Prior work on energy calibration of the Radphi lead-glass calorimeter
showed that the total light yield in an electromagnetic shower rises
somewhat faster than linearly with shower energy. This has been
explained as a consequence of the dependence of the average shower
depth on energy, coupled with light attenuation in the glass block.
Energy loss through leakage out the back of the block is also present,
but would cause the light yield to rise slower than linearly with
shower energy. At angles below 20
where showers see over
45cm of glass, it is plausible that the attenuation effect would win out
over leakage. But above 20
(almost 50% of the solid angle
subtended by the LGD is above 20
) the effective thickness
of the glass decreases, and somewhere one would expect the sign of the
nonlinear effect to reverse. Thus the nonlinear energy correction
must be a strong function of the polar angle of the shower. The same
is true of the
-centroid of the shower in the glass. This
report gives the results of a detailed Monte Carlo simulation that
reproduces the known nonlinear behavior of the LGD for normal-incidence
showers and provides the information required to carry out this
correction at larger angles where the shower leakage is substantial.