Comments about METRo's performance (METRo)

Français

Seth: Write your stuff here.

Here are some findings from some sensitivity tests I did to figure what has the biggest influence on the METRo forecast. This was related to the bug I submitted about METRo quickly diverging from road-T observations in the obs history and subsuquently these rapid jumps in temperature cause inconsistencies in the forecast time series from run to run.

Modifying the forecast cloud-cover values made very little difference.

Longer obs-history makes almost no difference.

Colder air-T obs make the forecast only slightly colder.

Colder air-T and road-T obs make the forecast quite a bit colder...

The road-T obs in the obs-history have a much bigger influence on the forecast than any of the other variables.

This indicates that having real road-T obs in the obs-history is important...and you can produce better forecasts if you have recent road-T obs in the obs-history.