@Amber Bowman your gut is probably right, but I'd rule out body shadowing first before committing to a bracket redesign — if the dropout angles correlate with the airframe structure sitting between the two antennas rather than just attitude change in open sky, that's the easier fix and it's worth confirming before you go down the polarization rabbit hole.
That said, polarization mismatch at long baseline is absolutely real and the 20dB theoretical worst case is not as theoretical as people hope when you're talking about a whip that's been rolled 60+ degrees. The MultiNav Pro+ front-end has solid receiver sensitivity which gives you headroom, but that headroom is finite and at the drone-to-drone range envelope you're essentially running at the edge of the link budget by design — there's not a lot of slack left to absorb a 10-15dB polarization hit on top of any path loss variance.
The RHCP on the ground station approach is genuinely the right move for your use case. Yes you take the 3dB circular-to-linear penalty on the airborne whip, but you buy yourself immunity to the worst of the polarization swings as the platform maneuvers, which is a much better tradeoff than having a link that's fine in level cruise and then falls apart in a bank. A lot of fixed-wing telemetry setups run exactly this configuration — helical or patch RHCP on the ground, standard whip on the airframe — and it works well. The consistent 3dB loss is predictable and you can plan around it; the random 15dB swing from cross-pol is not.
If you want to share the dropout angles you're seeing and roughly what attitude the platform is in when it happens, that would help narrow down whether it's purely polarization or if there's a body blockage component mixed in too.
