A few weeks ago, the StEEmcTreeMaker was changed so that if no TPC vertex was availabe, the BBC time bin was used to determine the vertex Z positon (with X and Y set to 0).  The values for the Z position per time bin was set based on the average value of the vertex Z position per time bin (when TPC vertex was available), using the one file

st_physics_7160005_raw_1030001.MuDst.root

Specifically, a histogram of TPC vertex Z position vs time bin was make, and then "profiled" in root to see the mean and st. dev. per bin.

To confirm whether the values are appropriate, the same process is applied to the stored vertex and time bin information in the produced EEmcTrees, using all available (343) EEmcTrees for 2006.  There were 2,340,147 events with a primary vertex defined and 342,759 with no vertex defined (13% of the total).  Figure 1 show the distribution of events in the BBC time bins for events where there is no primary vertex defined.  Figure 2 shows the average position of the Z vertex vs. time bin (the ROOT "profile") as well as the hard coded values currently in the StEEmcTreeMaker.

Figure 1: Counts per BBC time bin when no primary vertex is available

Figure 2: Average position of vertex Z per time bin (blue lines) and the current hardcoded values in the StEEmcTreeMaker (red markers).

Conclusions

I can make a plot using more of the MuDst files, to see if the hardcoded values are off because of the limited statistics.  However, it is likely that the difference is due to the trigger cut already being present in the EEmcTrees.  In this case, we will have to decide which set of values is more appropriate.