Last modified: 21 February 2024

How can I get the source variability [index] from the data?

In the CSC 2.1 we compute two types of variability:

  • Intra-observation variability refers to the probability that a source is variable in a timescale comparable to the duration of the Chandra observation (typically a few hours), and is measured by looking at the arrival times of the source events to see if they are consistent with a constant flux. The variability index (var_intra_index_〈band〉) in this case is computed from the Gregory-Loredo variability probabilities and the fraction of the light curve points that are outside the 3σ and 5σ levels with respect to the light curve mean, as explained in Gregory-Loredo Variability Probability: Assigning the Variability Index.
  • Inter-observation variability refers to the probability that a source is variable on timescales comparable to the timespan between different Chandra observations of that same source (months/years), and is computed by looking at the aperture photometry fluxes (and their confidence intervals) for the different observations in which the source is detected (or not). The variability index (var_inter_hard_〈band〉) is estimated based on a likelihood ratio between the null hypothesis (no variability) and a model that assumes variability, applied to the fluxes and their errors on each observation, as explained in Source Variability page.

For both types of variability, we compute variability indices on a per-energy band basis. So, for example, for the hard band (2–7 keV), the relevant catalog columns at the master source level are var_intra_index_h and var_inter_index_b. Please note that you can also obtain the intra index at the individual observation level (i.e. considering a single observation of a given source), and in that case the quantity would be called var_index_h for the hard band.

These columns, as well as the light curve data products (that refer to individual observations of a source, and therefore to var_intra quantities) can be obtained from our CSCview interface. In the lc3.fits data product, the variability index may be found in the VARINDEX header keyword.

In addition, if you prefer to access the tables and the data products from a Python environment, that is also possible using our TAP and Virtual Observatory-compatible services. Here is a basic tutorial on how to access catalog columns and data products in Python that will soon be posted in our documentation pages: CSC 2: Accessing release 2.0 of the Chandra Source Catalog with PyVO and CIAO Tools.

A final note: it is good to keep in mind that light curves should not be over-interpreted, especially in the case of low count sources. Because of the way they are produced using the Gregory-Loredo algorithm, some of them might show variability spikes that are not consistent across bands and are due to the smoothing of individual photon arrivals. That is why we also provide additional information in the catalog columns, such as the variability probability, the variability indices, etc. If those quantities indicate that the source is not variable but you still see a variable light curve, that might just be a processing artifact, rather than actually variability. On the other hand, if the variability index and probability are relatively high, you can trust more the structure seen in the light curve, especially if you have more than a few tens of counts in your source.