Add one hour to hourly knmi-data

[rubencalje]

The returned file for hourly knmi data states about the hours in hourly data:
# HH : tijd (HH=uur; UT.12 UT=13 MET; 14 MEZT. Uurvak 05 loopt van 04.00 UT tot 5.00 UT

So the timestamp is allready at the end of the period that the measurement describes (what we want right?).

However, the time is in UT. For daily data we set the index to standard time in the Netherlands with the following code:

# from UT to UT+1 (standard-time in the Netherlands)
df.index = df.index + pd.to_timedelta(1, unit="h")

However, we do not do this for hourly data, which I think we should.

[HMEUW]

This issue has my interest, because I have to work with timeseries of groundwater observations that are affected by the astronomical tide. Therefore, I am trying/working to apply a timezone to all timeseries.

My suggestion for the KNMI data is not to apply a timedelta on the index, but assign the appropriate timezone of the dataset.

[rubencalje]

That is a nice solution as well, although I tend to have issues with a timezone-aware DatetimeIndex. For example, comparing a timezone-aware Timestamp with a timezone-unaware Timestamp gives an error.

The important thing is that we do the same thing everywhere in hydropandas. If I am right, up until now we convert all timestamps to Dutch standard time (winter time). And for measurements that describe an intensity over a specific period, we set the index at the end of the period that the measurement describes.

[HMEUW]

You are right. Timezone-aware and -naive Timestamps gives an error. Even a 'simple' tmin='1990-10-12', does not mix with a timezone-aware DateTime. Therefore, I propose that Timezone-aware should not be the standard in hydropandas.

On the other hand, hard coded references to Dutch common uses are not very friendly for the international expansion plan.