If they are truly bolted in a way that induces rigidity in the joint, then by all means you can model them as such. My guess is that, however, the connections, even though bolted, probably aren't intended to provide rotational rigidity to the connection. Without seeing a detail of the connection configuration it's hard to guess.
What I was mostly saying above is that, if you are intent on comparing in-situ testing and measured deflection, you need to be very thorough in accounting for the reality vs the model. All structural analysis is an idealized structure and it will be fairly rare that you will get an exact one-to-one correspondence between an idealized model and the reality. If you want to be exact, you will need to measure the modulus of elasticity of the actual members you have used in the structure, account for the effects of acting loads, the width and compressibility of the support (it's probably sitting on piece of wood loaded radially), the amount of rotational stiffness in the connection (which you would have to measure), the fact that the roof isn't rigidly connected and also has other factors affecting it, etc. etc. It's not a small task. You might want to consider how important it is to go down that road.
