wrong end time when querying long traces

[niyiyu]

This is a known issue from obspy. When obspy reads an miniSEED stream, it reads only the start time, sampling rate and the number of samples. Then obspy will calculate the end time of the trace on-the-fly. Usually, the sampling rate is not accurate enough to have a perfectly accurate end time, meaning that the waveform will shrink or expand. This is negligible for short trace, e.g., 30 minutes for 1 data points for traces@100Hz. However, when you query a day-long trace, the misfit could be up to ~50 samples, which is about 0.5 seconds. The solution is querying a shorter stream when use obspy.read(), which is not implement by pnwstore yet.

[niyiyu]

It turns out that obspy bypassed this bug through passing a start/end time to obspy.read. Thus it is recommended to read a shorter trace at a time by passing a start/end time, instead of a day-long file. This also save bandwidth of data streaming from pnwstore1. If you are indeed need to read day-long mSEED file, be aware of such feature of obspy.

See commit d9b8468.