What is interesting here is ...

What is interesting here is the change from 50% combined chlorine to 0% combined chlorine (all free chlorine) by the time it reaches the reservoir.  This would suggest that chlorine dosing at the WTW is not under control as others have suggested, and that there is over-dosing such that break-point chlorination is achieved.  If you are using calcium hypochlorite  as the chlorine source, there may be a possibility that it has not fully dissolved where the sample is taken at the WTW, hence the higher readings downstream.