You're talking about data dropouts here - not real uncommon. If you plot the data in a 3-D grid it's very plain that an occasional point is inconsistent with the remainder, and should be properly omitted.
While I've certainly seen higher quality data, the tracks you have shown us are no cause for alarm - they're merely raw data that need some screening.
