GIS data layers developed in the past few months and meteorological station data are used to create a first "pilot" dataset to test the GAM regression in Oregon.
Input variables are: lat, lon, ELEV_SRTM, DISTOC, ASPECT.
Locations of stations data for Oregon were reprojected in NAD83 Oregon Lambert Conic Conformal (EPSG2991). All original raster input layers were reprojected from sinusoidal to EPSG2991 and windowed (i.e. spatially subset) to match the extent of the Oregon case study. Note that a distance to ocean (DISTOC) variable was created from the input land cover data available on E&O server.

The work was done through ArcGIS and IDRISI softwares but a r script is currently under development to translate the necessary steps to automate the proces.