Couple of blended issues here. GPS only approach or RNP-AR only approach?
RAIM prediction is run before you depart, and if there is an outage predicted, you cannot plan to use that approach. As it is a prediction, there are the RAIM calcs real time on the aircraft. That is what really matters.
RAIM needs a minimum of 6 satellites to provide the fault detection and exclusion.
With RNP-AR, you are required to have the RAIM provide the acceptable HIL/HPL for the applicable RNP level. You may still have GPS coverage sufficient for an RNAV GPS approach, but RNP -AR is different.
Enroute, you need to real-time monitor HIL/HPA to ensure that you meet the required RNP level for the procedure at arrival.
The HIL coefficient is a circle around the ac keeping it with the 2x RNP boundaries. It is not the 2x boundary, but an alarm to tell you that you are getting close. If it exceeds the required level, you will get the Horizontal Alarm Limit (HAL) on the procedure and are required to discontinue the procedure.
As noted in your example, the RAIM enroute may allow for an RNP 0.5 vs 0.3 RNP, but I have not really seen a procedure where there are multiple RNP levels with associated minima. I am not sure what you mean by increased minima by RNP level, or am I misreading this? Just so you know, if the procedure is designed for 0.1 it is 0.1, you cannot use it at 0.3 RNP
The northern latitudes are very well covered, so I see little issue on coverage, nor the prediction viability.
This is a Eurocontrol program. For flight ops in in North America use this
http://www.nstb.tc.faa.gov/24Hr_RAIM.htm note 0.1 and 0.3 HAL coverage.
The HAL of 0.3RNP is 556m but depending on the FMC, it will alarm before this, as in reality HIL and HAL or not the same.