General regional temperature biases:

Summary of 2m temperature errors:

In general, temperatures are forecast fairly well over the globe.  On average, systematic errors in forecast 2m temperatures are generally <0.5°C.  Biases in 2m temperature (verified over land) vary geographically, as well as with season, time of day and altitude.  Larger biases and errors occur over orography or in snow covered areas.  

Fig9.2.2-1: An example where ensemble forecast temperatures differ substantially from observed data.  The plot shows the Continuous Ranked Probability Score (CRPS).

Reasons for errors:

• surface temperatures can fall much lower than forecast (e.g. in this case in Sweden) where IFS has not captured the snow cover sufficiently or has retained too much cloud.
• surface temperatures can be higher than forecast (e.g. in this case in Croatia) if too little cloud is forecast implying too much nocturnal radiative cooling in the model.

Sometimes large errors will arise from a poorly forecast synoptic pattern (essentially not the case in this example).

2m Temperature:

2m temperature biases (verified over land) vary geographically, as well as with season, time of day and altitude.  Larger biases occur over orography or in snow covered areas.  

General biases for land areas: 

• In the northern hemisphere:
  • Maximum temperatures:
    • in summer generally show neutral or slight cold bias;
    • in winter generally show neutral or warm bias but
      • often a strong warm bias in Northern Europe and Russia.
  • Minimum temperatures:
    • in summer generally show warm bias but
      • cold bias in Europe, Middle East, China and SE Asia;
      • strong warm bias in Northern Europe and Russia;
    • in winter generally show warm bias but
      • cold bias in Europe and Middle East;
      •  warm bias USA southern Great Plains.
• In Europe:
  • Night-time temperatures in winter:
    
    • France and Mediterranean area generally show neutral bias;
    • Sweden, Finland, NW Russia often show a strong warm bias;
    • Rest of European area generally a cold bias.
• In the tropics:
  • Maximum and minimum temperatures throughout the year generally show cold bias but
    • winter minima in India, sub-Saharan Africa show warm bias.
• In the southern hemisphere: 
  • Maximum temperatures:
    • in summer generally show slight cold bias;
    • in winter generally show cold bias.
  • Minimum temperatures:
    • in summer generally show warm bias;
    • in winter generally show warm bias. 

These biases are not corrections to be made to every temperature forecast but merely highlight areas where errors have been identified when averaged over an extended period.  Users should consider the effects outlined below  

Diurnal range of temperatures

The amplitude of the diurnal cycle is generally underestimated over land (a deficiency shared by most forecasting models).  This is especially the case in Europe during summer when the underestimation of temperature range reaches ~2°C across large areas.  Near-surface temperatures are generally too warm during night-time and slightly too cold during the day.  However, the degree to which the amplitude of the diurnal cycle is underestimated depends on region and season.  Night-time 2m temperatures are about 1–2°C too warm and surface temperatures about 2°C too warm.  

Effects contributing to temperature biases

Near-surface temperatures are related to a variety of processes:

• cloud cover and cloud optical properties
• radiative transfer
• precipitation
• surface fluxes
• turbulent diffusion in the atmosphere
• strength of land-atmosphere coupling
• soil moisture and temperature

Some of the above processes in turn depend on land surface characteristics (vegetation, soil type, soil texture, etc.) and processes.  

Cloud cover effects

Much of the cold bias of night-time 2m temperature south of 60°N is associated with an underestimation of (low) cloudiness.  The wintertime night-time bias in Central Europe is smaller for days which are (nearly) clear-sky.   However cloud cover is not solely responsible and underestimation of cloud optical depth and/or incorrect forecast of cloud type or base height could also play a part.

Snow cover

Some warm bias in northern Scandinavia has been related to the modelling of snow.  Investigation suggests the snowpack surface does not cool rapidly enough.  There may be errors in the analysis or prediction of snow cover and depth. 

Currently snow is modelled as a single layer of snow which allows too much heat to be transferred up from the underlying ground.  A multi-layer snow scheme under development will enable more realistic heat transfer and better assessment of the albedo.  This will allow faster response to changes in the radiative forcing.  In wintertime in northernmost European countries when actual minimum temperatures are well below average (e.g. with snow cover), forecasts temperatures often are much too high (even by as much as 10°C).

Turbulent mixing

Biases in near-surface temperatures during winter conditions are very sensitive to the representation of turbulent mixing in stable boundary layers.  Comparison with radiosondes in the lower 200m of the atmosphere suggests underestimation of the temperature gradient; this is particularly pronounced at lower latitudes.  Full resolution of the details of the temperature structure in the lowest layers of the atmosphere is not possible with current computational resources.

Too much mixing increases the upward diffusion of heat,  hence reducing stability and/or temperature inversion and consequently the temperature fall at 2m and at the surface.  Errors tend to be much larger during low level inversion situations.  Note:

• • low level inversions are particularly common at high latitudes in winter
  • the closer the inversion is to the surface, the larger is the potential error.

Vegetation effects

Temperature  biases (particularly during spring and autumn) are in part related to the representation of vegetation (in terms of cover and seasonality), and evaporation over bare soil.

Orography and geographical effects

• Lake temperatures can have an effect on forecast of temperatures, particularly in deciding whether the lake is frozen or not.
• Overnight 2m temperatures tend to be too cold over rugged or mountainous areas.

Miscellaneous

• Forecast maximum 2m temperatures can be too low particularly during anomalously hot weather.
• Predicted humidity - if too low then maximum temperatures can be forecast to be too high.
• Post-processing (e.g. using a calibrated statistical technique) usually improves 2m temperature forecasts, sometimes substantially so.
• Model 2m temperature output corresponds to short grass cover (possibly snow-covered), because by meteorological convention observations are ordinarily made over such a surface.  In complex terrain - e.g. forests with clearings - this strategy may not work so well.

The forecaster should assess the potential for error due to the above factors.   He/she should:

• compare analyses of temperature, dew-point and soil moisture with observed data.
• assess future "background" conditions and any potential impacts (e.g. snowfall or cloud cover that might be different from model forecasts).

Summary of 2m dew-point and Humidity errors:

2m dew-point temperature biases (verified over land) vary geographically, as well as with season and time of day with a daytime dry (low dew-point) bias generally:

• In the northern hemisphere:  Generally dry bias both day and night and throughout the year but neutral or moist bias in Canada, Scandinavia, and Russia.
• In Europe:  Generally dry bias both day and night and throughout the year, especially during daytime in summer but with variable magnitude in winter.  
• In the tropics: generally variable throughout the year both day and night.  But a dry bias in India, sub-Saharan West Africa, and to a lesser extent Central America and parts of Brazil.
• In the southern hemisphere:  Generally dry bias both day and night and throughout the year.

Effects contributing to dew-point temperature biases

Near-surface temperatures are related to a variety of processes:

• cloud cover and cloud optical properties
• radiative transfer
• precipitation
• surface fluxes
• turbulent diffusion in the atmosphere
• strength of land-atmosphere coupling
• soil moisture and temperature

Some of the above processes in turn depend on land surface characteristics (vegetation, soil type, soil texture, etc.) and processes.  

Cloud cover effects:

Under clear-sky conditions there is generally little bias during the day, but a moist bias in the evening.  In cloudy conditions the daytime the bias is dry and is in part related to the representation of turbulent mixing, in particular in cloudy convective cases.

Turbulent Mixing:

Biases in near-surface dew-point temperatures during winter conditions are very sensitive to representation of turbulent mixing in stable boundary layers.  In the lowest 200m radiosonde data suggests the gradients are underestimated for temperature and especially humidity (giving a dry bias).  This is particularly pronounced at lower latitudes.  Full resolution of the details of the temperature structure in the lowest layers of the atmosphere is not possible with current computational resources.

Too much mixing increases the upward diffusion of heat and moisture.  This reduces the fall in temperature and dew-point at 2m and at the surface.   Biases in wind profiles in the boundary layer, and in wind direction at the surface, are related to the representation of mixing in convective boundary layers, particularly with the partition of momentum transport between dry and moist updrafts.

Vegetation, Soil moisture and Evaporation effects:

Errors in the representation of evaporation and/or soil moisture can also impact forecasts of near-surface humidity.   In particular, spring evaporation is too high, and summer vegetation gets into stress conditions too quickly (over-depletion of soil moisture).  Evaporation over bare soil is also problematic. 

Orography and Geographical effects

• Lake temperatures can have an effect on forecast of dew-point temperatures, particularly in deciding whether the lake is frozen or not.  Proximity of a lake can have an influence on the humidity at a downwind location. 

The forecaster should assess the potential for error due to the above factors.   He/she should:

• compare analyses of temperature, dew-point and soil moisture with observed data.
• assess future "background" conditions and any potential impacts (e.g. snowfall or cloud cover that might be different from model forecasts).

Summary of Soil temperature errors:

Ensemble mean values of soil moisture slightly overestimate the diurnal cycle of soil temperature:

• First (top) soil layer up to 2°C too cold at night.
• All other (lower) soil layers are always too cold.

Investigation suggests too much energy is exchanged between the atmosphere and the land.  During the night too much energy is extracted from the soil and transferred to the atmosphere. This results in:

• soil temperatures that are too cold.
• earth skin temperatures and 2m temperatures that are too warm.

Additional Sources of Information

• See the impact of Cycle 47R3 on representation of cloud and sunshine.
• Read more on Bias in IFS.