I am now trying to find a better way to reconstruct geometry and energy of EHE events for 22 strings analysis.

Here, I tried to reconstruct events by using channel-wise infomation rather than using event-wise information such as total (event-wise) Npe.

I hacked a bit again the waveform reconstruction codes, and change it to be able to rebin easily and handle the saturation level correctly. The easy rebinning functionality make us possible to rebin into 1 bin, and this is actually what I am mostly interested in due to the robastness for 22 strings analysis. This is I think similar to what muon-llh does. The difference is we use an assumption of EHE event track such as muons. I also checked the possibility to use the all waveform information, which seems to be a bit too ambisious.

- 1bin analysis (channel-wise Npe)
- rebin16 (16 bins are rebinned)
- rebin4 (4 bins are rebinned. nearly using all waveform info)

- 1bin analysis (channel-wise Npe)
- rebin16 (16 bins are rebinned)
- rebin4 (4 bins are rebinned. nearly using all waveform)

- The geometry reconstruction by hyper reco Thanks to Henrike;)
- The geometry reconstruction by channel-wise Npe reco
- The total Npe based energy reconstruction

- By using the channel-wise Npe, the energy reconstruction seems to work reasonably.
- There is a linear relation between the reconstruction and the true energy.
- The lightsaber tables with FADC gives the best results, and the energy resolution is about factor 6 (factor 4 (dist<500m)).
- Although, there is a systematic offset between true energy and the reconstructed one.
- The energy resolution by the channel-wise Npe is about factor 2 better than the one derived from total Npe.
- All the waveform information doesn't seem to improve the event reconstruction so significantly.
- I hope we can improve the energy resolution for far away events by improving the geometry reconstruction. But, the geometry reconstruction by the channel-wise Npe reco doesn't help...

Keiichi Mase Last modified: Fri Mar 14 21:17:55 JST 2008