A span of about four weeks centred on the mid-point of the M-climate period is used constructing the model climate.  This means that during:

  • winter and summer there are usually only small seasonal changes.  The range of maximum and minimum temperatures is less than around the equinoxes.  The M-climate value is rather more representative of the mid-point of the period. 
  • spring and autumn there are rather larger seasonal changes.  The range of maximum and minimum temperatures is greater than around the solstices (e.g. minima during the period late April to late May are included in the calculation of the M-Climate for early May).  The M-climate value is rather less representative of the mid-point of the period.

There is of course continuing day-to-day variation in minimum temperatures.

The effect during spring and autumn is to extend the tails of the M-Climate (as seen on Cumulative Distribution Function (CDF) diagrams) towards warmer and colder values - i.e. increase the spread beyond what it should be.  An investigation showed that the cold tail should change by about 1ºC per week, but of course the magnitude of this will depend very much on location.  It seems from case studies that the impact upon CDFs and EFIs for maximum temperatures is rather less pronounced than for minimum temperatures.  

The tails have a disproportionate influence on the Extreme Forecast Index (EFI) and for minima at least the extra, anomalous spread seen in the transition seasons can lead to a reduction in EFI values by ~0.2 from what they would be if the M-Climate had been evaluated over a shorter period.  The reason we don't evaluate over a shorter period is that the sample size would reduce, so we would be replacing an occasional over-spreading issue with a continuous under-sampling issue, which would make the EFI even less reliable.

In summary, in extreme situations during the transition seasons, cold EFIs in particular will typically be damped to be of smaller magnitude than they should be.  This is because the M-Climate range is artificially extended in those seasons.  Users should expect equivalent adverse impacts on the SOT also, which by definition focuses on the distribution tails where the issue lies.


Fig8.1.9.5-1: EFI forecast for 2m minimum temperature T+24 VT 14 May 2020, DT 00Z 14 May 2020.  Minimum temperatures over SE England between 0°C and -2°C observed widely.


Fig8.1.9.5-2: Central England Minimum temperatures (source Met Office) highlighting the changes in minimum temperatures through the year.  The black bars show the four-week period used to compute M-Climate minimum temperatures for the central point of the bar.  In practice re-forecasts, which will also show the seasonal variations in temperatures, are used to define the M-climate.  

On average, during the periods used for calculation of M-Climate (approx 4 weeks):

  • in spring and autumn, minimum temperatures rise/fall by around 4°C,
  • in summer and winter, minimum temperatures remain relatively constant.

A similar effect may be seen regarding maximum temperatures but changes during the spring/autumn calculation periods are less pronounced than for minimum temperatures.

One must expect differences from this general picture in other regions - e.g. in Siberia, which has a huge annual temperature range, the transition season changes would be much larger.