I’ve only really measured EW manually but found it very sensitive, particularly in noisy spectra, to deciding exactly where the continuum goes (linear, spline fit, what order to use ? etc) and where the line meets the continuum for the integration limits. How does IRAF do this? In your screenshot it looks like it is fitting gaussians to separate merged line components. ( I remember something similar in SPLAT) Is this part of the EW measurement?
Robin