For layout like this, you’ll need to compile and then do the final formatting in a word processor or other software with this capability (Word, Indesign, etc.). You can choose in Scrivener to compile just a single document or collection of specific documents, so you could select those items that need to be landscape and compile them separately, but you would still need to merge them back together into a single file after leaving Scrivener. The editor isn’t constrained to a page size–all of that is handled in the compile Page Settings–so you can focus on drafting and organizing your text in Scrivener without needing to worry about “portrait” or “landscape”, and then do that adjustment at the end, once your text is written.
In the case of a dissertation, I expect you’ll need to be compiling sections regularly for your advisor or others following your work, so this may need be something done more often. I expect there’s a macro or other quick way to post-process in Word or similar, but I’m not knowledgeable about that end of things. Some some of the other users who’ve been down the PhD path, or who are just pro Word users, may be better able to help you better with that.
The reason, for me, to use Scrivener is that it gets away from Word and it’s large scale incompetencies.
Word suffers from silent errors (pagination lacunae) as complex documents exceed several hundred pages; mix in pictures and bounded objects of varied sizes and Word gets so fragile that WYSIWYG breaks and it will print something other than it puts on screen. It becomes quite (and randomly) eccentric; changing a comma in one place might offset margins elsewhere or cause a blank 1/3rd of a page to appear 10 pages away.
My last 4 days of my MSc were a hell of trying to get Word to generate a .pdf which matched correctly what was on screen; I’ve learnt that it is famous for this.
And… Scrivener needs Word as an output engine? NOOOoooooo!..