How do you know how many cross-sections you need to survey for an analysis before you conduct a survey, construct a model, and interpret the results of the model?
I assume this varies greatly depending on analyses, but is there any general guidance for this?
You need to make sure you have cross-sections at grade breaks, change in cross-sections areas, change in roughness, and around bends. Below is a link that discusses a way to determine how many cross-sections based on bank full depth and slope. It is also important to make sure that the downstream boundary (assuming sub-critical flow) is far enough away to that it does not impact the area of interest.