Now that you've given me a handle on what the error is, I've been sitting here for the past half hour drawing spatial and time domain graphs trying to conceptualize precisely how the error is occurring, haha. I've even drawn up a little meteorological grid model to try to visualize what would happen to each forecast point as a front washes over.
Bottom line is - that artifact drives me nuts. It's egregious as hell, and it seems like there should be a relatively easy way to ensure there is better handling on the assigning of precipitation type to QPF. However, I'm not so dumb that I don't realize that if it were really so easy to fix, folks much smarter than me would have already fixed it.
Thanks for helping explain it though!