The General Frequency analysis allows the user to fit a probability distribution to many different types of data including flow, stage, precipitation, and wind speed (amongst others). Observed data can be in the form of either an annual maximum or partial duration series. When analyzing an annual maximum series, the user can fit one of 14 possible analytical distribution | fitting method choices, like Log Normal | Product Moments or Generalized Extreme Value | Linear Moments. However, the user can also graphically fit an empirical distribution. When analyzing a partial duration series, the user must graphically fit an empirical distribution.

Create a new General Frequency Analysis

  • Click Analysis | New | General Frequency to create a new General Frequency analysis.
  • Name the analysis "Randolph_Elevation".
  • Select the Randolph pool elevation data set within the Data Set drop down menu.

The data set will be named "Randolph Pool Elevation".

  • On the General tab, select the "Graphical/Empirical" radio button within the Distribution panel and the Do Not Use Log Transform radio button within the Log Transform panel. Leave all other options unchanged.
  • Move to the Options tab. Take note of the options that are available within this tab.
  • Move to the Graphical tab.
  • Under the User-Defined Graphical Curve table in the lower right, hit the "Compute Plotting Positions" button.  The ranked data set versus plotting positions will appear in the graph.
  • Enter values of 1498, 1496, and 1494 for the 0.2-, 0.5-, and 1.0-percent chance exceedance ordinates (in descending order) within the User-Defined Graphical Curve table in the lower right corner of the editor, as shown in Figure 1.
    • These values are used to initialize the frequency curve computations.

Compute, Enter Curve, and Visualize the Results

  • Click the Compute button at the bottom of the General Frequency editor. A window informing you that the computations are complete should appear. Click OK to close this window.
  • Use either the Multi Point Edit tool (  ) or manually enter ordinates to better define the entire empirical distribution.
    • To use the Multi Point Edit tool, begin by selecting the tool on the left side of the editor.
    • Then, define the curve by left clicking within the plot window at the desired locations.
    • You should see a line appearing on the plot as you draw.
    • To finish defining the curve, right click.

The Multi Point Edit tool will not extrapolate frequency information; it will only interpolate to the user-defined frequency ordinates specified on the Options tab.  Bold, dashed lines will be displayed on the plot that correspond to the output frequency ordinates defined on the Options tab. These vertical lines help to visually identify the locations where frequency information will be extracted from the drawn curve.

To include additional ordinates, check the box above the ordinates table on the Options tab, then add rows and frequency values where desired. 

  • Once you are satisfied, click the Compute button.
  • Select the Compute User Order Statistics radio button within the Confidence Limits panel. An equivalent record length of 32 years should be initially estimated within the Equivalent Years of Record entry field. Click the Compute button at the bottom of the General Frequency editor. A window informing you that the computations are complete should appear. Click OK to close this window. Investigate the computed curve as well as the confidence limits.
  • Within the Equivalent Years of Record entry field, enter a value of 100 years and click Compute.

Question: What effect did the change in equivalent record length have upon the computed confidence limits? Does this imply more or less uncertainty within the computed results? How will the confidence limits respond when an equivalent record length of 15 years is entered?

The confidence limits corresponding to an equivalent record length of 100 years are narrower than the confidence limits corresponding to an equivalent record length 32 years.  This implies that there is less uncertainty within the results when an equivalent record length of 100 years is used.  When an equivalent record length of 15 years is used, the confidence limits will be wider and imply more uncertainty in the results.

  • Click OK to close the General Frequency editor.
  • Click the Save button (  ) to save the study.

Continue to Task 4. Create and Parameterize a Duration Analysis.

Download the final project files here: Intro_to_SSP.zip