I see huge discrepancy between researchers (difference for the same class is somewhere up to 0.2). What determines the choice of the appropriate value ? What is the role of spatial resolution of the research in this ? Thank you.
Manning's n or roughness coefficient is actually a measure of energy or head loss along a channel reach. The roughness of the channel, bank and floodplain is a major determinant of Manning's n. However, all mechanisms that contribute to energy loss should be considered in selecting its proper value. For example, flow velocity and associated turbulence are important (e.g., if velocity is near zero, the Manning's n will be zero even when the river has high roughness). Similarly flow depth is important. So is the season (e.g., more vegetation and thus higher roughness in summer).
I think engineers should think of Manning's n as more related to energy loss under specific flow conditions than to only roughness.