I'm working on a project where I need to perform a viewshed on several lines for comparison that are about 100 miles long each. They share some common routes in the beginning and the end, and they diverge in the middle. I'm using 10m NEDs that I have merged and converted to a GRID file. Performing a viewshed on the vertices was taking days, so we rethought our strategy. We decided to treat the common routes as one viewshed analysis, and separated out all the other lines in between as separate lines. Also to cut down on analysis time we placed points along the lines every 500ft. The analysis still took a very long time, but at least it was hours and not days. When the viewsheds were complete the outputs of the common lines and each diverging line were combined to create a viewshed for each line. For the last part I used the mosaic to new Raster tool and in the Mosaic Operator field I chose Maximum. Looking back on my process though, is it possible for this piecemeal approach to yield inaccurate Value results?
I'm now running a new viewshed with all the points along the line and 10m NEDs converted to a GRID clipped to 4miles round the line. I started it a few hours ago, crossing my fingers that it will be done by tomorrow so I may compare results. Any thoughts or recommendations are welcome.