Spikes' Remotion

Once the light curves have been background subtracted according to the procedure described above, a technique aimed to identify phosphorescence spikes applies to the light curves: the purpose is to avoid possible biases to the estimates of the GRB fluences for the different GRBM units.

Figure: Background subtracted light curves of GRB000830, UT 11:39:08. From top to bottom: GRBM1, ..., GRBM4, AC1, ..., AC4.

The identification of a spike is based on the following criterium: when the net signal, i.e. the background subtracted counts for a given 1 s bin, is at least 5$\sigma$ significant in the GRBM band (40-700 keV), and the corresponding AC ($>$ 100 keV) background subtracted counts are lower than $0.3$ times the GRBM counts (see the HR condition expressed by eq. ), the GRBM counts are thought to be due to a phosphorescence spike, and not to the GRB. Whenever this condition triggers a spike in the GRBM band light curve, it changes the corresponding background subtracted light curve, by replacing the spike counts with the average value of the two adjacent 1 s bins.

A nice example of this spikes' replacement procedure is shown in the above example of GRB000830: from fig. , in GRBM unit 4 there is a spike preceeding the GRB by $\sim 30$ s, whose SNR is around 60$\sigma$, while in the AC 4 light curve no corresponding significant signal is visible. According to this procedure, the spike disappears in the corrected light curves shown in fig. .

When the HTR light curves are also available, the background fit results obtained from 1 s light curves are assumed, to subtract the background level from the 7.8125 ms light curves.