I have a generic digital surface model (DSM) and I want to simulate illumination in different parts of the country on the same day, all using the same DSM. (i.e., how does illumination differ in Seattle and San Antonio on March 4?) I am using the ArcMap v10.4 hillshade function, but there are other implementations of hillshade functions in various software packages.
The ArcMap hillshade function requires the solar elevation and azimuth as inputs. These can be calculated for a given location and date/time using NOAA's solar position calculator.
I want to know if the spatial location of the DSM affects the output of the hillshade function, or if the elevation and azimuth are the only variables that affect the output.
To test this, I created three separate rasters as inputs: my original projected DSM, the DSM with the projection removed, and the DSM projected but shifted an arbitrary number of X and Y units. In R:
require(raster)
#Input one
DSM <- raster("DSM.tif")
#Input two
DSM_nocrs <- raster("DSM.tif")
crs(DSM_nocrs) <- NULL
#Input three
DSM_shifted <- shift(DSM, x = 100000, y = 100000)
I inputted these three rasters into the ArcMap hillshade function using the same solar elevation and azimuth. I saved the outputs to my working directory. I then compared the raw values of each output in R.
DSM_hillshade <- raster("DSM_hillshade.tif")
DSM_nocrs_hillshade <- raster("DSM_nocrs_hillshade.tif")
DMS_shift_hillshade <- raster("DMS_shift_hillshade.tif")
all.equal(values(DSM_hillshade), values(DSM_nocrs_hillshade))
[1] TRUE
all.equal(values(DSM_hillshade), values(DSM_shift_hillshade))
[1] TRUE
As you can see, R indicates that the values of the three hillshade rasters are the same even though one wasn't projected and one had a different spatial location. This suggests that the spatial location of the DEM/DSM has no impact on the output of the hillshade function. Perhaps then, for my analysis, I do not need to change the spatial location of my generic DSM and only need to change the solar elevation and azimuth.
My question: Does this seem like a fair interpretation? Is there any reason to believe that the spatial location of a projected DEM/DSM has an impact of the output of a hillshade function?