How to check numerical noise in MWS?

Well, I got a simple idea for this check. Send an x-polarized plane wave and put 3 probes at a distant point: one of the probes oriented along x-direction, the other oriented along y-direction and the last one oriented along z-direction. Normally one shouldn't observe any electric field along y and z-directions. The amount you observe in the simulation should be due to numerical noise. I wonder if this idea is correct. I would like to take your opinion on this matter.

the problem you proposed above would test a limited number of features such as the modeling of plane wave source, absorbing boundary etc. Any conclusion you draw regarding the accuracy is for this kind of problem only. It is unlikely that you could generalize the conclusion to other problems, thus you should not treat it as a "numerical noise in MWS".