Initial beta release:  

Final version release: 

Minor Enhancements

Allowable Skew Coefficient When Computing an Expected Probability Curve

When using the Compute Expected Probability Curve Using Numerical Integration (EMA) option, the allowable skew coefficient was +/- 1.2.  If the adopted skew coefficient magnitude was found to greater 1.2, an error was thrown and the compute was halted.  Now, the allowable skew coefficient has been expanded to +/- 1.414 (Bobee, 1991) and only a warning will be issued within the Report file.

Added the MOVE.3 Record Extension Method from Vogel and Stedinger, 1985

The MOVE.3 computational method from Vogel and Stedinger, 1985 was added to the Record Extension analysis, as shown in the following image:

MOVE.3 (Full Extension) Computational Method

This method allowed for an extension to be performed for every non-concurrent value of the long-term site (N2).  However, a negative consequence of this assumption can be realized when the long-term site has many more years than the short-term site.  When this occurs, the information content of the extended record can be overestimated and the uncertainty of the extension is underestimated.  Thus, within Bulletin 17C, the maximum number of years allowable for record extension was redefined as ne.

More information pertaining to this method can be found here: Record Extension Analysis | Computational Method.  An example application of the Record Extension analysis can be found here: Annual Peak Flow Record Extension.

Bugs Fixed

The following bugs were present in previous versions and have been fixed within version 2.3.1.

Distribution Fitting Analysis Confidence Limits and Expected Probability Curves

In previous versions, confidence limits and expected probability curves computed within the Distribution Fitting Analysis were truly random and not reproducible.  The computed confidence limits and expected probability curves for two analyses with the same dataset, distribution, and distribution fitting method were different.  Confidence limits and expected probability curves are now pseudo random and reproducible for a given dataset, distribution, and distribution fitting method combination.  This bug was introduced in v2.2 and fixed within v2.3.1.

Variable Time Window Option Doesn't Include Perception Thresholds Prior to Start of Systematic/Historical Data

In previous versions, when using the variable time window option within the Bulletin 17 analysis, perception thresholds that were entered prior to the start of the first systematic or historical event were not included in the analysis.  This issue has been rectified such that all data is incorporated when using the variable time window compute option.  This bug was introduced in v2.3 and fixed within v2.3.1.

Selecting a Data Set Prior to Entering an Analysis Name Causes an Error When Re-Opening a Study

Selecting a Data Set prior to entering a Name in the Bulletin 17, General Frequency, Volume Frequency, Balanced Hydrograph, Duration, or Distribution Fitting analyses caused an error when the study was re-opened. Users must now enter a Name prior to selecting a Data Set in these analysis types. This bug was introduced in v2.3 and fixed within v2.3.1.

Correlation Analysis Drops Name, Description, DSS File Name, and Report File Fields After Successful Compute

In previous versions, the Correlation Analysis would occasionally drop user-entered information (e.g., Name, Description, DSS File Name, Report File, and location information) after a successful compute. This bug was introduced in v2.3 and fixed within v2.3.1.

Setting a Start and End Date When Importing from USGS in the Data Importer Caused an Error

When importing daily data from the USGS in the New Data Importer and a Start and End Date were specified, rather than importing the period of record, an error and no data was imported. This bug was introduced in v2.3 and fixed within v2.3.1.

Restrict the Allowable Extension when Using the MOVE.3-B17C Method

When using the MOVE.3-B17C method within the Record Extension analysis, two options are presented for computing the slope and intercept of the linear regression model: Use Most Recent Years or Attempt to Match Skew.  When using either of these methods, the allowable range of extension is limited.  As such, only the allowable years should be displayed, tabulated, and available for selection.  The behavior of the table and plot within the Record Extension tab has been updated to be more intuitive.

More information on these methods is contained here: MOVE.3 (Bulletin 17C) Computation Method Options.  An example application demonstrating the Record Extension analysis and available computational methods is contained here: Annual Peak Flow Record Extension.

Importing General Frequency Analysis Results within a Coincident Frequency Analysis

Previously, only Bulletin 17 analysis results could be imported to the Variable A tab within a Coincident Frequency analysis.  Now, results from both Bulletin 17 and General Frequency analyses can be imported to the Coincident Frequency analysis Variable A tab.  This bug was introduced within v2.3 and fixed within v2.3.1.

Manual Entry Duration Analysis Does Not Use User-specified Frequency Ordinates and Does Not Write Output to DSS

In previous versions, HEC-SSP would not use user-specified frequency ordinates when computing a manual entry Duration Analysis. The standard frequency ordinates were used. In addition, the program was not correctly outputting the results from a manual entry Duration Analysis to a DSS file. This bug was introduced in v2.0 and fixed in v2.3.1.

Distribution Fitting Analysis Equivalent Record Length Computations Did Not Account for Broken Record

Previously, the confidence limit equivalent record length (ERL) computations in the Distribution Fitting analysis did not account for time series with broken records or for the year specification (water year, calendar year, etc.) of annual maximum series data. In both cases, ERL was computed as the difference in time between the first and last values. Now, ERL computations account for missing data in time series data. In addition, ERL for annual maximum series data is computed as the number of annual peaks in the record.

The changes to the ERL computations will lead to changes in the confidence limits when compared to previous versions. These changes will be more pronounced as the amount of missing data in a regular time series increases.