N. M. 2000. The relationship between water vapour imagery
and thunderstorms. JCMM Internal Report, No. 110, 40 pp.
Russell, A., Vaughan, G., Norton, E. G., Morcrette, C. J., Brown-
ing, K. A. and co-authors. 2008. Convective inhibition beneath an
upper-level PV anomaly. Quart. J. R. Meteorol. Soc. 134, 371–
383.
Schmetz, J., Pili, P., Tjemkes, S., Just, D., Kerkmann, J. and co-authors.
2002. An introduction to Meteosat Second Generation (MSG). Bull.
Am. Meteorol. Soc. 83, 977–992.
Simpson, J. E. 1997. Gravity Currents in the Environment and the Lab-
oratory 2nd Edition. Cambridge University Press, Cambridge, 244
pp.
Thorncroft, C. D., Hoskins, B. J. and McIntyre, E. M. 1993. Two
paradigms of baroclinic-wave life cycle behaviour. Quart. J. R. Mete-
orol. Soc. 119, 17–55.
Vaughan, G. 2002. The UK MST radar. Weather 57, 67–73.
Vaughan, G., Price, J. D. and Howells, A. 1994. Transport into the
troposphere in a tropopause fold. Quart. J. R. Meteorol. Soc. 120,
1085–1103.
Ware, R., Solheim, F., Carpenter, R., Gueldner, J., Liljegren, J. and co-
authors. 2003. A multi-channel radiometric profiler of temperature,
humidity and cloud liquid. Radio Sci. 38, 8079–8092.
Tellus 61A (2009), 2