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.