Note: HRES and Ensemble Control Forecast (ex-HRES) are scientifically, structurally and computationally identical.  With effect from Cy49r1, Ensemble Control Forecast (ex-HRES) output is equivalent to HRES output where shown in the diagrams.   At the time of the diagrams, HRES had resolution of 9km and ensemble members had a resolution of 18km.

Other causes of errors in temperature and dew point at 2m

Errors associated with thick fog.

Some errors have occurred in forecasts near-surface data associated with cases of thick fog.  A bug in IFS has misrepresented the positive feedback between two interacting and imperfectly represented mixing processes in the near surface layers in the new moist physics scheme.  The problem has been added to Known IFS forecasting issues and a fix has been prepared with implementation in the next IFS upgrade expected late in 2022.

Errors associated with soil moisture.

Impact of heat and moisture fluxes

Errors in the analysis of heat and moisture fluxes from the underlying ground have an important impact on the model surface temperature and moisture values and hence the derived 2m screen temperatures.  Fig9.2.1.3-1 & Fig9.2.1.3-2 illustrate the problem.   Low-level moisture can impact upon temperature forecasts; if humidity is too low then maximum temperatures can be forecast to be too high (e.g. East England and Germany).

Land surface characteristics (soil moisture, leaf area index) have an impact upon temperature forecasts.    Significant differences in temperature can occur over a short distance where there is a sharp change of surface characteristics.   This can influence the location and development of subsequent convection.

Users should assess the analysis of temperature and moisture against observations in the area of interest and modify forecasts accordingly.


 

  

Fig9.2.1.3-1: An example of incorrect assessment of heat and moisture fluxes (left, temperatures; right, dew points), at Cordoba 12 June 2017.   Ensemble Control Forecast (ex-HRES) forecast temperatures and dew points (red) and observed temperatures and dew points (black).  Ensemble Control (ex-HRES) forecast has under-estimated the maximum temperatures by some 3ºC.  

The left panel shows that during this very hot spell the maximum temperature, on 12th, was under-predicted by 3ºC. This may be due to unrepresented local factors, such as urbanisation, though on the other hand the signal is also typical of what we often see during extreme summer heatwaves.  This bias is a subject of current research; it may be symptomatic of an IFS inability to generate the super-adiabatic near surface layers that one sometimes sees on radiosonde ascents.

The right panel shows that on this occasion the magnitude of the dew point errors was even larger overall.   Again there are many possible reasons, but one candidate would be mishandling of moisture fluxes to/from the surface.  In turn these could relate to soil moisture errors, or errors in handling the biology of evapotranspiration.  

An influx of moist low-level air might also occur locally (e.g. effects of a strong sea breeze).  This can influence the location and development of subsequent convection.


Fig9.2.1.3-2: Soil moisture 00Z 11 June 2017.  It is possible that there was too much moisture in the soil (yellow) when more arid conditions (brown) would have been more appropriate.  This is suggested by the observed lower dew points during the day on 12th June in Fig9.2.1.3-1.  Dew point errors are more likely to be indicative of soil moisture errors during the day, because there is much more convective overturning then. Conversely night-time dew point errors could be much more a function of very local effects - e.g. proximity of a lake or river.



Fig9.2.1.3-3: An example of possible connection between IFS forecast 2m temperature (here using forecast EFI as a proxy) and forecast soil moisture content.  EFI lower than -0.8 suggests extreme low temperatures compared to the model M-Climate and appear to be associated with high soil moisture content.  During the 10 day period, soil is forecast to dry out and the 2m temperature is forecast to become unexceptional.

The reasons are not clear why IFS shows temperature and moisture differences and uncertainties in analysis and forecasts.  Possible causes and correlations:

  • There is a dry bias during the dry season and first guess 2m temperatures can be too cold.  During the analysis process, soil moisture increments attempt to better represent 2m humidity.  These increments can be substantial.  However, this results in excessive soil moisture near the surface despite no precipitation.  This moisture in IFS can then incorrectly percolate to lower layers or even runoff on the surface.  In contrast, the increase in soil moisture from actual rainfall is normally realistic and much less.   Increments can override the impact of any noteworthy rainfall.
  • Surface ground moisture limits the amount of diurnal surface warming and night time cooling leading to lower 2m temperature and higher 2m relative humidity being observed and forecast.
  • Vegetation could be lush and green after previous wet period and/or evapotranspiration could be mis-represented.
  • Errors in forecast advection of boundary layer air.  Diurnal influxes of moist sea air may replenish moisture and/or not be well forecast either in depth or penetration.
  • Observation representativeness.  Insufficient information on atmospheric structure due to the lack of radiosonde. 


 

Fig9.2.1.3-4: Mean soil moisture analysis increments for the top soil layers over the winter period 2024/25.  Systematic increments during the analysis process are:

  • positive, implying an increase in IFS model soil moisture, over central and Southern Africa, India and SE Asia (but notably not central South America).
  • negative, implying a decrease in IFS model soil moisture, over Indonesia and central South America.

The increments shown are only a guide and do not imply that they occur in every case. 

Summary of soil temperature errors:

Soil moisture and temperature is modelled in four soil levels but there is a considerable lack of real-time observations of soil condition and moisture content.  Nevertheless heat and moisture fluxes have an impact on model surface and 2m temperature and moisture. 

The 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.

Flooding may occur after heavy or prolonged rainfall but will not be modelled.  Incorrect soil characteristics and/or a water surface will cause errors in the forecast low-level temperatures.

Suggested considerations to offset soil temperature and moisture errors

The forecaster should assess the potential for error due to the above factors by:

  • comparing analyses of temperature, dew point and soil moisture with observed data.
  • assessing soil moisture and whether any increase is due to forecast or observed rain, or perhaps due to model soil moisture increments.
  • assessing future "background" conditions and the potential impacts thereof (e.g. snowfall or cloud cover that might be different from atmospheric model predictions).


Users should recognise the impact that low-level moisture has upon temperature forecasts; if humidity is too low then maximum temperatures can be forecast to be too high.  ECMWF is currently investigating problems with soil moisture increments.

See also Section 2.1.4.5 Modelling soil structure.

(FUG Associated with Cy49r1)

  • No labels