The USGS EROS site has some interesting info on how the hydro1k dataset was built up from gtopo30 data. Most interesting to me are sections 3.1.2, 3.1.3 and 3.1.4, which seem like they might be very adaptable to generating DEMs to fit a known set of rivers. This could be useful to folks like redrobes working at ME-DEM.

Not sure exactly where to put this and it may not be as useful as it seems to me at the moment. I just ran across this while trying to find geodata for the area around the Tarim Basin. I'm trying to build up a cartography portfolio for a job interview. With the USGS no longer providing seamless data outside the US, I may have to give up on my Silk Road map idea. It would have been glorious!