All

There is also a more recent USGS WS paper on this interesting subject.
I have pasted the following from a one-pg handout that I give out in my
CIVL6159 'Form and Process in Alluvial Channels' course (I had 35 mm
slides made of all the colour photos in both documents; and when we are
feeling tired of number crunching in CIVL6159 we kick back in a dark
room and discuss them):

1. Barnes H.H. 1967. Roughness characteristics of natural channels.
USGS Water Supply Paper #1849, US Govt printing Office, Wash. DC, 213 p.

50 stations, data collected over 15 years. n from 0.024 to 0.097.

Sites presented:

no pure sand-bed channels.

hydrometric station present in each reach.

Qhigh from vel.-area method (measurement) or use of reliable stage-Q
curve with clear high-water marks from the event(s).

relatively uniform reaches.

all flows were in-bank flows save one, no significant floodplain
flows.

bed-material was sampled later (years after the event(s)).

each reach consisted of 2 to 17 cross-sections.

n computed using a method "virtually the same as" the slope-area
method.

for some sites more than one Qhigh was investigated (larger one
described first).

2. Arcement G.J. and Schneider V.R. 1989. Guide for selecting
Manning's n roughness coefficients for natural channels with flood
plains. USGS Water Supply Paper #2339, USGS, Denver CO, 38 p.

this article emphasizes densely-vegetated floodplains. n from 0.10
to 0.20.

Aldridge and Garrett (1973) had previously presented a rationale for
determining n based on:


nb = base value of n,
n1 = surface irregularity roughness component,
n2 = channel irregularity roughness component (prismaticity),
n3 = roughness component to explain obstructions,
n4 = roughness component to explain the effect of vegetation,
m = related to sinuosity.

Arcement and Schneider modify the Aldridge-Garrett system for
floodplains


nb = base value of n,
n1 = surface irregularity roughness component,
n3 = roughness component to explain obstructions,
n4 = roughness component to explain the effect of vegetation.

and present detailed flow-charts and tables for systematically
determining the appropriate n for a floodplain. The vegetative
component was studied in great detail by measuring the total stem area
of plants/trees in the flood plains. Depths were a few ft.

On the subject of hydraulic resistance, I like the paper:
3. Bray D. and Davar K. 1987. Resistance to Flow in Gravel-bed Rivers.
Can J of Civil Engineering, 14(1):77-86.

David
Halifax, NS