Yeah, it amazes me how often I see the cloud cover myth mentioned...
Clouds specifically don't affect GPS accuracy that much, but the entire atmosphere the signal is passing through does. An overhead satillite's signal travels through less error inducing atmosphere than one near the horizon. This is the main reason that even the miliary and land surveyors can only get +/- 5 metre accuracy with only a single GPS receiver, so 20' accuracy is indeed good. To get results better than that, either the readings need to be taken for long periods of time and then averaged (Static GPS) or be compared with another independent GPS unit measurement from a known point to compute a correction vector (RTK or Post processing). With either of these methods (and big $$$) the accuarcy is down to a couple centimetres.
