Forecasting Severe Convective Hazards

Use of EFI with CAPE and CAPE-shear

Deep moist convection happens when deep instability (steep mid-tropospheric lapse rates), moisture and lift are all in place together.  If any of these ingredients is missing, deep moist convection is impossible. 

Presentations of Extreme Forecast Index (EFI) for CAPE and CAPE-shear can help with the early identification of the potential for severe weather related to convection. 

  • CAPE gives information on the the convective energy and the availability of low-level moisture.  It is only one measure of the potential for severe convection and thunderstorms - sufficient lift to overcome any capping inversion and to release that instability is also necessary.
  • CAPE-shear combines CAPE and deep-layer bulk shear (computed at ECMWF between 925hPa and 500hPa).  High values of this parameter indicate the potential for well-organised and long-lived convective storms such as MCSs and supercells.

A combination of convective EFI parameters and a precipitation forecast (probability of precipitation (PoP), available on ecCharts) highlights areas where severe convection is likely and is a powerful indicator of timing and location given some method of initial uplift to release the instability.


Fig8.1.9.6-1: 500hPa contour heights, forecast data time 00Z 21 June 17, T+48 verifying at 00Z 23 June 17.   CAPE-shear EFI for the period T+24 to T+48 coloured Yellow >0.4, Orange > 0.7, Red > 0.8.


Fig8.1.9.6-2: As Fig8.1.9.6-1 but with HRES precipitation totals over 9hrs added: purple > 10mm.  In practice, the fact that there is precipitation indicates sufficient availability of moisture while the very high EFI indicates that unusual (i.e. climatologically high as defined by M-climate) convective available potential energy (CAPE) is available in the north Germany area.  Precipitation totals in the very active storms that are likely to form will be greater than ENS or HRES show (here HRES precipitation) and with associated significant downdraught gusts.

 

   


Fig8.1.9.6-3: CAPE-shear EFI from a sequence of forecasts data times 00UTC on 18, 19, 20, 21, 22 June 2017.  Note the increasing Extreme Forecast Index (EFI) and the Shift of Tails (SOT) above 0 and reaching above 1 at T+24 on the last forecast over North Germany and Poland. 

 

Fig8.1.9.6-4: CAPE-shear EFI, data time 00UTC 22 June 2017, valid for 00-24UTC 22 June 2013 (as on Fig8.1.9.6-3).  EFI colours orange and red taken as indicating an extreme event likely.  SOT values indicate the ratio of departures of ENS forecast values from the M-climate extreme considering the greatest 10% ENS members.  The other charts show CAPE-shear values in the M-climate (derived on 19 June 2017) wherein only 1 in 10 occasions (central chart) and only 1 in 100 occasions realises more than the values shown.  The existence of significant EFI and SOT, even some days in advance, should not be overlooked, particularly if the actual forecast CAPE-shear values are much greater than the M-climate values (at say the 90th or 99th percentiles) for the area.

 



Fig8.1.9.6-5: Maximum gusts (kph) during the period 12UTC to 18UTC 22 June 2017.  Over 100kph in Central Germany associated with the widespread active thunderstorms.



Fig8.1.9.6-6: Rainfall (mm) in 6hrs during the period 12UTC to 18UTC 22 June 2017.  Over 50mm of rain fell in Central Germany associated with the widespread active thunderstorms.

Care needed in Interpretation of ecChart Presentation.

It is tempting to simply observe on forecast charts where large CAPE or CAPE-shear EFIs coincide with high rainfall from HRES when assessing the release of severe convection.  HRES forecast rainfall may be used in combination with convective EFIs in the short-range (up to T+48hr), but it should be remembered that HRES (& CTRL) is just another individual possible forecast.  In the short-range it is probably the most likely one, but in the medium-range its relative weight compared to ENS members decreases and ensemble Control (and HRES) become just as likely as any other ensemble member.  Then it is best to use a probability of precipitation forecast (PoP > 1mm/24hr) rather than a simple precipitation forecast throughout the whole forecast period (both short-range and medium-range).  These concepts are discussed below using one case as an example.


All the charts below correspond to the same example. All are for data time 00UTC 6th August 2017, and we focus on the forecast for 8th August.  Fig8.1.9.6-7 and Fig8.1.9.6-8 show 6-hour HRES precipitation forecasts for 00UTC on the 8 and 9 Aug 2017 as displayed by ecCharts.  It appears an area of significant rainfall associated with an upper trough moves from southwest France to Austria.   However, precipitation data is not shown for 12UTC on 8 Aug.  Meanwhile 24-hour total precipitation EFI (0.9) (Fig8.1.9.6-9) and CAPE-shear EFI (0.85) (Fig8.1.9.6-11) are available for 00UTC 9 Aug and show very high values.  CAPE EFI (0.6) (Fig8.1.9.6-10)  is only moderate illustrating the significant impact of bulk shear to give the high CAPE-shear EFI values.  The precipitation meteogram for the western Alps shows heavy rainfall in that area during the day and this is confirmed by data on Fig8.1.9.6-8 and Fig8.1.9.6-12 (note that these charts have different but overlapping validity periods). 

 

Fig8.1.9.6-7: ecChart showing 300hPa height with stratiform and convective rainfall (convective rainfall is plotted on top of stratiform) over the last 6 hr for T+48hr verifying at 00UTC 08 Aug 2017 based on HRES data time 00UTC 6 August 2017.

 

Fig8.1.9.6-8: ecChart showing 300hPa height with stratiform and convective rainfall over the last 6 hr for T+72hr verifying at 00UTC 09 Aug 2017 based on HRES data time 00UTC 6 August 2017

 

Fig8.1.9.6-9: ecChart showing 300hPa height with total precipitation EFI at T+72hr verifying at 00UTC 09 Aug 2017 based on ENS data time 00UTC 6 August 2017

 

Fig8.1.9.6-10: ecChart showing 300hPa height with CAPE EFI for the 24h ending at 00UTC 09 Aug 2017 based on ENS data time 00UTC 6 August 2017

 

Fig8.1.9.6-11: ecChart showing 300hPa height with CAPE Shear EFI for the 24h ending at 00UTC 09 Aug 2017 based on ENS data time 00UTC 6 August 2017

 

Fig8.1.9.6-12: ECMWF chart showing stratiform and convective rainfall over the last 6 hr for T+60hr verifying at 12UTC 08 Aug 2017 based on HRES data time 00UTC 6 August 2017. Also shown are surface isobars.

 

Fig8.1.9.6-13: ecChart showing the probability of precipitation ≥20mm in 24hrs ending 18UTC 9 August 2017. The forecast probability of heavy rainfall is concentrated at about 6ºE, in amongst the forecast CAPE-shear EFI maxima in Fig8.1.9.6.12.

 

Fig8.1.9.6-14:  15day meteogram with M-climate for 45N06E based on ENS data time 00UTC 6 August 2017.  An exceptional event is forecast for 8 Aug; the median lies above the 99th percentile of M-climate (green line).

 

Fig8.1.9.6-15: CDF and associated EFI for west Alps region (45N06E).  Forecasts with a data time of 00UTC 6th, as on the other plots above, are denoted by the darkest of the two dashed blue lines.  There is a consistently high EFI for rainfall (over 80%) which is sufficient for forecasting a significant and maybe an extreme rainfall event.  Some ENS members show rainfall totals close to the M-climate maxima.  The slope of the precipitation CDF shows the variation within ENS members, but all members show greater than M-climate values.  For greater confidence the slope of the CDF should be more vertical.


Mesoscale Convective Systems (MCS)

The aim of any weather prediction scheme is to be able to forecast extreme weather events.  Some of the most significant severe convective outbreaks are associated with Mesoscale Convective Systems (MCS).

An MCS may be defined as: 

“A cloud system that produces a contiguous precipitation area, typically 100km or more in extent in one or more directions, containing deep embedded convective cells and active thunderstorms (e.g. squall lines, bow echoes, Mesoscale Convective Complexes (MCCs)).” 

MCSs are often associated with prolonged heavy rain, active thunderstorms together with strong wind gusts, and, occasionally, damaging hail and tornadoes.  Most severe weather occurs during the initial or developing stage of the MCS.  However, heavy rain and flash floods tend to occur in later stages of more mature systems. 

Typically MCSs develop mid- to late afternoon and then persist through the evening and well into the night.

MCSs tend to fall into two types:

  • Type 1: in association with widespread strong forcing (e.g. along a frontal zone).
  • Type 2: driven by their own cold pool.

A feed of warm and very moist air into the system is desirable for the most active MCS.


The resolution  (~9km) of HRES (& CTRL) allows reasonable capture of the area of a large MCS –  but detail of convective areas within, nor any narrow features (e.g. squall lines), won’t be very well defined.

The current resolution of ENS (~18km) means MCSs are less likely to be represented in the forecast, and individual storms within the MCS won’t be identified or tracked.  However, the scale of an MCS is such that it can substantially alter the surrounding atmospheric environment and potentially affect larger and therefore resolved scales.  ENS can usefully identify environmental conditions promoting deep moist convection and MCSs and hence products such as the EFI of CAPE or CAPEshear can give an indication of potential for MCS development.


An investigation has shown that:

  • HRES has high skill in predicting MCS in the first 24hr or so but skill falls away beyond 36hr.  Nevertheless, warnings of the potential for extreme weather are very important, even at short lead-times (Fig8.1.4.5.16).
  • ENS can’t track individual storms but is good at predicting the environment that favours development of extreme convection.
  • EFI for CAPE and CAPE-shear shows high skill to day 4 and there is still good correspondence between EFI and severe convective outbreaks even at day 7 (Fig8.1.4.5.17).
  • HRES and ENS can discriminate well between days of intense convective activity and days of less convection – they don’t over-predict MCS.  Generally, MCSs are not predicted during periods of less active weather and convection, and this corresponds well with observations.

Fig8.1.9.6-16: Probability of Detection and False Alarm Ratio results from an initial investigation on the ability of HRES in capture of MCS.



Fig8.1.9.6-17:  ROCA Diagram showing skill of CAPE and CAPE-shear Extreme Forecast Index (EFI) at recognising severe convective outbreaks (verified against observed MCS).  The area under the Relative Operating Characteristics (ROCA) curve gives an indication of skill (1.0 = High Skill; 0.5 = No Skill).  The EFI is verified against severe weather reports in the European Severe Weather Database (ESWD) averaged over the April-September periods between 2017 and 2020.


Fig8.1.9.6-18: Comparison between HRES output and observed distribution of MCS areas over Europe.  HRES data time 23 June 2021 12UTC, Verifying time 24 June 2021 00UTC. Note MCS are persisting during the night.

Supercell examples

Example 1:

Right-moving supercells (highly-organised thunderstorms with cyclonic flow at the mesoscale) developed over NE Spain producing giant hail and floods in Zaragoza.  Large deep-layer shear (over 30 m s-1 0-6 km shear) coexisted  with a quite large MUCAPE.  ENS mostly about 1500J kg-1 with extreme above 2000 J kg-1.   Very low values of convective inhibition (CIN) in the moist air in the lowest layers and the level of free convection (LFC) and lifted condensation level (LCL) were strong signals for the very active convection.   

Fig8.1.9.6-19: Forecast vertical profile for Zaragoza DT 00UTC 6 July 2023 VT 12UTC 6 July 2023.  Large MUCAPE can be released after overcoming small CIN. Strong bulk wind shear.  The ENS cumulative distribution function for precipitation at Zaragoza during the 24hr period 00UTC 6 July to 00UTC 7 July 2023.  The CDFs for DT 00UTC 5 July (purple), DT 12UTC 5 July (blue), DT 00UTC 6 July (red) all show about 5% of ensemble members produce precipitation greater than the extreme of M-climate for Zaragoza.

Fig8.1.9.6-20: Channel-9 and Channel-12 imagery VT 15UTC 6 July 2023 showing supercells over NE Spain.Example 2:

Example 2:

During the winter months, one can easily downplay the signal from the convective EFIs.  This is because the model climate for CAPE and CAPE-shear don't have particularly extreme values at that time of year.  Thus almost any signal of CAPE or CAPE-shear is often portrayed as extreme.  However, high CAPE and CAPE-shear values should not be underestimated.

In the case illustrated (Figs8.1.9.6-21 & 22) the structure of the convective EFIs patterns differs.  These are:

  • extreme CAPE (for the season) across southern Ireland to northern England. 
  • extreme CAPE-shear (for the season) across Celtic Sea towards Wales.  In this example CAPE-shear is dominated by the extreme wind shear.

Severe thunderstorms, even supercell thunderstorms, can develop where the CAPE and CAPE-shear areas overlap and high storm helicity as all the necessary ingredients for severe convection are in place . 

 

Fig8.1.9.6-21: Forecast Extreme Forecast Index for CAPE with Shift of Tails for 24hrs to VT 00UTC 27 Dec 2023, DT 00UTC 26 Dec 2023.


Fig8.1.9.6-22: Forecast Extreme Forecast Index for CAPE-shear with Shift of Tails for 24hrs to VT 00UTC 27 Dec 2023, DT 00UTC 26 Dec 2023.


 Fig8.1.9.6-23: Forecast wind hodograph relative to storm movement.  Curved forecast hodographs in the lowest 3 km give high helicity relative to the  storm - even in the lowest 500 m.  This is one of the predictors for tornadoes.

Example 3:

A major outbreak of severe convection occurred on 26 May and CAPE-shear EFI covers the wide area of severe convection.  Within the CAPE shear areas 43 tornadoes and hail very large hail (over 50mm diameter) occurred mainly in the west of the area and severe convective gusts were reported mainly in the east.  Tornadoes and very large hail usually occur in the presence of supercells.  The CAPE-shear EFI signal was much stronger than that of CAPE EFI. This gives information about the environment – large instability but also presence of strong deep-layer wind shear.  In such kind of environment well-organised convection tends to develop if enough lift is provided.

Fig8.1.9.6-24: Observed severe weather over USA 26 May 2024 (left diagrams). EFI forecast of CAPE and CAPE-shear  DT 00UTC  24 May 2024, VT 00UTC 26 May to 00UTC 27 May 2024 (right diagrams).  The CAPE-shear EFI signal is much stronger than that of CAPE EFI which illustrates the importance of strong instability together with significant shear when forecasting severe convective weather.  

Considerations when forecasting Mesoscale Convective Systems (MCS) and Supercells

When using IFS output, the user should keep in mind:

  • the limited ability of HRES (& CTRL), and particularly ENS, to resolve a potential MCS in detail.  Individual convective elements won’t be resolved.        
  • the characteristics of the airmass, particularly the moisture content of any convergent flow.  A persistent inflow of high moisture air encourages more activity.
  • changes in the forecast IR cloud output, lightning, and precipitation fields together with CAPE and CAPE-shear can point to likely areas for potential MCS formation.
  • that under certain circumstances of vertical wind shear, forcing, and cloud structure an MCS can comprise one or more supercells.  The MCS can split and significantly alter the MCS’s track and development.  Alternatively some supercells can back-build and become stationary.   These effects are unlikely to be captured by HRES (& CTRL).
  • most severe weather tends to occur during the initial or developing stage of an MCS.  However, heavy rain and flash floods continue in later stages of more mature systems.
  • MCS tend to develop mid- to late afternoon and then persist through the evening and well into the night.
  • that observed surface temperatures and dew points may differ from forecast values.  Users can then assess possible modifications to the lowest levels of the forecast vertical profiles and amend the convective inhibition accordingly.
  • Winter M-climate EFI values are low and moderate values of CAPE and CAPE-shear can appear extreme but should not be ignored as unlikely. 


Additional Sources of Information

(Note: In older material there may be references to issues that have subsequently been addressed)