Yes tobias.koetter you are quite right - that is the problem. Originally that database was used on a computer with automatic DST, for 3,5 years, after that the station was put into non-DST, normal local time because of all side effects it caused. On that normal local time the timestamps were recalculated, and usually the gaps in data were filled in with linear data. Must have missed that date.
That problem now shows because some other reconstructions are done “off line” on with the database on another desktop computer with automatic DST, so then the KNIME and system found out what it ought to do in normal circumstances. I will run a test with settings using only standardtime. (Please excuse if I didn’t use correct time standard names!)
.