I want to have a raster stack of two different resolutions. One is from VSNIR and one from SWIR sensors (hyperspectral). The VSNIR has a resolution of 25 cm and the SWIR of 80 cm and both data have the same importance for my study. If I upscale the VSNIR and downscale the SWIR rasters to 60 cm, the resulting stack could have the minimal loss of accuracy from both bands. The number 60 comes as the area of the square which is the average of the areas of the two pixels 0.5 x √(25 x 25)+(80 x 80). I have doubts, that I have to consider also that the upscaled raster will have a greater loss of accuracy (the result is extrapolated) and add a parameter which changes the optimal pixel resolution.
Any ideas of what would be the optimal common resolution for bands of such different resolutions? I will use this data for spectral regression analysis.