Precipitation over mountainous coasts and islands

Too much precipitation can be forecast over mountainous coasts and islands.  At a location on a mountain the model height can be significantly lower than the true height of the of the land surface.  See a schematic of model representation of orography and Tenerife as an example of a mountainous island.  Thus temperatures at the location can be forecast to be too warm.  This can then result in in a reduction of CIN from true values and the release of convection, possibly with large MUCAPE, over the mountains.

Higher forecast temperatures inland are likely to induce more onshore flow of moist air from nearby sea or lake.  This can substantially alter the structure of the forecast airmass and possibly encourage release of convection.  It is possible there may be excess convergence in ensemble control and the ensemble.  This would result in greater vertical velocity and moisture convergence that could lead to errors in forecast precipitation over steep or poorly resolved orography.

It is for the forecaster to assess critically the expected and changing structure and evolution of the airmass at a given upland or upslope location.  Forecasters should not rely on a single model forecast, but view the ensemble of forecasts and a whole.  Further, although precipitation could be heavy over the mountains, it might not be widespread nor extend to adjacent low-lying areas.   Equally, dry zone underlying the altitude of the mountain convection may reduce precipitation penetrating to lower levels. 

Orographic and Rainshadow effects

Orographic and rain shadow effects can be strong in unstable onshore air flow, particularly when an unstable marine airmass meets coastal mountains.   However, precipitation forecasts can be incorrectly represented.

The forced uplift triggers immediate development of parametrised showers.  The modelled convective precipitation (snow in the illustrated case) then falls immediately and vertically to the ground.  

In reality the showers take time to grow while also being driven downwind.  The snow that falls from them also drifts downwind as it falls.  However, neither mechanism is represented in the IFS and the net effect.  In reality snow spreads across a much larger distance downwind than in the raw model output.  

Fig9.6.2-1:  The diagram shows an area of NW Scandinavia with snow accumulation indicated by colours (large accumulations blues, small amounts, green).  Topographically the area is complex. The key features are steep upslopes near the exposed NW coast of Norway. There is a line of mountains reaching about 2000m interspersed with lower lying gaps.  

The top left diagram shows accumulated snow derived over a 15 day period ending 00UTC 1 May 2023.  Forecast accumulation of snowfall is mostly on the exposed NW-facing mountainous coastal areas.  But little or no snow accumulations are shown in the lee of the mountain ranges (shown by dashed line).

The central diagram shows the ECMWF analysed accumulation of snow at 00UTC 1 May 2023 which uses observations supplied by the relevant meteorological service but also uses the predicted accumulations given in the top diagram.  Thus there remains a bias towards the clearer area near the dotted line despite the observations.

The bottom diagram is the snow depth analysis by the Swedish meteorological service.  There are more observations than are shown plotted, and the rain shadow effect is not as well marked as suggested by model forecast or analyses.

Effect of open water within polar ice.

Cy50r1 in May 2026 introduced new techniques in modelling water temperature and ice concentration.  Ice cover modelled in polar waters can be broken and non-continuous, particularly around the edges of the ice area.  Leads and polynyas in polar sea ice could introduce sources of strong destabilisation (e.g. snow surface temperature -20C, sea temperature 1C).  In a very cold atmosphere the available heat and moisture could give active and possibly deep convection with large CAPE.

Unrealistic extreme convective precipitation focussed near coastlines (Cy49 and earlier)

This effect can occur with onshore cyclonic flow of maritime air that is marginally unstable to sea surface temperatures.  At certain times SSTs can be higher close to coastlines than offshore.   In these cases the inshore sea surface temperature can be high enough for the convection scheme in IFS to trigger release of convection with high CAPE values.  Just upstream the lower sea surface temperatures offshore cannot overcome convective inhibition at low or mid-tropospheric levels.    The IFS convective scheme triggers Instantaneous shower development at each step but in Cy49 and earlier does not advect the showers down wind.  This results in repeated convective rainfall over the same inshore locations which can add up to large or implausibly large record-breaking near-coast totals (Fig9.6.2-2).  Cy50r1 transfers some of the convective moisture and precipitation to the large scale precipitation calculations.  This helps carry the effects of the convective precipitation inland from windward coasts.


Fig9.6.2-2: EFI for 24h precipitation for period 00UTC 6 Nov 2023 to 00UTC 7 Nov 2023, DT 00UTC 6 Nov 2023.  The chart shows unrealistically large or extreme precipitation totals on coasts exposed to the northwest.  These are caused by instantaneous shower development by the convection scheme over inshore waters that are slightly warmer than offshore.

(FUG associated with Cy50r1)


  • No labels