For those researches that try to reconstruct scenes or environments from the past, 3D modeling and simulation seem most appropriate. Like the Gettysburg project 3D modeling of Roman and Hellenistic architecture and environments, not perfect information to reconstruct the whole environments is not available. By 3D modeling using such sparse information, it is possible not only to reconstruct what exactly happened, but also come up with hypothesis what might happened from one event to another and conduct more research to reconstruct as much close as possible.
I think manual modeling makes the most sense when replicating items from the past like broken pots and rusted knives. Also, I remember that I saw physical models of ships for all kinds purposes such as war, fishing, and transportation.
Based on the experience from the last class activity, procedural modeling is very good for modeling a world with given sets of instructions or rules. Although it requires quite a lot of computer resources for running, it can generate fully realized models of past environments.
Photogrammetry obtains reliable information about physical objects and the environment, interpreting photographic images. In the last class, we used dozens of images for creating a 3D model of an artifact. I think that this technology can be very helpful in 3D printing.
For Roman and Hellenistic project, I think incorporating manual and procedural modeling, and photogrammetry together would yield more realistic environments. One idea that I have for now is first creating a physical model of buildings as much close as possible. Then, digitize it using photogrammetry. Lastly, insert those digitized models into the virtual world so that the environments now look more authentic and realistic.