Page History

...

• Temperature
  • Inversions and low cloud can be poorly represented, notably in winter or high pressure areas. This has an impact on fog extent and persistence and also 2m temperatures. 
  • Supercooled liquid at  top of boundary layer cloud.  The limited vertical and horizontal resolution and uncertainties in the parameterization of physical processes near the cloud top leads to an overall reduction in supercooled liquid water with a detrimental impact on shortwave and longwave radiative fluxes, and increased 2m temperature errors over land.
  • See also the summary of 2m temperature errors and regional temperature biases.
• Precipitation
  • Investigations over USA have shown from April to September, an underestimation of precipitation occurs in the first half of the night while an overestimation takes place during the morning, both peaking in mid-summer.  Systematic biases are less pronounced during the rest of the day.
  • Sub-grid variability gives apparent biases of over-prediction of small totals, and under-prediction of large totals.  This is most seen with convective precipitation which has greater sub-grid variability than large scale precipitation.
• Snow
  • Snowfall in marginal situations and winter precipitation type.  The elevation, thickness and extent of the melting layer can be incorrectly modelled. This is particularly important where precipitation falls through it.
  • Surface fresh snow often takes too long to melt although ground temperatures are forecast above 0°C.
• Convection
  • Substantial convective outbreaks can be under-represented by HRES at higher latitudes and over flatter areas in Europe.
  • Sub-grid variability gives apparent biases of over-prediction of small totals, and under-prediction of large totals.  This is most apparent for convective precipitation which has greater sub-grid variability than large scale precipitation.
  • Inland penetration of showers triggered by sea-surface temperatures has improved but remains insufficient.
  • Convection can be over-represented in mountainous areas.
• Wind
  • There can be a bias towards under-prediction of wind strength at mountain sites.
  • Windstorms related to mountain waves can be completely missed by HRES (although captured by a Local Area Model (LAM)).
• Fog
  • Correct prediction of fog dispersal is notoriously difficult, because situations tend to be finely balanced.  The representation of near-surface inversions can be incorrect due to difficulties with energy coupling with ground and lakes.
• Monsoon
  • Indian and SE Asian Monsoons are too strong.

...