Page tree
Skip to end of metadata
Go to start of metadata

image © Mandy Cherundolo/iStock/Thinkstock

Convective Cloud Processes and Precipitation

The convection scheme does not predict individual convective clouds, but only their physical effect on the surrounding atmosphere in terms of latent heat release, precipitation and the associated transport of moisture and momentum.  The scheme differentiates between deep, shallow and mid-level convection but only one type of convection can occur at any given grid point at any one time.  An increase in the amount of super-cooled liquid water held by the convection scheme at colder temperatures (down to -38C) improves the development of convective precipitation.  New ways of forecasting the degree of sub-grid variability in precipitation totals have also been developed (Point Rainfall).  Future updates to the IFS may allow some of the convective precipitation (mainly as snow) to be advected downstream into adjoining grid boxes.     

The effects of the model convection (changes to the temperature or humidity) drift downwind with model winds.  However, any (convective) precipitation that is developed is considered to remain within the column and fall vertically downwards instantaneously (i.e. taking zero time to reach the surface).

In consequence, model showers are not advected with the wind during their life-cycle when in reality showers normally are.  In particular, any showers that the model develops over the sea do not penetrate beyond the coast.

Users should allow for penetration of maritime showers inland from windward coasts, especially in winter or with wintry precipitation because snowflakes fall more slowly than raindrops and thus advect further inland before reaching the ground.


CAPE and CIN are computed in the IFS according to parcel theory, assuming pseudoadiabatic ascent and no entrainment.  This is exactly what one should see if one analyses the tephigram.  CAPE and CAPE-shear EFI and SOT computations sample the hourly CAPE and CAPE-shear values during the 24-hour period and the maximum values are what is used. 

At any given grid point the convection scheme inspects the temperature and humidity structure progressively from the surface to 300hPa and if there exists a level of free convection (LFC) it evaluates the CAPE.  Entrainment of surrounding air is not considered and thus the CAPE is likely to be a slight overestimate.  The technique currently in use for estimating CAPE allows for the discovery of elevated instability, even at night, despite low-level stability.  Convective Inhibition (CIN) is assessed from the IFS model atmosphere in a similar way.  CAPE and CIN are computed in order to help the user assess the likelihood of severe convective storms.  CAPE-shear is a combination of bulk shear (vector wind shear in the lowest 6km of the atmosphere) and CAPE and is used to identify areas of potentially extreme convection. 


Vertical wind shear tends to promote thunderstorm organisation, although excessive wind shear can be detrimental to convective initiation by increasing entrainment of environmental air into the storm.  But if active convection is indeed established, then the larger the wind shear the higher organisation and severity of convection tends to be.  For example supercells can be very long-lived (more than 6 hours in some cases).  They produce the majority of strong to violent tornadoes and very large hail (more than 5cm in diameter) and they tend to occur in environments with strong wind shear (0-6 km shear > 20 ms-1).   Therefore for diagnostic purposes both CAPE and CAPE-shear should be used together, or alternatively one can examine together CAPE and wind shear as separate parameters.  The CAPE-shear EFI may be used to anticipate well-organised severe thunderstorms.  For example in the case of very strong wind shear but relatively modest CAPE (e.g. a few hundred J/kg) we can have development of well-organised severe thunderstorms but the EFI for CAPE will give a much weaker signal than the EFI for CAPE-shear.  For extremely severe thunderstorms we need high CAPE and high shear; therefore both EFIs for CAPE and CAPE-shear should show a strong signal.

Fig2.1.23: Rough guidelines on how to use CAPE and CAPE-shear (EFI) values together. Bear in mind also that EFI and SOT are computed relative to reference model climatologies, so "severe" in one region will tend not be at the same level as it is in another.

It is vital to try to diagnose whether or not convection will initiate before giving considering to convective severity, as suggested by CAPE and CAPE-shear. Some broad guidelines, based on the level of Convective INhibition (CIN) can be:

  • small CIN (e.g. <50 J/Kg): diurnal heating and/or local topographic features would be sufficient for triggering
  • moderate CIN (e.g. 50 J/kg to 100 J/Kg): needs more substantial uplift than provided by diurnal heating alone
  • high CIN (e.g. >100 J/kg) needs very substantial uplift (e.g. a well-defined airmass boundary with strong surface convergence), and depending on the CIN level even that may not be enough

These values are not definitive.  It is the users responsibility to assess the impact of local effects (e.g. convergence, changes in the temperature and moisture structure, sea breezes, low. cloud advection etc.) upon the amount of energy required to overcome inhibition. 

Forecast charts:

Some indication of severe precipitation may be deduced from the values of CAPE  and CAPE-shear, coincident with indications of significant rainfall intensity and/or indication of an upper contour pattern favourable for forced broadscale ascent.

Considerations in interpretation of CAPE charts

When deducing the forecast rainfall distribution, in convective situations, it is important that users do not rely simply upon CAPE and CAPE-SHEAR charts alone. At first sight these charts appear to signal areas of high probability of deep and active instability.  However, they do not give information on the amount of available moisture and hence no information is given on the initiation or even potential existence of moist convection and consequent showery precipitation. This is especially important when there is a possibility of very heavy or severe instability-related precipitation.  Users should investigate closely all aspects of the forecast model atmosphere.  In particular the charts of forecast precipitation should be viewed to identify areas where there is an overlap with the forecast CAPE or CAPE-SHEAR areas - showers are generally not likely to happen if no forecast precipitation is indicated, no matter how large the values of CAPE or CAPE-SHEAR.  It is vital to view the precipitation fields and vertical profiles in connection with CAPE and CAPE-SHEAR fields.

CAPE6 charts have been introduced to show the greatest CAPE within the previous six hours. This is in order to limit data overload from too many single CAPE snap-shots; instead now we can cover all hourly values with just data for the main forecast times.  This can give the user a much better indication of the potential for active convection.

Users should note that evaluation of CAPE differs amongst models at individual forecast centres.  Until the method of computation of CAPE is standardised it is unsafe to compare the magnitude of CAPE derived by different forecast models though of course the changes in magnitude of CAPE derived from each forecast model remain useful.

An example - Northern Greece, 11 July 2019

Large values of CAPE lie in a zone across the Aegean Sea and parts of mid-Greece coincident with a belt of strong vertical wind shear resulting in very high values of CAPE-SHEAR (Figs 2.1.24 & 2.1.25).  In particular, high forecast values of CAPE and CAPE-SHEAR are indicated at Pilio while much lower forecast values are shown at Kavala.  This might suggest at first sight that any instability that is released in the region of Pilio would be very active with the possibility of severe storms and rainfall.  At the same time much less showery activity might be expected at Kavala on the northern flank of the CAPE and CAPE-SHEAR zone.  Such a snap assessment would be incorrect.


Fig2.1.24a: Forecast CAPE (Blue high, Red low) and Fig2.1.24b: Bulk Wind Shear (Orange high, Yellow low).  T+24 VT00UTC 11 July 2019, DT00UTC 10 July 2019.  


Fig2.1.25a: Forecast CAPE-SHEAR (Purple high, Blue low).  Fig2.1.25b: Max CAPE-SHEAR (Red high, Blue low). T+24 VT00UTC 11 July 2019, DT00UTC 10 July 2019.  Very high values are indicated in the vicinity of Pilio.  More modest values are indicated in the vicinity of Kavala on the CAPE-SHEAR chart but note that the maximum CAPE-SHEAR chart shows there have been much higher values during the previous 6hrs.

The forecast precipitation field (Fig2.1.26) shows a belt of rainfall across North Greece, Albania and Bulgaria.  This indicates that, as a minimum, in this area there is sufficient moisture in the forecast atmosphere to provide precipitation.  The area of forecast precipitation intersects the northern flank of the forecast CAPE and CAPE-SHEAR areas and thus it is this area that is more likely to see release of deep and active convection with availability of plenty of moisture.   Little or no precipitation is indicated in mid-Greece but nevertheless these lie within the areas of very high CAPE and isolated but local very heavy showers are possible and, bearing in mind the high bulk shear and CAPE-SHEAR values, local storms cannot be ruled out. 

Fig2.1.26: Forecast precipitation (12hr). T+24 VT00UTC 11 July 2019, DT00UTC 10 July 2019.

The corresponding diagnostic charts for the probability of high rainfall (>40mm/24hr) and Extreme Forecast Index (EFI) for precipitation (Fig2.1.27) identify the areas at greatest risk of a major precipitation event.     

Fig2.1.27a: Probability of total precipitation >40mm (24hr).  Green shading represents 35-65% probability.  Fig2.1.27b: Precipitation extreme forecast index (EFI).  Red shading represents EFI>0.8, Dark red >0.9 EFI. T+24 VT00UTC 11 July 2019, DT00UTC 10 July 2019.

Forecast vertical profiles are very helpful in assessing the potential for severe events.  The forecast vertical profile at Pilio shows large CAPE but with relatively dry convection, possibly released by high surface daytime temperatures.  Very little moisture is indicated and precipitation looks very unlikely.  However some moisture is available locally over mid-Greece (Fig 2.1.26) mainly at medium levels producing possible local showery outbreaks given some form of dynamic uplift.  The relevant wind shear to consider for this is probably between medium and upper tropospheric levels rather than between lower and medium levels (the bulk shear).  Inspection of the hodograph suggests the upper tropospheric shear is not great, so shower organisation/activity would lack this element of support.  Note, however, that heavy medium level showers can penetrate downwards through underlying dry layers more than IFS forecasts tend to suggest, even reaching down to the surface.  HRES and some ENS members do show a very humid boundary layer at Pilio, but it would require large energy input at the surface (2m temperatures above about 35°C) to overcome the large CIN and to lift the low level moisture to release moist convective cells.

The forecast vertical profile at Kavala shows rather less CAPE and CAPE-SHEAR but with an almost saturated atmosphere and absolute instability at around 750hPa.  Recall also that the max CAPE-SHEAR over the previous 6hrs was higher.  So very active moist convection is extremely likely in the northern Greece region.  Inspection of the hodograph suggests significant shear throughout lower and medium layers allowing separation of up draughts and down draughts with persistent active precipitation cells.  

Violent storms with local hail swept across northern Greece overnight 10/11 July 2019 causing seven deaths and widespread damage.

Fig2.1.28: Forecast vertical profiles for Kavala and Pilio, Greece. T+24 VT00UTC 11 July 2019, DT00UTC 10 July 2019.

A sequence of forecast EFI charts gives early indication of forthcoming severe weather potential (Fig2.1.29), and some idea of the confidence that may be placed on the forecast event.  In this case, northern Greece is identified as being at moderately high risk of an extreme event (EFI ~ 0.6) four days before, rising steadily to a very high risk of an extreme event (EFI ~ 0.9) two days before the occurrence of the severe weather.  Note how there is consistent indication of a very high risk of an extreme event (EFI ~ 0.9) over the Balkan states through the sequence of forecast runs.  The consistency in the areas shown at risk leads to a higher confidence in forecasts of severe weather.  Users should inspect forecast fields using ecCharts and vertical profiles as outlined above to assess forecast details, and also add in the influence of additional factors using local knowledge (e.g. regarding topographic influences) wherever possible.   

Fig2.1.29: Sequence of EFI precipitation charts from four EFI runs at 24hr intervals (DT 12UTC on 6, 7, 8, 9 July 2019). Increasingly high EFI precipitation values identify the areas at greatest risk.

Points to note regarding IFS forecasts of convection and convective precipitation.

The IFS currently shows:

  • a bias towards:
    • insufficient convective precipitation in arid regions (e.g. parts of West Africa, the Middle East, and central Australia).
    • excessive convective rainfall near orography.
  • a tendency to:
    • under-forecast precipitation amounts from large scale convection (e.g. MCSs),
    • over-forecast convective precipitation amounts otherwise.
  • imprecision in the diurnal cycle of convection.  In particular:
    • the convective inhibition (CIN) tends break too easily.  CIN is not currently well evaluated and consequently there can be unreliable timing of release convection – generally too soon. 
    • there is too rapid increase to a peak in convection (by about 3-4hrs, i.e. around local noon rather than in the afternoon).  
    • there is too rapid decay in convection.  CAPE is destroyed too quickly and showers die away too soon.  With active convection some showers may be expected to persist much longer and linger into the night.  In general, showers die out:
      • 2-3hrs too early in west Europe,
      • 1-2hrs too early in east Europe,
      • about right in USSR.

Considerations regarding medium level instability in drier areas

It is important that moist medium level instability is modelled sufficiently as even relatively small CAPE can produce precipitation.  Users should check forecast vertical profiles against local observations and profiles.

Heavy precipitation developed aloft from medium level instability can have drop sizes sufficiently large that they will penetrate through dry air to reach the ground.  IFS tends to evaporate precipitation from medium levels too much during descent (in part due to limitations of assumed drop size distribution), and consequently insufficient rain is forecast to reach the ground.  

Lightning associated with medium level instability is often indicated on forecast charts although no precipitation is forecast at the surface. Whilst lightning activity tends to be over-predicted (sometimes considerably) it can be a reasonable indicator of the potential for active medium level instability.  Forecast lightning activity often covers a greater area than does forecast surface precipitation. 

Convective Available Potential Energy (CAPE) is very sensitive to the humidity in the boundary layer.   A slight change in dewpoint, particularly within the boundary layer will leads to a significant change in CAPE.  Thus any medium level showers that do penetrate to the surface can locally increase boundary layer moisture - observed surface dew points can become several ºC higher than forecast T2m dew points (up to ~12ºC difference has been observed).  This leads to a local, possibly major, reduction in CIN and an increase in CAPE.   Further instability may then be released inducing further showery activity.   Forecast charts of surface precipitation are not likely to capture all such details.

Where medium level instability is forecast above a dry lower atmosphere, users should use forecast lightning charts and forecast vertical profiles to extend and improve precipitation forecasts.   Where medium level instability is forecast (even with only moderate CAPE), some additional showers should be forecast within the areas of forecast lightning.  Owing to resolution issues, forecast intensity of lightning strikes gives only a rough idea of regions where there is more active medium level instability but it does not reliably indicate that showers will penetrate to the surface, nor their intensity if they do so.  However, probability of precipitation should be increased. 

An example - Central Australia, 17 January 2019

Fig2.1.30: Forecast IFS data for central and northwest Australia 17 Jan 2019.  Local time is about 10hrs ahead of European time zones. The circled triangle locates Alice Springs.

  • Fig2.1.30a: Total HRES 6hr precipitation T+21: DT 12UTC 16 Jan 19, VT 09UTC 17 Jan 19.
  • Fig2.1.30b: Lightning density in 6hr (flashes 100km-2hr-1) T+21: DT 12UTC 16 Jan 19, VT 09UTC 17 Jan 19.
  • Fig2.1.30c: ENS probability of total precipitation >1mm: DT 12UTC 16 Jan 19. VT 00UTC 17 Jan 19 to 00UTC 18 Jan 19.
  • Fig2.1.30d: Total HRES 6hr precipitation T+21: DT 12UTC 16 Jan 19. VT 09UTC 17 Jan 19, and Observed lightning flashes VT 09UTC 17 Jan 19.
  • Fig2.1.30e: Forecast vertical profile at Alice Springs T+18: DT 12UTC 16 Jan 19, VT 06UTC 17 Jan 19.

Medium level thunderstorms developed and extended well into central parts of Australia (Fig2.1.30d, observed lightning) but no underlying surface rainfall is forecast (Figs2.1.30a & 2.1.30d), nor any probability of rain (Fig2.1.30c).  Forecast lightning flashes (Fig2.1.30b) is overly extensive in northwest Australia but although there is some indication in central parts it is under-indicated (compare with Fig2.1.30d). 

The model boundary layer was generally dry in central Australia but observations showed much higher dew points where showers have occurred.  Near Alice Springs the model T2m dewpoint was 4.2C lower than the observed dew point, and at a location to the northwest the error was 11.8ºC.  Both discrepancies were probably due to storms that the model didn't represent.

The forecast vertical profile for Alice Springs (Fig2.1.30e) shows possible (surface-based) medium level instability with just moderate CAPE (e.g. cyan line construction).  Note that some ENS members have higher low-level dew points which means a lower CIN to initiate medium level convection with greater CAPE (e.g. red dashed line construction).  Further, if the boundary layer is moistened after any medium level showers penetrate to the surface then there is a higher likelihood of more energetic convection being released afterwards with much greater CAPE (e.g. black dashed line construction).

In Central Australia, no precipitation is indicated; any precipitation in the model is being evaporated before reaching the ground.  However, the lightning activity chart suggests that, though the deep moist convection isn't very well-organised, scattered thunderstorms appear likely.  This was bourne out by observations.  Note also that the model greatly over-predicted lightning activity over northwest Australia.

Additional Sources of Information

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

Updated/Amended 30/12/19 - Link to issue with CIN

Updated/Amended 02/07/20 - Removed link to issue with CIN. Not applicable with 47R1.

Updated/Amended 24/10/20 - amended chart links to open access.

Updated/Amended 05/11/20 - added section on CIN, CAPE and Cape-shear.

Amended/Updated 24/03/21 – Addition of Section Considerations in interpretation of CAPE charts.

  • No labels