It is now possible to execute DMI groups from rules. The user can associate one or two DMI groups with a rule. Whenever the rule is executed, the DMI groups are executed, one as the first step of rule execution, one as the last. Values imported by the pre-execution DMI group are available for use by the rule’s statements. To add the execution of a DMI group, select rule --> add pre-/post-execution DMI group.

An example application of this new feature is executing an external water quality model at each timestep. In this example, the water quality model might take as input RiverWare values which reflect the current state of the system and return a recommended reservoir release value. At each timestep the rule would first execute an output DMI to export current values from relevant RiverWare slot(s), e.g., inflow on a reach. An input DMI then runs the external water quality model (which reads the data from the output DMI) and then imports the resulting recommended reservoir release values.

The behavior of the data_date DMI keyword has been modified as follows: If the data_date is earlier than the slot’s start date (after the start_date keyword has been processed), the slot’s start date is set to the data_date. Similarly, if the data_date is later than the slot’s end date (after the end_date keyword has been processed), the slot’s end date is set to data_date. In either case a diagnostic message is posted.

The DMI now allows spaces in file paths if the user selects the “allow spaces in file paths” option in the DMI editor dialog. Previously the DMI would replace spaces in paths with ‘_’. Note that if users have written their DMI executable to recognize white space as the end of a path, and they begin using spaces in path names, then the executable should be modified.

When all slots are successfully processed by a DMI, an informational diagnostic is generated. Previously this was a warning message. If not all the slots were successfully processed, a warning is still posted.

A new command, Output <outputName>, has been added to the RiverWare Command Language (RCL). The command allows output files to be generated directly from batch mode, so long as the output configuration has already been specified in the Output Manager of the model. Note that Plot Pages cannot be generated. If an Xmgr Data Format File is to be generated, the option to Save to File (rather than Send to Plot Program) must be selected.

RCL has been modified to allow the batch script to set user parameters with the following command:

Multiple user parameters can be set. The parameters and their values must be valid, otherwise the batch script fails. If the model is saved, the parameters and their values are not saved with the model file.

The workspace has been reimplemented in Qt with many new features. A brief overview of the new features is provided below. Please see the workspace documentation in the RiverWare help for more details.

The Run Control dialog has been reimplemented in the Qt GUI interface. After changing the run start/end date, users must click a green check mark to accept or red x mark to cancel the changes before the run can be started.

The Multiple Run Management (MRM) dialog has been reimplemented in Qt and has many new features. These new features are described briefly below. For more detail, see the MRM documentation in the RiverWare help.

The MRM now supports multiple configurations contained in one model file. The Multiple Run Control dialog opens with a list of the configurations saved with the model file. Users must select the desired configuration to edit or run it. This flexibility could be useful for models in which users may want to run an MRM run with two or three different rulesets.

It is now possible to send the output from a MRM run to multiple files, potentially one file for each slot. To do this, users can modify the output control file using the standard DMI syntax: object.slot file=fileName. Previously, the output from a MRM file was always in the form of a RiverWare Data Format (RDF) file which could be used with ExcelWriter.

It is now possible to run an output DMI after each MRM run. This works like a normal DMI, e.g., outputting one file per slot rather than a combined RDF file.

When running multiple traces in MRM, it is now no longer necessary to write an input (or output) DMI for each trace when the only difference in the DMIs is the trace number. To make use of this new feature, the DMI executable can reference the last command line parameter passed from RiverWare: -STrace=traceNumber.

It is now possible to use environment variables when specifying a policy set, thus allowing portability across machines. To use an environment variable, users can directly type in the path to the policy set rather than using the file chooser.

In the default MRM configuration (now called Classic), MRM runs execute once for each combination of each DMI with each index sequential or concurrent run. With the Lite configuration, MRM runs execute once for each DMI-index sequential/concurrent run pair. The Lite configuration is needed to run the CRSS Lite model.

MRM can now call DMI groups. If a DMI group is called during MRM, all the DMIs in the group must be input DMIs or all must be output DMIs.

The diagnostics for table interpolations have been improved. When an interpolation fails, RiverWare now posts a diagnostics which describes the slot name, date (when applicable), and value that was used for the failed interpolation.

The CLEAR_STATE and RESET_STATE diagnostics have been removed from Simulation. These messages were deemed excessive and not useful.

It is now possible to configure the user units for all columns of a table slot at the same time. (This functionality already exists on AggSeriesSlots and other columnar slots.)

The RPL expression in an expresison slot is now only displayed if the expression is currently invalid. Selecting View->Show Expression (Alt+Shift+S) will display / hide the RPL expression.

Values in RPL expression slots which are evaluated “Never” or “Only Interactively” are now no longer cleared at the beginning of a new run. Previously these values were always cleared and would remain cleared until the user interactively evaluated the slot. Users should be aware that values in an expression slot evaluated “Never” or “Only Interactively” could be the result of a previous run, i.e., the run after which the slot was evaluated.

The Export Copy function of the SCT now supports optional inclusion of displayed (non-hidden) row header fields associated with the cells selected for Export Copy. This option is available by first selecting several cells, then selecting Export Copy from the Edit menu, then in the resulting Export to Clipboard dialog toggling the ‘Include Row Headers’ box.

The SCT now has a new ‘Go To’ feature which allows the user to find a slot in the SCT and “jump” to that position in the SCT. By selecting Go To Find Slot..., the user can search for a slot in two ways: by searching for a slot beginning with a specified string, or by searching for a slot that matches a given wildcard pattern (i.e. *Inflow*). If dividers have been added to the SCT, the user can jump to the slot following each of the specified dividers.

In completing the migration to the new SCT, it is now no longer possible to create an old style (1.0) SCT. Now when users select SCT -> New SCT... the new version (2.0) of the SCT is automatically created. When 1.0 version SCTs are loaded and saved they are automatically converted to the 2.0 version SCT.

The SCT no longer displays the hours for models with a timestep of days or longer. This is valid because for daily and larger timesteps the hours are always aligned at midnight (24:00).

RiverWare model file timestamps are now updated each time the model file is saved, rather than the first time after the model file is loaded.

Previously RiverWare always used the ".bak" suffix; now it uses the old version number when a model file is saved with a new version for the first time. For example, when model.gz-, last saved with version 4.5- is saved with version 4.6, the backup name will be model.4.5.gz (rather than model.bak.gz). This ensures there will be a model file compatible with version 4.5 until the user explicitly removes it.

The RPL editor dialogs have been reimplemented in Qt with several new enhancements. These changes are described briefly below. Additional detail is available in the Ruleset Editor Dialog documentation in the RiverWare help.

RiverWare 4.6 includes the initial implementation of the new Qt "Grand Unified Selector" for the selection of Simulation Objects, Physical Slots, and Accounts. All former uses of the old Galaxy-implemented Multiple Object and Multiple Slot selectors are replaced by this new selector, including within the SCT, Diagnostics Settings dialogs, Output Configuration Manager, Snapshot Manager, Multiple Object Method Selector, Edit (user-defined) SubBasins manager, Supplies Manager, Scenario Manager, MRM Slot Store, and the Reservoir Selector (e.g. for ListSlots on the Control Point object).

New features include:

The sedmtr file that is packaged with RiverWare releases for use with CRSS has been extended by copying the first 15 simulations to the end of the file. This extension allows MRM to now perform 100 simulations for CRSS.