I'm reading a paper about a recent x-ray burst from a suspected magnetar (A. Dai et al 2016) where they show a light curve of a burst that lasted about 10ms. (Their figure 1). The value of count rate for the peak of the burst from the graph is $6\text{ counts/s}$ over a timespan of less than $0.01s$. This corresponds to about $6*0.01=0.06 \text{ counts}$ for that time period. What does this mean? Is it not counts of photons? What is 0.06 photons? Is there a step in the data analysis I'm missing?

Swift/BAT mask-weighted light curves in different energy bands (Fig. 1 Swift/BAT mask-weighted light curves in different energy bands)

  • $\begingroup$ An average rate in counts per second hardly needs to be integral! $\endgroup$ – Carl Witthoft Oct 3 '16 at 12:38
  • $\begingroup$ @CarlWitthoft I don't follow, how do you mean? $\endgroup$ – zeitoon Oct 13 '16 at 19:52
  • $\begingroup$ Try it: take 100 random integer samples and calculate the mean value. Odds are it'll be a non-integer value. $\endgroup$ – Carl Witthoft Oct 13 '16 at 19:53
  • $\begingroup$ Oh yes, I understand that, but my problem with 0.06 counts isn't that it's non-integer but that it's less than 1. If less than a single photon was detected in some 2ms time bin what was actually detected? Rob's answer below helped me answer that. When you said "integral" I thought "sum over total time span" rather than "integer" $\endgroup$ – zeitoon Oct 13 '16 at 19:55

I think there is a missing piece of information. The BAT is a coded mask telescope. The imaging is done by photons passing through a mask and falling onto an array of 32768 detectors. http://swift.gsfc.nasa.gov/about_swift/bat_desc.html

The "mask-weighted" light curve is produced after a complex ray tracing exercise using an estimate of the position of the source. Looking at some of the software specifications (eg https://heasarc.gsfc.nasa.gov/ftools/caldb/help/batmasktaglc.html ) it seems that the count rate is divided by the number of active detectors.

So in this case it looks like the detection rate was more like 200,000 counts/s in the peak 2ms of the event. There would be 400 detected counts, assuming there are no other corrections for sensitivity or vignetting. However, I think there probably are - the mask must take out about 50% of these counts and there are gaps between the detectors. A background rate has also been subtracted. So overall I guess that the first point in the burst is due to around 100 detected photons, and so on.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.