Great discussion this morning! 

Before I forget, I wanted to send out two papers.

Bolton et al 2020 is an example of data fusion between Landsat 8 and Sentinel-2 to create a CONUS scale land surface phenology model. Nice recent example of a similar project. 

Thuiller et al 2003 is the paper I mentioned comparing Generalized models with classification tree approaches. They are trying to predict species distributions for plants, but I think it's similar enough to provide some reference to how GAMs perform across a range of spatial resolutions.