There are a number of checks in the process of determining clear and cloudy fields of view.
One such check looks at how much colder the individual fov's are than the warmest band 5 value within the processing box.
As can be seen by looking at the 'old' derived product image of total precipitable water, this test was much too tight (too many cloudy boxes in clear regions).
Relaxing this test to 3 times the noise results in a great reduction of questionable cloudy boxes.
