Maybe one could produce a small note about the base change for closed immersions, similar to the smooth base change proposition in Arakelov 1. There are some paper notes related to this, and also an email conversation with Jakob: search gmail for “Base change for closed immersions”.
nLab page on Motivic base change