Page History
...
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.
Diurnal temperature changes are strongly influenced by incoming and outgoing heat flux which itself is governed by the extent and thickness of cloud cover. Uncertainty in the model analysis of cloud can have a strong impact on forecast errors.
Most Most of the large errors seem to occur when the surface temperature is very cold, and the lowest levels may levels may become extremely stable. In such very stable air tiny amounts of energy can correspond to large temperature changes at the surface because surface because there is no convection to mix energy through the lower atmosphere. This is the main physical reason for large errors being relatively commonplace in such circumstances. Temperature errors often don’t depend strongly on the forecast range.
Effects contributing to temperature errors
Near-surface temperatures are related to a variety of processes:
The near-surface inversion is likely to be most influential and errors more likely with high pressure and calm conditions. It is vital to compare the observed and forecast thickness and extent of low cloud and the temperature and humidity structure of the lowest atmosphere.
Effects contributing to temperature errors
Near-surface temperatures are related to a variety of processes:
- cloud cover and cloud cloud cover and cloud optical properties
- albedo and radiative transfer
- precipitation
- surface fluxes
- turbulent diffusion in the atmosphere
- strength of land-atmosphere coupling
- soil moisture and temperature
...
- cloud cover (from observations or satellite pictures) with cloud analyses or forecasts.
- cloud structure (from observed and background vertical profiles).
...
Fig9.2.1-1: Examples of the difficulty in forecasting temperature in lowest layers. At Kryvyi, although it is cloudy in reality yet the observed temperature is still a lot colder than forecast. The forecast boundary layer temperature is too warm (by ~5°C) and the cloud cover is not represented. At Lulea the inversion top is not well captured, the moisture (in relative humidity terms) is not well portrayed, and the surface temperature is too warm (by ~5°C) - but had more cooling been forecast near the surface then the very lowest layers would have been correctly captured.
Fig9.2.1-2: Small errors in the extent of low cloud and fog can have large impact on forecast 2m temperatures. In the example positive temperature errors (model too warm) are over parts of Spain dominated by fog and low clouds. The contrast of temperatures between these foggy ares and the rest of the country are quite large – subfreezing temperatures in some areas with freezing fog in contrast to +12°C outside of the fog. Errors are relatively large.
The example illustrates how small errors in the extent and thickness of model forecast low cloud and fog can have large impact on forecast 2m temperatures. Over Spain, 12UTC temperatures approached 12°C in areas where low cloud broke and cleared but remained cold where low cloud and fog persisted (subfreezing in some areas of freezing fog). The model forecast low cloud to be variable in thickness and rather greater in extent than in reality.
Thus over Spain at 12UTC there were:
- relatively large positive temperature errors (model forecast too warm, yellow areas) where fog and low clouds broke and cleared in the model but persisted in reality. Too much cloud implies less insolation and persisting cold temperatures.
- relatively large negative temperature errors (model forecast too cold, blue areas) where fog and low clouds broke and cleared in reality but persisted in the model. Breaks in cloud cover allows temperatures to rise.
The forecast vertical profiles show that Low clouds are forecast to be more widespread than in reality. Observations show fog in some areas at 12 UTC. The forecast vertical profiles shows the model was fairly good at representing temperature inversions. But However, the model locally missed clouds and fog by a small margins in places. Small errors can have Such small errors had a large impact on surface temperatures. Too much cloud implies less insolation and persisting cold temperatures. Where the model breaks the inversion with consequent breaks in cloud cover allows temperatures to rise. The warm area in central northern Spain associated with high ground above the overnight inversion.
Snow cover effects
The analysis and forecast of snow depth, snow compaction, and snow cover are important for forecasting lower-layer and near-surface temperatures and they can have a significant impact on forecast accuracy. However, the relationship between snow cover and temperature is complex:
...
- its age (the model facilitates slow, natural compression),
- melting (small amounts of snow on the ground tend to take too long to melt, even if the temperature of the overlying air is well above 0°C).
- interception (of rain)
- addition of new snow.
The The albedo is related to the extent (and age) of snow cover and snow characteristics in analysed and forecast fields have an effect on the radiation that could be absorbed. This has a corresponding impact on forecasts of 2m and surface temperatures. Better assessment of the albedo when the multi-layer snow scheme is introduced will allow faster response to changes in the radiative forcing.
...
Currently snow is modelled by a a multi-layer snow scheme allowing a fairly realistic heat transfer.
(Note: Previous to June 2023 (Cy47r3 and earlier), only a single layer snow model was available. There was no mechanism to deal with density variations in the vertical within the snowpack. This had an impact on energy fluxes which in turn had potential to adversely affect the forecasts of 2m temperature. For example, when new low density snow falls onto old dense snow, the atmosphere might be "re-insulated" from a ground heat source, allowing 2m temperatures to drop lower in reality than in the model. In practice this particular problem will be exaggerated by temperature sensors ending up closer to the snow surface when snow has fallen (assuming they are not elevated manually)).
Fig9.2.1-32: The snow depth in the vicinity of Murmansk is shown as a shade of green (5-10cm). A snow depth of 10cm (actual snow depth, not water equivalent ) is the threshold for the IFS to assume the entire grid box fully snow covered (snow cover fraction = 1 ). Thus a difference around this threshold value can change the tile partitioning and thus snow coverage may not be uniform or continuous over the grid box. The snow-free tiles would have less insulation from the soil underneath so maintaining the average skin temperature to higher temperature compared to a fully snow-covered grid box. This can potentially impact the 2-metre temperature computation.
...
Occasionally lower tropospheric temperature data has been given low weight during the analysis process. Usually this relates to problems with assimilating the boundary layer structure in situations with a strong inversion, coupled with the fact that the background is a long way from the truth. The analysis procedures tend to give lower weight to observations that show major departures from the first guess and, particularly if there is little support from adjacent observations, such data can even be rejected completely. In consequence, the analysed temperature structure of the boundary layer may only move a small way towards correcting errors in the background (Fig9.2.1-4). From a mathematical standpoint it is also (unfortunately!) more difficult to correctly assimilate data near the surface than data higher up.
Fig9.2.1-43: Examples of the difficulty of assimilating temperature and humidity data in the lowest layers. At
- At Kryvyi the first guess (blue) is too warm, and also too dry (in relative humidity terms). The analysed structure (red) after assimilation of the observed data (black) is slightly less warm and has captured saturation within the inversion base but remains
...
- generally drier (in relative humidity terms) and warmer in the boundary layer. The forecast boundary layer temperature is too warm (by ~5°C) and the cloud cover is not represented.
- At Lulea the analysed temperature structure remains similar to the first guess (blue) despite the observed much colder near-surface temperature
...
- and warmer inversion top. The inversion top is not well captured, the moisture (in relative humidity terms) is not well portrayed, and the surface temperature is too warm (by ~5°C) - but had more cooling been forecast near the surface then the very lowest layers would have been correctly captured.
Differences Differences between observed and first guess values such as these may lead to very low weight being given to the observation, or to it even being rejected. In In many cases the analysed temperatures remain similar to first guess values despite the observations. Users Users beware!
Miscellaneous
- Forecast maximum 2m temperatures can be too low particularly during anomalously hot weather.
- If the predicted humidity is 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.
...
Overview
Content Tools