Thanks all!
Here's the process I used:
1) In the field: I took 3 shots of the sky, each at ISO 6400, 20s, and f/2.8. These were taken right after each other. I then followed up with a single exposure for the ground at ISO 1600, 180s and f/2.8. You can take more shots of the sky to get more noise reduction, but alignment gets harder since stars further away from Polaris move more than those closer to the it - this can lead to more streaking.
2) Back home, I opened all of the files in lightroom. Each "sky" image was processed using the same parameters, but I processed the ground differently (different white balance, contrast, noise reduction, etc.)
3) I opened the ground image and one of the sky images in Photoshop as layers. I manually blended these images to form the base for processing. When you stack MW images, the ground gets blurry which makes blending a challenge. I figured that if I started with a pre-blended base, then I could avoid having to work around the blurry foreground in the stacked images. I used some extra noise reduction on the sky layer since parts of it will be used in the final image and I didn't really care about saving details since those would come from the stack.
4) I then opened all three sky images as layers in a separate photoshop file. I masked out the ground in each layer (so the alignment would just be applied to the stars) and then used edit-auto align to align the stars.
5) I then removed the layer masks, selected the three layers, and converted them to a smart object. I set the blending mode to "median" as Jim noted above.
6) I then copied that smart object and pasted it into my base image from step 3 above.
7) Finally I did a manual blend of the smart object (sky stack) with the base, staying away from the ground, and did a slight crop to remove some of the edge issues created from alignment.
From there, salt and pepper to taste! Hopefully this wasn't too confusing
