The so-called last layer corrections algorithm is being upgraded. The threshold energy will be lower with limited efficiency around 100 MeV but good resolution (below 10%).
The spatial coverage on the calorimeter is extended by taking into account the variation of deposited energies as a function of postion shift. This is illustrated below for on-axis 300 MeV photons.
The previous version of this method used events well centered on the CALs and far enough from the edges (the black dot population). The coloured countours are the new areas taken into account now, adding 1 cm to the edges each time. The deposited energy in the CAL as a function of contour is shown below. One sees that the distributions can still be used as density functions (i.e. they peak, but at different values, generating a different density function when normalized for the likelihood) except for the most outlying contour which is smeared out. For on-axis events with an available track, this seems to be as well as one can do using position cuts.
The purple population has events that lie within the peak zone of the others. Could these be events that actually are within a "good" zone but reconstructed outside of it? The RMS of the reconstructed position is approx 1 cm, the FWHM being close to 4 cm (it looks energy independent from 300 MeV to 10 GeV) hence we are probably limited by the TKR recon close to the edges.
There is an important smearing for each area dependent on the conversion height in the TKR. This is illustrated below. As much as 70% can be gained at 100 MeV-ish events when density functions are made conversion-height dependent.