I have two elevation datasets which I am trying to subtract to calculate thickness (using ArcGIS raster calculator). The first datset is a tif which is oriented north-south and has a cell size of 5 ft x 5 ft. The second data set is a surfer grid which has a cell size of 10 ft x 10 ft but is rotated by approximately 30 degrees to the east. When I subtract the two rasters, I select bilinear re-sampling in the environments section of raster calculator. However, when I spot check my resulting output raster of thickness, I often get a thickness that is vastly underestimated (e.g. thickness should be 10 ft but the raster output equals a thickness of 2). Strangely enough, other cells that I spot check will have a thickness that is close to what I expect. I also re-ran the same calculation but used nearest neighbor re-sampling; I got the exact same results.
What is causing this discrepancy? Is it the rotated grid? How can I fix this?