CONUS point ndvi data geojson, the CONUS point geometry (coordinates) geojson, and an example of a GAM.pkl file run on those inputs. 

Two jupyter notebooks are included:
1) phenology_data_preprocessing_EROS.ipynb: takes the input geojsons and gets through running the gam model. I've re-run and tested it here, so hopefully, all you need to do is provide the path strings to the input geojsons and a couple for output files.
2) GAM_to_geotiff_EROS.ipynb: this one takes the gam model pkl file you generate in (1) and a folder of landcover geotiffs as inputs, and generates daily ndvi geotiffs. I think we'll need to run this one together since a lot is hardcoded depending on the land cover geotiffs, and rather than mess it up by trying to generalize the code (which I'm admittedly not great at), I just left it thinking we could probably walk through it and change things to match the EROS stetup.

I also included an edited version of the conda .yml file for the environment. The manual editing I did probably messed it up enough that you can't simply build a conda env with it, but it got rid of a ton of packages that you don't need and might help figure out package versions if there are issues. I think the thing to do is make a conda env with python 3.7,  geopandas and jupyter notebooks, and then add from there based on the error messages from running imports in the notebooks.