This looks like a good project to be added to:
I do wonder about your decision for a Raspi or Jetson. The cost of the brains part would be very small compared to this PCB and for the marine environment I would settle for nothing less then a brains PCB that is specified for full industrial temperature range. Olimex for example also makes similar Linux brain PCB’s and they can be bought with industrial temperature range for a small price difference (EUR 10 or so). Olimex is also very friendly to the Open Source “community”. Much more so then Raspi, which locks down their camera to their firmware and more of such things. Raspi is also the company that dragged on 32-bit os (and all the extra time needed to keep supporting it) for 10 years after it should have been obsoleted. (As you can read. I don’t like Raspi much, so I do not follow them closely)
Running such a PC from an SD-card is also not the most reliable, but I’m not sure if cheap eMMC is much better. There are companies which make extremely high reliability and endurence SD-Cards. I like the way hardkernel solved it. they use eMMC modules with a connector which can be swapped if worn out, or just to try another OS version or configuration. It also leaves the uSD card slot empty so you can use it for removable media such as what uSD is intended for.
Another thing I would consider is to have some stand-offs, and on that mount a spare brains PCB, which has no electrical connections to the rest. At a minimum this is just a footprint with 4 mounting holes on your PCB, and silk screen text"spare Brains box". That way the spare PCB has a home and will never be lost. Any technician servicing or troubleshooting the panel, or during emergencies at sea will find it if it’s needed. Best would be to leave that PCB in it’s own anti-static bag, and then just poke the mounting screws (or plastic snap clips) through the bag.
About the enclosure…
(Very dimly related to your enclosure).
Some time ago I had some interest in a weather station and thought about how to make a reliable enclosure, which is still easy to make and cheap. I came up with the Idea to use regular PVC pipe and then glue on some “inspection caps”, which have a screw-on lid and rubber seal. That would still leave differential air pressure from wanting to get in on some days, and for that I thought of some flexible bladder, which could for example be made from a bicycle inner tube. Then throw in an extra bag of desiccant for good measures and that should be plenty enough.
One of the problems I heard about is moisture seeping in from between the inner and outer isolation of multi strand cables.