Run 9 TPC alignment artifacts

In trying to help Grant and Na complete the TPC alignment work for Run 9, I've observed some troublesome artifacts in the along-the-padrow residuals. I provide here two examples, looking at residuals [cm] vs. padrow for sectors 1 (left) and 24 (right). Note that not all sectors look like this, but these two are demonstratively odd.

These plots are generated with very large transverse inner sector hit position errors used in tracking. In other words, the hits on padrows 1-13 should not influence the track fit. This is done to determine alignment, and usually results in a flat (at 0) distribution of residuals on padrows 14-45, and linear distribution of residuals on padrows 1-13, whose slope and intercept are related to the alignment rotation and shift (which are applied to the outer sector).

However, the above plots show a large wiggle over padrows 1-13 (on the order of +/-100 microns). This does not correspond to a misalignment between the inner and outer sectors. When averaging over all sectors, I get the following plot of residuals [cm] vs. padrow vs. Z [cm]:

So there appears to be some Z dependence (these plots are made from Run 10118050, which was a dedicated low luminosity run, and I have made some effort to remove any possible remaining pile-up). Which brings up the possibility of SpaceCharge+GridLeak...

Using Hao Qiu's preliminary SC+GL calibrations, I find the distortion correction maps for SpaceCharge, GridLeak, and SpaceCharge+GridLeak look like this in [cm] vs. radius [cm] vs. Z [cm]:

Here we see that the GridLeak correction introduces a maximum of something close to 30 microns of shift between padrows 13 and 14 (and probably close to 15-20 microns on average). SpaceCharge makes the inner padrows dip in a systematic way, but does not introduce a wiggle.

So, at this point, I do not understand the features I am seeing in the residuals. Any ideas are welcome...