LIGO Open Science Center
LIGO
is operated by
California Institute of Technology
and
Massachusetts Institute of Technology

Plot GalleryTo make any plot shown on this page, start with the following code to read the strain data and data quality information from the LIGO data file that you downloaded in step 1 of the introductory tutorial.readligo refers to the
sample reader functions
described in step 4, which
read data from a LIGO data file:
Click on any plot to see the code used to make it.
Plot the time seriesThe strain time series, or h(t). This is the most direct plot of LIGO data. The time series is dominated by low frequency noise, which appears as smooth oscillations at around 20 Hz.Show the code FFT with Blackman WindowA Fast Fourier Transform, or FFT, is one method to transform the data into the frequency domain. Here, we first apply a Blackman Window to force the data to zero at the time boudaries, and so reduce edge effects in the FFT. The LIGO noise spectrum is seen to have the most power at low frequencies (f < 50 Hz), due mainly to seismic noise. The most sesitive band, often called "The Bucket", is roughly 100  300 Hz.Show the code Power Spectral Density (PSD)The Power Spectral Density is another representation of how the power in the data is distributed in frequency space. Matplotlib'spsd method use's Welch's method to estimate the PSD. The method
divides the data into segments with NFFT samples, computes a periodogram of each segment
(similar to the FFT plotted above), and then takes the time average of the periodograms. This averaging over
several segments reduces the variance in the result, which is why the PSD looks "thinner" than the FFT.
Show the code Amplitude Spectral Density (ASD)The ASD can be obtained by taking the square root of the PSD. This is done to give units that can be more easily compared with the time domain data or FFT. More information about the LIGO ASD can be seen on the Instrumental Lines page.Show the code SpectrogramThis spectrogram shows how the PSD of the data varies as a function of time. The main feature we see is that the low frequencies contain more power than the high frequencies, as in the above plots.Show the code Spectrogram renormalized to the median power in each frequency binIn this spectrogram, the power in each frequency bin has been normalized by the median power for that bin. This "whitens" the data. The remaining spectrogram shows how the power fluctuates in time. Such whitened spectrograms are useful for visualizing transient signals. For example, this simulated inspiral is plotted using a technique similar to a whitened spectrogram. You can also see how to visualize a Burst hardware injection in the "Find a Burst Injection" tutorial.Show the code 