This illustates problems in creating a shaded-relief image of an area east of Thule. Because DEMs are always altered in the process of reprojection, it makes sense (except with large areas that substantially change shape in reprojection) to calculate the shaded-relief image on the raw image and then reproject it. This is explained in Ontario work, but was not tested until I got to Thule. A one-second pixel is more than four times as high as it is wide, so parameters must be adjusted when hillshading the one-second DEM. To keep things as simple as possible, I lit the landscape from the west (bearing=270°) and I avoiding possible complications of reprojection by simply interpolating extra rows. To avoid having ArcMap outsmart me, I did the work at the arcinfo command line and in python. | ||
One-second DEM shaded from the west with standard .0000111 factor, then shaded image squeezed to equal-area proportions. As we expect, it looks flat. | One-second DEM squeezed to equal-area proportions, then hillshaded. | One-second DEM shaded from the west with standard .0000444 (vertiacl exaggeration of 4), then shaded image squeezed to equal-area proportions. We are surprised that this differs from the middle image. |