The idea of the "Preservation format" is only useful if you get to choose (or influence) the format of the material. This is common when the material is being digitised by the same organisation that is going to preserve it. The term is often used in contrast to the "Access format" (often a lossily-compressed version), and so may also be referred to as the "Preservation master".
Similarly, many archives operate under a deposit model for digital content where the primary "master" and "access" forms are decided by the original authors, and where the repository has to make some policy decisions about how to understand and handle those authorial decisions. In this case, "preservation formats" tend to go hand-in-hand with normalisation strategies, generating a secondary copy in a format that is believed to have a better chance at longeivity (even if this does mean some risk of loss).
For web archives, forensics, etc. it's not up to us to choose the format. We can still call the original WARCs/images the "preservation master", but the formats within reflect the values and choices of the original authors, and cannot be overridden lightly (apart from anything else, it's usually not possible/feasible to negociate with those original authors in these cases).