FLOOD07843

• County: Statewide
• Basin: Statewide
• Title: Nationwide Summary of US Geological Survey Regional Regression Equations for Estimating Magnitude and Frequency of Floods or Ungaged Ssites 1993
• Date: 1/1/1993
• Prepared By: USGS
• Floodplain - Doc Type: Educational/Technical/Reference Information

0.23. Some sites in the testing had 500-year peak dis-
charges which exceeded the maximum envelope values
in Arizona, Oklahoma, and Utah. Consequently, the
user must be aware that some T-year peak discharge
estimates may exceed the maximum flood envelope
value for that site. Careful attention should be given to
determining in which maximum flood region a basin is
located (Crippen, 1982).
The same procedures were used in comparing
500-year estimates to the Crippen and Bue maximum
envelope value for States without 500-year equations.
The mean ratio of extrapolated 500-year peak dis-
charges to the Crippen and Bue maximum envelope
values was 0,17, Some sites in the testing had 500-year
peak discharges which exceeded the maximum enve-
lope values in Arkansas, Connecticut, Kentucky,
Nebraska, New Mexico, New York, South Dakota,
Tennessee, and Texas. Again, the user must be aware
that some T-year peak discharge estimates may exceed
the computed maximum flood envelope value for that
site.

The mean standardized skew residual for the
sites with 500-year equations was 0.155 with values
ranging from 6.46 to -1.93. The mean of the RMS devi-
ations of the log residuals of the Stale equation T-year
peaks from the smooth log-Pearson Type III curve was
0.00437 with values ranging from 0.0389 to 0.0001,
while the mean of the smooth-curve RMS deviations
was 0.3667 with a maximum of 0.97 and a minimum of
0.11.

The mean standardized skew residual for the
sites without 500-year equations was 0.104 with values
ranging from 10,1 to -2.57. The mean of the RMS devi-
ations of the log residuals of the Stale equation T-year
peaks from the smooth log-Pearson Type III curve was
0.00565 with values ranging from 0.0623 to -0.0099
while the mean of the smooth-curve RMS deviations
was 0.33 with a maximum of 1.39 and a minimum of
0.06.

Results of the testing indicated that the frequency
curves generally fit a log-Pearson Type III distribution
by virtue of the small RMS deviations of the log resid-
uals of State equation T-year peak discharges from the
smooth fitted log-Pearson Type ill curve. The low skew
errors suggest that the skew coefficients, computed
from the frequency curves by NFF, are very similar to
the generalized skew coefficients computed for the
United States (Interagency Advisory Committee on
Water Data, 1982).

I
\

Extrapolation Testing for the SOO-Year Flood

Estimates of 500-year peak discharges for 149
stations used in the testing were obtained from pub-
lished flood-frequency reports or from USGS District
offices. The extrapolated 500-year peak discharges
were obtained by using station frequency curve values
for 2-year through l00-year peak discharges and then
extrapolating to the 500-year recurrence interval using
the extrapolation procedures described earlier. These
extrapolated 500-year peaks differed by an average of
0.04 percent when compared with the 500-year peak
discharges from the station frequency curves which
indicated that the extrapolated peaks were similar to,
and on the average slightly higher than, the station 500-
year floods.

Regional/State Boundary Testing

I

Currently, NFF allows computations of fre-
quency curves for basins that span more that one
hydrologic region within the same State, This is
accomplished on the basis of percentage of drainage
area in each region. The user should verify that the
resultant curves reflect the flood characteristics of the
regions by consulting the respective State flood-
frequency report and by examining plots of the com-
puted frequency curves,

Regional flood-frequency computations for
watersheds that span State boundaries may give differ-
ent results depending on which State's equations are
used. Nine sites were evaluated using the previously
described methods to examine the application of NFF
to basins that cross Stale boundaries. Currently, NFF
does not allow the user the option to compute a
weighted frequency curve by drainage area for basins
which cross State boundaries. Because of this limita-
tion, the user must perform this procedure manually,
which can be accomplished by applying NFF for each
State using the basin's full drainage area. Next, the user
must manually weight the frequency curve estimates
based on the percentage of the basin's drainage area in
each State, For example, two frequency curves were
computed for the Sucarnoochee River at Livingston,
Alabama; 320 square miles of the basin's total area of
606 square miles is in Mississippi, and 286 square
miles of the basin is in Alabama. Table 1 shows the fre-
quency curves computed using the full drainage area in

i

I

18

Nationwide Summary 01 U.s. Geological Survay Raglonal Regression Equatlone lor Estimating Magnitude and Frequency of
FI_lor Ungaged Sites, 1993