spike correction should be ADC subtraction instead of the interpolation
SeiyaNozaki opened this issue · comments
Just let me report:
If predicted spikes are located at the pulse peak position, spike correction reduces the pulse height.
As @tysaito2008 explained in the reco meeting, Tomohiko @toka2018 is studying the spike feature and implementing the new spike correction (ADC subtraction) into lstchain.
Ok, keep in mind that this should also go into ctapipe_io_lst, so that it works for the DL1 stage tool of ctapipe which we will eventually use as the LST standard.
Implemented now in #96 together with the PR for the new drs4 pedestal tool that also calculates spike heights
Additionally to the reported case of lowering the peak like here:
It can also increase the peak, like here:
