@Michael Hawkins yeah your multipath diagnosis is almost certainly right — water is one of the worst surfaces for this because it's a near-perfect reflector, so you're essentially getting a second set of "ghost" signals arriving slightly delayed and confusing the receiver into thinking it's somewhere it's not. The circular drift pattern you're describing is pretty classic for that.
The first thing I'd try is antenna placement and orientation. If the antenna is mounted low on the hull or anywhere near the waterline, you're basically giving those reflected signals a direct path in. Getting it as high as possible and making sure it has a good ground plane underneath it (a metal disc, even a piece of copper tape glued to a flat surface) helps reject low-angle signals, which is where most of the multipath is coming from off the water.
The 18Hz fix rate on the MultiNav Pro+ is actually working against you a little here in one sense — at high update rates the receiver doesn't average as aggressively, so the noise from multipath shows up more in the position stream. If your autonomy stack allows it, you can smooth this on the software side with a simple moving average or a Kalman filter over the position output, which helps a lot for a stationary or slow-moving platform like an anchored boat.
One thing worth checking in your setup — are you fusing the GNSS with the IMU/gyro data at all? If you've got the RFOXiA Sensors Module in there too, combining that accelerometer/gyro data with the GNSS position is probably the cleanest fix for this, since the IMU won't care about reflected satellite signals. Even a basic sensor fusion approach will kill most of that circular drift.
