.. sectionauthor:: Duncan Macleod <duncan.macleod@ligo.org>

.. currentmodule:: gwpy.timeseries

Comparing the same `Channel` at different times
###############################################

I'm interested in comparing the amplitude spectrum of a channel between a
known 'good' time - where the spectrum is what we expect it to be - and a
known 'bad' time - where some excess noise appeared and the spectrum
changed appreciably.



First, we import the `TimeSeries`

.. code-block:: python

   from gwpy.timeseries import TimeSeries


And we set the times of our investigation:

.. code-block:: python

   goodtime = 1061800700
   badtime = 1061524816
   duration = 120


Next we :meth:`~TimeSeries.get` the data:

.. code-block:: python

   gooddata = TimeSeries.get('L1:PSL-ISS_PDB_OUT_DQ', goodtime, goodtime+duration)
   baddata = TimeSeries.get('L1:PSL-ISS_PDB_OUT_DQ', badtime, badtime+duration)


and calculate an `amplitude spectral density (ASD) <TimeSeries.asd>` using a 4-second Fourier transform with a 2-second overlap:

.. code-block:: python

   goodasd = gooddata.asd(4, 2)
   badasd = baddata.asd(4, 2)


Lastly, we make a plot of the data by `plotting <FrequencySeries.plot>` one `~gwpy.frequencyseries.FrequencySeries`, and then adding the second:

.. code-block:: python

   plot = badasd.plot(label='Noisy data')
   ax = plot.gca()
   ax.plot(goodasd, label='Clean data')
   ax.set_xlabel('Frequency [Hz]')
   ax.set_xlim(10, 8000)
   ax.set_ylabel(r'Noise ASD [1/$\sqrt{\mathrm{Hz}}$]')
   ax.set_ylim(1e-6, 5e-4)
   ax.grid(True, 'both', 'both')
   plot.show()

.. plot:: ../examples/frequencyseries/compare.py