7812

• UCREFRP Catalog Number: 7812
• Author: Many
• Title: Rivers, Studies in the Science, Environmental Policy, and Law of Instream Flow
• USFW Year: 1990
• USFW - Doc Type: Rivers
• Copyright Material: YES

COMPARISON OF FLOW SYNTHESIS METHODS
The two methods used to synthesize
monthly flows from existing flow records
for ungaged and/or regulated rivers in
New England are referred to here as (1)
the New England ABF method, and (2) the
flow duration method.
The major difference between the two
methods of calculation is that the New En-
gland ABF method primarily uses monthly
mean values without regard to biases
caused by data variability, while the flow
duration method uses median values,
which are less sensitive to biases induced
by highly variable data.
The New England ABF method portrays
typical unregulated New England stream
flow based on calculations made from USGS
gage data from selected unregulated wa-
tersheds throughout New England. The
gages used in this analysis were selected
according to the following criteria (USFWS
1981):
1. Essentially unregulated flows (i.e., if
regulation exists, it must occur at times
of the year that will not influence cal-
culation of the ABF);
2. A minimum drainage area of 50 square
miles;
3. A minimum period of record of 25
consecutive years;
4. A record quality rating of good or ex-
cellent as indicated in the comments
appended to USGS stream gage data.
The USFWS indentified a total of 48
stream gages in the New England area that
appeared to fit the above criteria, distrib-
uted as follows: 13 gages in Maine, 11 gages
in New Hampshire, 13 gages in Vermont,
10 gages collectively in Massachusetts,
Rhode Island, New York and Connecticut,
and one in Canada (bordering on New
Hampshire) (Table 1).
For each stream gage, an annual August
mean flow value was calculated for each
year of record by averaging the daily mean
flow data for the month. For each gage, the
entire period of record available at the time
of-analysis was used. The median value was
then selected from among the array of an-
nual August means. The resulting flow val-
ue for a given river was then divided by
the respective drainage area to determine
the cfsm value for the river at the gage
location.
These August median cfsm values for
all 48 selected New England rivers were
then averaged and termed the New En-
gland August median flow. This average
cfsm value, determined to be 0.48 cfsm for
New England, was then rounded by the
USFWS to 0.5 c£sm and was defined as the
ABF recommendation.
The New England ABF method for cal-
culating a median August flow value for a
given gage employs the median of a series
of mean values. A mean value can poten-
tially be skewed by extremes in one tail of
the distribution (Huntsberger 1970). In or-
der to use the mean August flow value as
a statistic of central tendency for a given
year of record, the assumption must be
made that the daily flow data are symmet-
rically distributed within that month. One
example of this is the familiar bell curve
(Sokal and Rohlf 1969). Unregulated
monthly streamflow data for August in
New England typically form a skewed dis-
tribution, since data are clustered around
the lower extremity of the value range, with
few values in the high range. A small num-
ber of intense storm events (a common
phenomenon) can substantially skew the
monthly flow distribution. In this type of
data distribution, the mean flow value re-
sides at a higher point than where the me-
dian flow values occur, and the median is
a more useful statistic than the mean for
describing the central tendency of such a
distribution (Chow 1964; Sokal and Rohlf
1969). This statistical problem of data dis-
tribution was not addressed by USFWS
when deriving the New England ABF pol-
icy (G. Russell, personal communication).
To illustrate this data distribution prob-
lem, this study examined data symmetry
for two unregulated gages in Maine. For a
period of 24 years (1961-1984 inclusive),
the individual annual August median val-
ues for the Mattawamkeag River and the
Piscataquis River (at Dover-Foxcroft) were
determined and plotted from daily USGS
data. Figures 1 and 2 illustrate that extreme
variation exists among years and that the
median occurs nearer to where most data
lie than does the mean and is a more rep-
resentative statistic of typical flow. This
1N- 10 Rivers • Volume 1, Number 1 January 1990