Hi @julianslipk,
This is a methodology problem rather independent of the use of LAMMPS. Do not hesitate to discuss this with people around you.
There are issues to be careful with when using GK methods, mainly convergence. So there are numerous papers in the literature dealing with error estimations of the measured values. A small search made this one pop up in my browser but this is not the first I see. It might be a first lead to help you find something that suits your needs.
As a rule of thumb, you can also make several independent simulations and compute an error-bar as with any measurement.