My apologies for not getting back to you sooner. I've been on bereavement leave and then had to spend a little time researching your issue.
So, in 3.1.x the behavior is that AP related alarms are reported to the site where the controller is located. Since your device count in that site is 0 (indicating that the controller is not located here), it's expected that the alarms for the down APs will not be reported here. This should be consistent across the UI and API. If you are seeing alarms for this particular site in the UI, please reach out to me via email (in my profile) and we can dig deeper.
We realize that this is not optimal behavior, so in 3.2 (which has recently been posted to cisco.com for download) we changed the behavior to report these AP-specific alarms in the site where the AP is located.
Thanks for your reply.
This makes sense, although I do not know why it should be designed that way. We only use Prime for wireless management and monitoring and don't use it for any other type of device other than controllers and access points.
We will look at upgrading to 3.2 in the future and thanks again for your detailed response.