I want to perform zonal statistics on PRISM precipitation rasters using watershed shapefiles as my input zone features. I'm operating under the precept that I should project all my data into the same PCS before doing zonal stats. Original PRISM data is in NAD 1983.
Yet, I am having a couple issues understanding why Project Raster apparently adjusts the original PRISM rasters.
First, regardless of interpolation method I choose (nearest, cubic, bilinear), the position of the resulting projected raster is the same. Not sure it matters but dataframe is NAD 1983. From what I've read, generally speaking, cubic is best for continuous data like precipitation and that it should change the cell's values (it did in my projection process). Though, I expected spatial location of the original raster data to be preserved using cubic as opposed to nearest interpolation.
Second, after running Project Raster, the cell size of the original raster changed from 4x4 km (that's the resolution of publicly available PRISM data) to ~ 5x5 km in the projected raster. These are approximate values using measurement tool.
Why does the spatial location and cell size change during the projection process? I didn't set a Snap Raster or change the Output Coordinate System or Raster Storage in Env't Settings either. Any link to a scientific paper or blog/online article post discussing the best practices for Project Raster?