The only information we have on streamflow in many streams is that recorded at gauges, and yet, to make the measurement easier, gauges tend to be located in unnatural places - bridges, weirs, along riprapped banks etc.
While this ensures a stable cross-section, it's not so great for figuring out where the bankfull stage is at the gauge. So we have to go up- or down-stream to the nearest natural bank and hope that it is close enough that values of slope, Manning's n, etc. will remain relatively similar to those measured at the gauge so that we can transfer the hydraulic geometry.
