There is no way to "fix" this. All we have to work with is the data received from the boat. In this cases it appears that DE send an incorrect position on 6/30 at 4:49AM. The following position at 8:48 having the correct position - hence the variation in the reported COG and SOG around those two points.