I believe it is stated in RAS documentation that calculated WSE values as low as 0.01 are considered negligeable. That being said, if a target calibration of 0.05 cfs is desired, I would imagine RAS not being sensitive to roughness coefficients. Instead, what I would try is increasing or decreasing the # of XS’s in the model because the averaging between XS data will alter results but to the hundredth of a decimal place…I am not sure. Good luck