RF power measurement it's a very different beast from small signal S-parameter measurement. Neither a regular VNA nor a SA are the right tools for it.
If you want to do it right (i.e., with the lowest possible measurement uncertainty), the usual way to go is a a power metering setup for which you'll need:
- A power meter. It's a "generic" instrument.
- A power sensor. It's the application-specific part of your test setup. It must fit your power level, dynamic range, frequency range and mechanical interface (connectors, waveguide, etc.) requirements.
- Additional test accessories. Attenuators, isolators, power splitters, power couplers, switches... depending on what you want to do.
This is how a power meter and several power sensors look like:
I highly recommend you to read Agilent's application note 64-1C on Fundamentals of RF and Microwave Power Measurements so you can get a good grasp about it. Here you can find more or less the same information in a "keynote slides" format.
In that AN you'll find a method for calculating measurement uncertainty ("error") and for selecting an adequate power sensor. For example, given your high power requirements (+30 dBm) and frequency range, you should be looking into the following tables (pp. 92-93):
Diode sensors have better dynamic range that thermocouple sensors, but worse frequency range. 8482H or 8481H could be good fits for your needs.
Follow this link to find more details about how to make a complete power metering setup like this:
There are also non-linear VNAs, which of course are by far the best option, but I suspect they must be prohibitively expensive.
NOTE BELOW: I'm in no way affiliated to Agilent Technologies. I'm referring to their instruments and applications notes because I've had practical experience measuring power with those. Instruments from other manufacturers could be equally or better suited to your application that Agilent's (it's up to you to verify that).