FLOOD06431

• County: Statewide
• Basin: Statewide
• Title: Estimating Design-Flood Discharches for Streams in Iowa Using Drainage-Basin and Channel-Geometry Characteristics
• Date: 1/1/1993
• Prepared For: Iowa City
• Prepared By: USGS
• Floodplain - Doc Type: Educational/Technical/Reference Information

channels had stabilized. Commonly, the
active-channel portion of the channel will adjust
back to natural or stable conditions within
approximately 5 to 10 years after channelization
occurs (Waite Osterkamp, U.S. Geological
Survey, oral commun., October 1992). Two data
sets thus were compiled for the
channel-geometry multiple-regression analyses:
a 157 -station data set that did not include
bankfull measurements and a 111-station data
set (a subset of the 157-station data set) that
included both the active-channel and bankfull
measurements.

Channel-Geometry Characteristic
Equations

Analysis of Channel-Geometry Data on a
Statewide Basis

Multiple-regression analyses initially were
performed on both data Eets. Statewide
equations were developed for each data set
using the ordinary least-squares (OLS) and
weighted least-squares (WLS) multiple-
regression techniques previously described. The
best equations developed in terms of PRESS
statistics, coefficients of determination, and
standard errors of estimate for E,ach data set are
listed in table 3. The channel-geometry
characteristics identified as most significant for
the Ill-station data set were bankfull width
(BFW) and bankfull deptb (BFm. The
channel-geometry characteristJ.c identified as
most significant in the 157-stat:ion data set was
active-channel width (ACW). Table 9 (at end of
this report) lists the average values for BFW,
BFD, and ACW for the streamflow-gaging
stations analyzed in the 111- and 157-station
data sets. Appendix C (at end of this report)
outlines the procedure for conducting channel-
geometry measurements of these
characteristics.

Comparison of the average standard errors
of prediction listed in table 3 indicate that the
data set that included bankfull measurements
provided better estimation accuracy for the
design-flood discharges investigated in this
study than did the active-channel measure-
ments in th~ other data set. The size and shape
of the channel cross section is assumed to be a
function of streamflow discharge and sediment-
load transport. The bankfull channel is a longer

term geomorphic feature predominately
sculptured by larger magnitude discharges,
whereas the active channel is a shorter term
geomorphic feature that is sculptured by
continuous fluctuations in discharge. Because
the design-flood discharge equations developed
in this study estimate larger magnitude
discharges, a multiple regression relation with
better estimation accuracy was defined using
bankfull characteristics.

In an attempt to further improve the
estimation accuracy of the equations, each
gaging station was classified into one of six
channel types for which separate multiple-
regression analyses were performed. Gaging
stations were classified according to channel-
type classifications described by Osterkamp and
Hedman (1982, p. 8). This classification is based
on the results of the sediment-sample analyses
of percent silt-clay content (SCbd) and diameter
size (Dso) of the channel-bed particles, and the
percent silt-clay content of the left (SC1bk)and
right bank (SCrbk) material. The channel-
geometry flood-estimation equations developed
using this procedure were inconclusive because
the estimation accuracy of some channel-type
equations improved while the estimation
accuracy of other equations decreased. An
analysis of covariance procedure described by
W.O. Thomas, Jr., (U.S. Geological Survey,
written commun., 1982), wherein each channel-
type classification was identified as a
qualitative variable, was used to test whether
there was a statistical difference due to
channel-type classifications. Based on the
results of this analysis, there was no significant
difference between the channel-type equations
and the equations developed without
channel-type classification. Because of the
results of these two channel-type analyses,
statewide channel-geometry equations classi-
fied according to sediment-sample analyses
were determined to not significantly improve
the estimates of design-flood discharges for
streams in Iowa.

Analysis of Channel-Geometry Data by
Selected Regions

Examination of residuals for both sets of
statewide channel-geometry equations listed in
table 3 indicated evidence of geographic bias
with respect to the Des Moines Lobe landform

22 ESTIMATING DESIGN. FLOOD DISCIlARGES FOR STREAMS IN IOWA

,
,