In Reply Refer To:
WGS-Mail Stop 415 December 16, 1985
OFFICE OF SURFACE WATER TECHNICAL MEMORANDUM NO. 86.02
Subject: PROGRAMS
AND PLANS—Low-Flow Frequency Estimation at Partial-
Record Sites
Instructions and guidelines on how to estimate low-flow
statistics at partial-record sites are provided in several documents. In all cases, the recommended approach is to
make base-flow measurements at the partial-record site and, in some way, relate
them to concurrent daily flows at a nearby index station. The D-day, T-year low flow at the
partial-record site is estimated using this relationship and low-flow
statistics at the index station.
Techniques of Water-Resources Investigations (TWRI) Book 4, Chapter B1,
recommends using graphical correlation techniques for establishing this
relationship and transferring D-day, T-year low flows from the index station to
the partial-record site. Surface Water
Branch (SWB) Technical Memorandum No. 79.06 expands upon guidelines in TWRI,
Book 4, Chapter B1, but still recommends the same graphical technique for
estimating low-flow statistics at the partial-record site. Furthermore, SWB Technical Memorandum No.
79.06 provides an evaluation of the accuracy of low-flow statistics estimated
in this way. Copies of
Open-File Report
85-95 “Low-Flow Frequency Estimation Using Base-Flow Measurements” by J. R.
Stedinger and W. O. Thomas, Jr., were distributed with SWB Technical Memorandum
85.09. This report recommended
estimating low-flow statistics at partial-record sites by (1) estimating the
mean and variance of the annual D-day low flows at the partial-record site and
(2) using those statistics to estimate the D-day, T-year low flow. Finally, Office of Surface Water (OSW)
Technical Memorandum No. 85.17 recommended the continued use of the graphical
correlation techniques described in TWRI Book 4, Chapter B1.
The purpose of this memorandum is to provide further
guidance on estimating low-flow statistics at partial-record sites. As the use of computers became more
prevalent, many analysts began to use linear regression analysis to transfer
low-flow statistics to partial-record sites in lieu of using the graphical
correlation technique. In Open-File
Report 85-95, Stedinger and Thomas point out that the linear regression
approach of transferring low-flow statistics is biased. Attached is a brief summary and
clarification of the important technical points and conclusions presented in
that report. Based on the attached
summary and Open-File Report 85-95, the following recommendations are made:
1. The linear regression approach should not be
used to transfer
low-flow
statistics from an index station to a partial-record
site.
2. If a mathematical technique is utilized,
then either the
moment
approach suggested by Stedinger and Thomas or the
maintenance
of variance extension technique (MOVE.1) should
be
utilized. Approximately 10 base-flow
measurements should
be
available to use either of these mathematical techniques.
3. The graphical correlation technique
described in TWRI Book 4,
Chapter B1,
can be utilized but the analyst should give more
weight to
the lower data points in the base-flow measurement/
daily mean
flow relationship. This graphical
correlation
technique
may be most appropriate when there are fewer than 10
base-flow
measurements available or when there is a nonlinear
relationship between base-flow measurements at the partial-
record site
and concurrent daily mean flows at the index
station.
Estimates of low-flow statistics at partial-record sites
should be entered in the Streamflow/Basin Characteristics File. These estimates should be qualified with a
code of SL (synthetic low flows) as described in the WATSTORE User’s Guide
Volume 4, Chapter II.
Any questions about the attached summary or Open-File
Report 85-95 should be referred to Will Thomas in the Office of Surface Water.
Verne R. Schneider
Acting Chief, Office of Surface
Water
WRD Distribution: FO, SL