HEC-FFA: Flood Frequency Analysis
VERSION/DATE: Version 3.1 (2/95)
OPERATING SYSTEM: DOS
DOCUMENTATION (87 Pages):
HEC-FFA: Flood Frequency Analysis User's Manual May 1992 By US Army Corps of Engineers Hydrologic Engineering Center
This version will run on an IBM or compatible microcomputer that has the following:
640 Kilobytes (KB) of Random Access Memory (RAM) (300,000 bytes available for program execution) MS DOS 2.1 or greater 1 Megabyte free space on hard disk
The HEC-FFA user's manual describes capabilities of, input to, and output from the Flood Frequency Analysis (FFA) program. The manual includes changes that have been made to the program to reflect techniques described in the revised, "Guidelines for Determining Flood Flow Frequency," Bulletin 17B, Water Resource Council, September 1981, hereafter referred to as the Guidelines.
The computation methods are basically as described in "Section V, Determination of Frequency Curve," in the Guidelines. A very brief description of how the computer program treats specific conditions follows:
Graphical Analysis: The data are arrayed and the plotting positions may be computed by the Weibull, median or Hazen formulae.
The Distribution: The log-Pearson Type III distribution is used in the computation of frequency curve.
Skew Coefficient: The computed skew coefficient is weighted with the Input generalized skew coefficient.
Broken Record: A broken record is automatically analyzed as a continuous record.
Incomplete Record: Missing data at the low end is indicated by a negative number (-1) and the conditional probability adjustment is used to determine the frequency curve.
Zero Flood Years: Any flood events of zero are automatically deleted and the conditional probability adjustment is used to determine the frequency curve (p. 15 and Appendix 5).
Outliers: Initially the program calculates the station skew coefficient for the systematic record which is presented under preliminary results in the output. The program then tests for high or low outliers in an order depending on the value of the station skew as discussed on pages 17-19 and shown on the flow chart on page 12-3 of the Guidelines. Basically if the skew is greater than 0.4, tests and adjustments for high outliers and historic peaks are made before testing for low outliers. If the station skew is less than -0.4, tests and adjustments are made for low outliers first. If the skew is between 0.4 and -0.4, tests for both high and low outliers are made based on systematic record statistics before any adjustments are made.
Historic Events: Weighted plotting positions and statistics are computed incorporating any input historic events.
Confidence Limits: The .05 and .95 confidence limit curves are computed unless other limits are specified.
Expected Probability: The frequency curve ordinates are computed with and without the expected probability adjustment.
GENERAL INPUT AND OUTPUT INFORMATION
The input is designed to be flexible, and default values are provided for all decision variables. Any option or nonstandard item activated by the J1 or J2 record will remain in effect for all succeeding station data or until modified by another J1 or J2 record. The only records actually required for a flood frequency analysis at a station are three or more annual flood peaks (QR records) and the end-of-data (ED) record. Input data preparation is described in detail in Appendix B.
The example problems in the next section illustrate input preparation and output. The program output has been arranged to enable the tables to be copied for report purposes. When special conditions are encountered in the analysis, such as historic data, high or low outliers, etc., the preliminary results (based on the systematic data only) are output before the final results.
Output options allow for printing summary tables for multistation applications (Figure 2a and 2b) or to suppress unwanted printout. There is also an option to output statistical summary records for each station analyzed.
HEC-FFA was formerly called HECWRC. The name was changed to HEC-FFA, or simply FFA, with this release to be more in keeping with other HEC computer program names, and to go back to something closer to its original name. The new program follows the same procedures as HEOWRC, but many of the routines were rewritten in a top-down, structured-program style. This was done to ease further improvements and maintenance. Some new capabilities were also added to the program as noted below.
HECWRC was originally a modification of the computer program FREQFLO written by Leo R. Beard and David Ford (Center for Research in Water Resources, the University of Texas at Austin) under contract to the Water Resources Council (WRC). The original program (FREQFLO) and documentation may be found in Appendix 13, Guidelines for Determining Flood Flow Frequencies, WRC, Bulletin 17, March 1976. The latest version of the Guidelines (Bulletin 17B) does not contain computer program documentation.
The input and output formats of the original program were restructured, a number of improvements and options were added, and a few computational errors were corrected.
DIFFERENCES BETWEEN HECWRC AND HEC-FFA
Change in High-Outlier Specification. Bulletin 17B provides for a historic weighting adjustment for all peaks above a threshold value. HECWRC determined the threshold value as the minimum of the specified high outliers or the historic peak flows. If the minimum specified historic flow determines the threshold, then all systematic flows greater than the threshold, even those that are not designated as outliers, will be adjusted for the historic weighing. FFA differs from HECWRC in that all peaks above a specified threshold (HITHRS) will be adjusted via historic weighing. If a historic peak is less than the specified threshold, then that peak will not be used to estimate the frequency curve except for determining the historic period. If the threshold is not specified, then FFA chooses the threshold as the minimum historic peak as in HECWRC.
Changes in Low-Outlier Specification. The low-outlier value will be calculated according to Bulletin 17B procedures and will automatically eliminate all peaks below this value, unless otherwise specified. In addition, a low threshold (LOTHRS) base can be specified by the user, and this value will override the base determined by 17B procedures. This differs from HECWRC in that, a lower threshold could be specified, but could not lower the base below the Bulletin 17B low-outlier base. This prevents including a Bulletin 17B computed low outlier in the analysis.
Plotting Positions. In HECWRC, historic records were added to the bottom of the list of systematic records regardless of their chronological order. FFA compares the first historic peak with the first systematic peak, if the historic peak has an earlier date, all the historic peaks are placed before the systematic records in the plotting positions table, otherwise all historic peaks are put at end of the systematic record. This affects only output display; if does not affect the computed frequency curve or the plotting positions.
Conditional Probability Adjustment. In the event of zero-flow years, the preliminary frequency curve now is calculated using preliminary statistics and is printed. The conditional probability adjustment is then made on that curve, then printed out. Thus, the frequency curve always corresponds to the statistics below ft. (See test no. 5.) The conditionally adjusted curve is readily apparent because "-1"'s fill the expected probability and confidence limit columns.
Input of Frequency Curve Statistics. FFA now allows the user to read in statistics, either with or without flow data, and compute the frequency curve ordinates.
Printer Output Format. An extended character set is now used to build the output tables. If a printer is used without this capability, set the IEXT variable to "1" (see J2 record).
Output to HEC-DSS. Computed frequency curves, confidence limits, and plotting positions can be output to the HEC Data Storage System (HEC-DSS). DSPLAY and other DSS programs may then be used to manipulate and plot the data.
Output to HP Laser Jet. FFA can write the Hewlet Packard printer codes to a file. This file can be printed on a HP Laser Jet Series II (or HP compatible) printer to produce a report quality frequency curve.
FFA Menu Operation. A Menu program was written for FFA similar to ones provided In other HEC software packages. The FFA Menu program carries out the DOS commands to identify files, call the COED editor, execute the programs FFA, DSPLAY, etc., and display the output results using the FGRAPH utility.
FGRAPH Utility. The FFA package includes a utility program that plats the final frequency curve of an analysis to the screen. The FGRAPH utility is limited to screen output only and can be called directly from the FFA Menu.
Use of CD ROM Data. FFA can read a peak flow file generated by the Earth Info CD ROM package. To use this capability, refer to CD record description In Appendix B and the example in Appendix D.