With regard to HEATHROW DIRECTOR's comment on temperature forecasts:
- The source of weather information for the mass media may vary indeed. The quality of their weather forecasts will depend on the quality of their raw data and even more on the compentence of their in house staff to process the information and prepare it for broadcast.
- Often the media don't allow enough time for a detailed forecast in their weather transmissions.
- Making good temperature forecasts is not a trivial task. You either need a lot of local experience in a forecaster, which is expensive as any manpower is.
The output of the numerical models is of limited use, as, see my above posting, the boundary layer near the surface is not simulated with sufficient accuracy.
However a strong combination is to make a statistical approach to the numerical data. So you need to build a lot of preferredly fully automatic weather stations and make a statistical approach to their temperature measurements with regard to the various weather patterns.
After a couple years the system can say 'This particular weather pattern will probably yield this temperature at this particular location'
That is an automated way of gaining 'local experience'. A private company here in Germany is doing this, and from what I can tell with great success.