computed vs. measured s11

I attach a picture of the s11 I got for my TEM cell [ ../imgqa/eboard/EM/EM-s5tge5uqggc.jpg ]
I have two questions:
1. why the difference between minimas of s11 curves (measured and simulated) is increasing with frequency?
2. Why does s11 varies like that as a function of frequency? I mean why do we have minimas and maximas (and they do not even repeat periodically with frequency; for instance, in computed curve, I have a minima at about 0.2 GHz, the next one is at about 0.55 GHz, the 3rd one is at about 0.85 GHz)

those two graphs look pretty similar to me. What were you expecting, 100% identical prediction vs measured? Join the real world!

- - - Updated - - -

of course, not identical, but a bit more closely. One more question: the theory says that I should have resonant frequencies of TE10 mode at about 700 and 970 MHz (I'm not interested in what is above). I found them in the measured and also in computed values. Yet, also in measured and computed, there are two more resonances which I do not know where do they come from. Any idea?

Above 9 GHz, there are just a whole bunch of non-tem waveguied modes going on. But it looks like there are some lower frequency narrowband waveguide modes too. I would get a piece of eccosorb and place it in various places in the TEM cell to see what orientation makes which glitch go away. That might give you a clue as to exactly what waveguide mode is being excited.

Thank you for your answer! But (I'm sorry if I ask stupid question) what do you mean by "narrowband waveguide modes"? As I read, the first excited mode (theoretically) is TE01 (which is actually too weak I think (I don`t have obstacles in the cell) then is the TE10. And, to my cell, the first resonant freq. of TE10 is -predicted by theory- at 700 MHz. Then where from does it appear the one at 600? It's weird that it appears in CST (where the model is idealised)

for instance a ~ 7.2 ghz there is some narrowband glitch on both the simulated and measured data. That looks like some sort of cavity mode caused by the L x W x H of the tem cell. But since it is mostly air dielectric and a good metal conductor, it is a high Q resonance. Since it is in both measured and simulated data, it is most likely real (as opposed to a loose SMA connector, kink in a measurement cable, etc). I do not know what it is, something like a TE126 mode of something like that. Why do you not see other modes? Because you have to have 2 things happen at once:
1) the L x W x H has to be right for the mode
2) there has to be some way of exciting that mode, like the center conductor has a little discontinuity that excites energy into other modes