Abrupt end to heatwave

After a brief spell of hot weather, enough to satisfy heatwave criteria over a large part of England and Wales peaking with a temperature of 33.4°C recorded at Heathrow, a change of type was preceded by an outbreak of locally severe convection across parts of northern England on Friday 26 June.

Overnight, elevated instability had been brough north and released within a marked θw (wet-bulb potential temperature) plume at 700hPa on the approach of a progressive Atlantic trough. The remnants of this can be seen as layers of decaying altocumulus on the visible satellite imagery valid at 12 Z (Figure 1, right) moving out into the North Sea and across southern Scotland. The previous afternoon this was responsible for some severe mid-level thunderstorms across the Western Isles, where large hail and localised flooding caused significant impacts, such as damage to the Rhenigidale Road in Harris which was washed away. This is a less common synoptic setup with a more meridionally oriented mid-level flow, a type identified and described as 'Modified Spanish Plume' (Lewis and Gray, 2010) which allows the plume to advect further north, and which typically can affect a much broader northern and western part of the UK instead of just the southeast of England.

Figure 1 shows an annotated visible satellite image taken at 12 Z.

Area A: This was the focus area for potential severe convection highlighted by the Mesoscale Discussions issued by the Guidance Unit through the morning. Surface temperatures were initially suppressed in this zone following rain and extensive medium-level cloud through the morning, however clearance by late morning allowed strong surface heating and eventually a cumulus field to develop here. This appeared to develop along the leading edge of a moisture rich boundary extending inland from the coast of SW England and S Wales.

Area B: A shallow boundary layer consisting of cooler and moister air had moved eastwards overnight following the plume of elevated mid-level instability, manifesting itself as areas of low stratus over the sea and windward coasts of England and Wales.

Discussion of available severe weather parameters

CAPE (Convective Available Potential Energy) is the most referenced parameter when judging the likely severity of convection – this uses both the moisture and heat available in the atmosphere to give an indication of how much energy there is to accelerate an air parcel vertically, and is calculated by vertically integrating between the environmental lapse rate and the theoretical parcel lapse rate, between the level of free convection and the equilibrium level (i.e. the area between these two curves on a tephigram). Values of CAPE, as well as their distribution in the column are both important, with for example taller, narrower profiles referred to as “Skinny CAPE” being generally more precipitation efficient than indicative of severe thunderstorms, however they do tend to have lower values of CAPE.

Figure 2,, is a forecast tephigram from the Global Model for a grid point close to Sheffield for 15 Z on Friday 26 June 2020. It shows an unadjusted CAPE of 1456 J/kg and a more representative MLCAPE of 672 J/kg (Mixed Layer CAPE - Mixing lowest 50–100hPa and lifting the resultant average virtual temperature and dewpoint to account for mixing within the boundary layer), both reasonably high values for the UK (Holley et al. 2014) and would indicate the likelihood of organised and potentially severe convection. Note that the higher CAPE value calculated here assumes that the tephigram is entirely representative of the wider environment and neglects significant processes such as entrainment. Also worthy of note here are the Bulk Richardson Number of 165 and the Hail Size parameter of 1.12 inches.

The Hail Size diagnostic uses the Fawbush and Miller hail technique – which is discussed in more detail in a previous blog (CS7).

The Bulk Richardson Number approximates the ratio of CAPE to shear (how efficient the atmosphere is at distributing and organising the available energy in a potential storm’s updraft) in order to diagnose the likely potential severity of any developing convection. Anything between 15 and 40 is generally considered to be supportive of the development of supercells.

Crucially the tephigram shows CIN (Convective Inhibition – the amount of energy required to lift a parcel of air to its level of free convection) of 0 J/kg, implying that an earlier cap on convection is expected to have been entirely eroded by this point in the day. A “loaded gun” scenario, often associated with Spanish Plume type events, has a small but surmountable amount of CIN, in order to store and allow a build of heat/energy at the surface, which can then all be released at once rather than gradually through "popcorn" convection. This is eventually achieved either through the removal of a cap through some large-scale dynamic process or strong mesoscale forcing such as a convergence zone, with values of 50 – 250 J/kg usually quite readily overcome.

The forecast from the UM suite Global Model across this part of northern England shows significant CAPE but only relatively low and generally uni-directional shear. A temperature of 27°C is enough to push an air parcel all the way to the troposphere and this air parcel is likely have enough buoyancy to overshoot towards 40,000 feet.

It is important to note that whilst the environment is only weakly sheared directionally, there is still enough speed shear of approximately 35 knots to separate and protect the updraft from the downdraft, to allow sustained and organised convection to form. Furthermore, the Lifted Index (diagnosed as the difference between environment and lifted air parcel at mid-levels, which on GPP output is diagnosed as 700 hPa but can more traditionally refer to 500 hPa) is between –3 and –5 and would indicate a risk of hail, more on this later. Finally, the precipitable water value of 33.9 mm is high and indicative of a very warm and moist airmass typical of mid-summer in Europe, likely to support some very high rain rates and significant precipitation loading, capable of producing strong downdrafts and surface gusts.

Figure 3 is a forecast ascent from further north at Durham for the same time. It is worth noting here that although the forecast CAPE/MLCAPE is slightly lower, the Bulk Richardson Number is within the favourable region (36, between 15 and 40) for the development of supercells. The main reason for this is a greater amount of shear, particularly between 850 and 600 hPa combined with a slightly lower CAPE. Shear of between 0 and 6 km of greater than 15–20 m s , approximately 30–40 KT is generally necessary to support supercells (Thompson et. al, 2003) and at least 15–20 KT to produce multicells and organsiation to convection. Therefore, whilst the thunderstorms initiating across the Midlands would be in a less favourable environment for severe convection, they would quickly move toward a more favourable zone later in the afternoon as they developed and moved northward.

A rapidly developing situation

As discussed earlier, by 12 Z a well-developed cumulus field had formed across the Midlands. An hour later (see Figure 4, 13 Z visible image, below right) this area had expanded and started to rapidly develop into a large area of scattered towering cumulus, perhaps aided by the convergence of the low-level flow and moisture as already discussed.

There was also a notable convection free area across eastern East Anglia behind the sea breeze here with surface temperatures a few degrees lower than the 28-30°C recorded widely inland (highest in area of interest was: Coningsby 30.9°C and Cranwell 30.8°C), as well as to the south of London with cooler onshore flow and subsidence behind the upper trough and the rapidly deepening convective zone.

Development into CB was even more rapid after this point, with the first cell forming close to East Midlands airport at 1330 Z and becoming active with frequent lightning strikes recorded by 1350 Z. Another group of active cells also developed shortly afterward along the sea breeze front extending N-S along the meridian. However, the most severe convection developed to the southeast of Sheffield between 1430 and 1500 Z. The satellite imagery (see Figure 5) and radar loop (see Figure 6) from the period immediately before this illustrate how quickly convection upscaled from the initial cumulus field into a large and well organised cluster of CBs, with the radar signature of a “Splitting Supercell” evident (divergent motion) as well as large bright echoes likely to have been created by a hail dominant core.

The visible satellite image at 15 Z (see Figure 7) shows clearly the venting of deep convection with the flow generally increasing from the south towards the tropopause, and a developing overshooting cirrus canopy quickly being taken forward of the storms to the north. In this most active cluster over Yorkshire large hail was reported in a few cells, one near to Sheffield (see Figure 8, as posted to Twitter by @IamEmmaB and @PaulHopwood14) at just before 15 Z, and another downwind and to the left of the same cluster across Leeds and Wakefield around 16 Z. Hail size was reported as >5 cm (2 inches) in Sheffield and widely there were reports of 2–4 cm diameter hail, coded as ‘GR’ in TAFs and METARs.

Forecasting hail

The empirical techniques available for forecasting hail size generally include the Lifted Index (LI) and CAPE, or some other proxy for estimating the buoyancy of an air parcel in the storm environment and therefore the updraft strength and likelihood of it being able to support large hail. All have their various pros and cons; so for example CAPE (maximum updraft strength = √(2CAPE)) will often overestimate significantly maximum updraft strength because it neglects processes such as entrainment and precipitation loading, and furthermore it doesn’t factor how the CAPE is distributed with concentration in a layer more supportive of stronger updrafts. The LI is better here because it makes a direct measure of buoyancy at one level of the atmosphere, and can this can be deliberately chosen to be the level at which hail/graupel is likely to be developing.

However, all these techniques are imperfect and breakdown to become a lot less useful when considering complex structures and processes such as those involved with the development of supercells.

Often it is better to use a combination of techniques (a poor man’s ensemble) in order to diagnose the risk of large hail, and then find where there is a good level of agreement. For example, in this case the Fawbush and Miller hail technique (see Figure 2 and Figure 3) was predicting hail in excess of 1 inch (the WBFL was <10,500 FT and therefore no further modification was required). It is worth noting that this technique is employed in the GPP output and that this therefore should be internally consistent.

Using MLCAPE of 500-700 J/kg would give an estimate of w of 30–40 m s-1 . This would, according to Knight and Knight (2001), suggest storms capable of supporting up to 5 cm (2 inch) hail – before dropping out of the cloud due to their fall speed overcoming the updraft. Finally, all the above becomes irrelevant if there is insufficient shear in order to separate the updraft and downdraft, and provide the storm with longevity, generally this is required in the UK to be around 35–40 knots in the lowest 6 km (~20,000 feet).

All of these techniques pointed to the risk of large hail, and indeed this was communicated in the Mesoscale Discussions issued by the Guidance Unit throughout the event.

Severe Gusts

This technique doesn’t always work, but it does allow the forecaster to quickly determine a useful rough approximation of the fog clearance time (within a 2-hour BECMG group range).

Another common feature of severe convection, particularly MCSs but also supercells, is sudden and locally severe wind gusts formed by a strong downdraft and its associated cold pool. There are a number of different empirical techniques that can be used to forecast the likely magnitude of these, though it must be stated that like with the techniques used to diagnose hail these are basic techniques and fail to take account of the myriad of complex processes that can alter any potential severe gust. The Fawbush and Miller and Ivens techniques are both discussed in more detail in a previous blog but I will summarise them briefly here again.

The Ivens method (developed in 1987) is used at KNMI in the Netherlands to predict the maximum wind velocity associated with heavy showers or thunderstorms. It is based on two multiple regression equations that were derived using about 120 summertime cases (April to September). The amount of negative buoyancy is estimated by the difference in wet-bulb potential temperature at 850 and at 500 hPa, and available momentum aloft is estimated using horizontal wind velocities at one or two fixed altitudes (250hPa and 850hPa). These are used to estimate the maximum wind velocity.

The Fawbush and Miller technique (developed in 1954) simply uses an SALR (Saturated Adiabatic Lapse Rate) from the Wet Bulb Zero level to the surface as a proxy for the “downdraft temperature” and then compares this to the ambient temperature outside the thunderstorms to determine the strength of peak gusts as per diagram. The greater the difference, the stronger the potential gust.

In this example the Ivens method predicted a maximum gust of 45 knots and the Fawbush and Miller method (often the higher of the two, for a worked diagram using the Watnall 11 Z ascent see Figure 9) predicted a gust of 0 knots, with an error bar of between 54 and 68 knots. Both predicted significant and potentially damaging gusts. The caveat to highlight here are that these are NOT to be used in TAFs but as a very low point probability that can either be expressed in warnings, general area forecasts, or verbally to the customer.

In this event a cell downdraft passed close by to one of the stations in the Met Office observing network, Warcop Range, which is situated in Cumbia just to the east of the Eden Valley and SW of the Great Dun Fell mountain range. The maximum gust of 56 knots was recorded at Warcop Range (see MMS trace in Figure 10) and the MMS digital observations from that period show also a dramatic fall in temperature concurrent with the strong gust of over 9 degrees Celsius.

Conclusion

This was an interesting event that brought some locally severe and disruptive weather to parts of northern England through the afternoon of 26 June. The favourable environment for severe convection was picked up several days ahead, and the high-resolution model output nearer the event was able to diagnose and generate both large hail and severe gusts. Additional tools such as empirical techniques combined with careful study of real-time radiosonde/AMDAR data and satellite imagery and surface observations allowed the nowcasting arm of the Guidance Unit to give excellent advice on where and when severe convection was likely through Mesoscale Discussions, and more importantly fast dissemination of this information out to responders and the appropriate customer base with a clear and consistent message communicated across the office.

Unfortunately, due to being on shift at the time I didn't have time to source the Doppler radar imagery which I'm sure would have helped to highlight more interesting structures of these storms, if anyone knows where or how I could source these, or indeed has some of this data then please get in touch.

Addendum

Small hail – in METAR/TAF code ‘GS’ (graupel) is a very different phenomena restricted to winter or Arctic airmasses outside of winter, where the freezing level is lower, typically below 4000 feet and is formed by the accretion of supercooled droplets in convective clouds formed over a relatively warm sea, over hills or through convergence.

The Boyden index (critical value 93–94) generally performs better overall of the various TS indicies, however the Jefferson index (critical value 25–26) has been found to perform better during winter over southeast England.

Lightning strike rate from UKV (see Figure 11 for the verification of the 15-16 Z forecast from the 12 Z UKV on Friday 26 June 2020) verified well against actual ATD strikes in this forecast period. This uses the vertical flux of graupel in the model at the MS 15 level (the region shown to generate charge separation) and the total ice (including graupel) mass in the model grid column. Note that brightest colours represent 5 strikes per hour within a GPP 2 x 2 km grid box and therefore the large areas of red/pink represent many grid boxes and thousands of lightning strikes, often an overestimate.

Finally, the NCM (National Climate Message - see Figure 12) for 26 June 2020 – highlights the extremely localised nature of the storms with Bingley recording 11.2mm and nearby Bradford only 0.4mm with the highest total picked up at Pately Bridge Ravens Nest (36.4mm). Thanks for reading!

References / Further reading

Fawbush, E.J. and Miller, R.C., 1953: A Method for Forecasting Hailstone Size at the Earth's Surface. Bull Am Meteorol Soc, 34, 235-244.

Fawbush, E.J. and Miller, R.C., 1954: A basis for forecasting peak wind gusts in non-frontal thunderstorms. Bull Am Meteorol Soc, 35, 14–19.

Holley, D.M., Dorling, S.R., Steele, C.J. and Earl, N., 2014: A climatology of convective available potential energy in Great Britain. Int. J Climatol., 34, 3811-3824.

Ivens, R. A. A. M.: 1987, Forecasting the maximum wind velocity in squalls. Symp. Mesoscale Analysis and Forecasting, ESA, 685–686

Knight, Charles, and Nancy Knight, 2001: Hailstorms. In Severe Convective Storms, C. A. Doswell III, Ed., Meteorological Monographs, volume 28, No. 50. Published by the American Meteorological Society

Lewis, M.W. and Gray, S.L., 2010: Categorisation of synoptic environments associated with mesoscale convective systems over the UK. Atmos. Res., 97, 194-213.

Thompson, R.L., R. Edwards, J.A. Hart, K.L. Elmore, and P. Markowski, 2003: Close Proximity Soundings within Supercell Environments Obtained from the Rapid Update Cycle. Wea. Forecasting, 18, 1243-1261