COMPUTATIONS--Revision of Annual Flood-Frequency Analysis Program J407 According to U.S. Water Resources Council Bullentin 17B
Guidelines




In Reply Refer To:                                            July 30, 1981
EGS-Mail Stop 415




SURFACE WATER BRANCH TECHNICAL MEMORANDUM NO. 81.06

Subject:  COMPUTATIONS--Revision of Annual Flood-Frequency Analysis Program J407
                        According to U.S. Water Resources Council Bulletin 17B
                        Guidelines

The annual flood-frequency analysis program (J407) has been revised to conform
to the new U.S. Water Resources Council (WRC) guidelines (Bulletin 17B, in
publication).  The new WRC guidelines will not be ready for distribution until
early next fall, but the revised WATSTORE User's Guide for volume 4 describing
the program was recently distributed.  The new WATSTORE User's Guide describes
some of the revisions in J407, but many of the revisions are transparent to the
user of the computer program.

A more detailed description of the major revisions to J407 follows:

1.  The station and generalized skew are now weighted inversely proportional
to their mean square errors.  The mean square error of the generalized skew
must be input to the program unless the generalized skew map in Bulletin 17B
(and 17A) is used, in which case a default value is used.  The mean square error
of the station skew is computed in the program.

2.  There is a new test for detecting high outliers and a revised test for
detecting low outliers.  Both tests are based on a 10-percent significance level
for the normal distribution (zero skew).  The revised low-outlier test will detect
more low outliers than the old test which was based on generalized skew and a
1 percent level of significance.

3.  There are revised procedures for the application of the conditional probability
adjustment for zero-flow years, low outliers, and peaks below the gage base.  The
computation of flood-frequency estimates in the presence of zero-flow years, low
outliers, and peaks below the gage base is now more consistent with complete record
computations.

These changes are documented in great detail in WRC Bulletin 17B which hopefully
will be ready for distribution this fall.  A few other changes in Bulletin 17B
which are not apparent in the computer program are as follows:

1.  There are revised guidelines for estimating generalized skew.

2.  There is a revised description of the two station comparison procedure
(Appendix 7).


                                                                            2

3.  There is a revised description of confidence limits (Appendix 9).  No change,
however, in the way confidence limits are computed.

4.  An expanded table on estimated risk is provided (Appendix 10).

5.  There are revised examples illustrating the new computational procedures
noted above (Appendix 12).

The revised version of J407 will become the production version of the program
on August 7, 1981.  After this date the cataloged procedure given in the re-
vised WATSTORE User's Guide (June 1981) will access the program based on WRC
Bulletin 17B and not the previous program based on Bulletin 17A.  Anyone working
on a flood-freuqency project or about to begin one should be aware of this
change.  The old program based on Bulletin 17A will remain on the system for a
few months.  After August 7, 1981, the old program can be accessed by modifying
the execute card as follows:

                          // EXEC PKWRCA,PRO/G=J407A

Any questions concerning the revised program or changes in WRC Bulletin 17B
should be directed to Will Thomas (FTS 928-6837) in the Surface Water Branch.
As soon as WRC Bulletin 17B is published, it will be distributed to all Districts.




                                             Stanley P. Sauer