I would like to sample terrain elevation values along stream-beds.

[DavisTolman]

I will then create a regional surface that corresponds to the stream-bed elevations.  I've thought of converting the flow-lines into a series of points, sampling the terrain at these points and then using one of the geostatistical tools to create a "smoothed" surface from these points.  Anyone have thoughts on how to best accomplish this this?  

[RickeyFight]

Davis Tolman‌

Do you just need point elevations along a stream bed? What are you end goals with this? To push online? 

I might have done something like this. 

What are you looking for in a smoothed surface? 

Do you have any screen shots? 

[DavisTolman]

I have had some success in predicting flood risk by building a "residual

surface" from detailed LIDAR data. I construct a smoothed surface across

the watershed (often a 7th to 10th order polynomial surface) and then

subtract the original detailed terrain surface, leaving the residual. This

surface provides a great visual representation of drainage systems across

a broad area. Rather than using a sampling from the entire terrain data

set, I'd like to try using just the axis of the drainage systems. My

hypothesis is that it may be even a better predictor. I've have been

working in ArcMap and ArcGIS Pro.