eWater is pleased to announce the latest Production Release of Source version 5.10. Key enhancements in version 5.10 include improvements to the Operations Tabular Editor, the ability to Hot Start GR4J and Sacramento rainfall-runoff models, significant enhancements to the Calibration Tool, updates to the Environmental Flow Node and Manager and additions to Scenario Input Sets to make it easier to work with large models with significant scenario modelling.
Improvements to Source
Calibration Improvements
The calibration tool has been significantly extended, you can now:
- calibrate Schematic Scenarios,
- calibrate any parameters in the model that are available through Scenario Input Sets, and
- specify relationships between meta-parameters.
This is a significant step forward for the calibration tool, it's currently called Flow Calibration Tool, however, this should be updated soon.
The Calibration tool and Bivariate stats now support multiple parameters for Objective Functions. The Square-root Daily, Exceedance, and Bias function now has three parameters, Alpha (α), Lamda (λ) and Mu (μ). Details of the implementation are here: Bivariate Statistics SRG
These can now be set in both the Calibration tool and when looking at the Bivariate stats in the Results Manager:
Calibration using Parallel runs within Source
When using the Shuffled Complex Evolution optimisation, you can now run multiple instances of Source to improve the speed of the calibration:
Environmental Flow Node and Environmental Flow Manager
There has been a number of improvements in Environmental Flows. We have extended the Environmental Flows node's Spell action. Functions are now available for:
- Number of spells in a season,
- Number of spells for success in a season, and
- Spell duration.
The values of these functions can't change every day during a simulation. The values of the functions are taken from before the start of the season. To allow for the forcing of spells during a season, the values need to be taken allowing for travel time before the start of the season.
There is also a new option to allow the fall phase of spells to occur after the end of the season.
Scenario Input Sets and the Export summary now include more information about Environmental Flow Actions. The path for the existing Scenario Inputs Sets will need to change from Rules → Actions. API details are in RM-16591. Terminology in the User Interface for one of the Initial Values at the start of run has been corrected during the process of developing the full API. Time since last successful season has been changed to Time since last successful spell in a successful season.
The Environmental Flow Node and the Environmental Flow Manager now both support the addition of Notes.
Scenario Input Sets in Node / Link feature editors
When working with large models with significant scenario modelling with Scenario Input Sets, it can be hard to recognise that Input Sets may change the run-time configuration. When there are multiple Scenario Input Sets, we have included Scenario Input Sets in each feature editor to show the number of changes to the current node/link across the Scenario Input Sets:
Splitters can now be considered regulated in Rules-based Ordering
The Option Treat Splitter as a Regulated Supply Node in Rules Based Ordering is now available in the controlled splitter node:
This can be used when connecting an unregulated section of a network to a regulated section. Previously, the confluence node would only allow you to consider branches to be regulated if there was a Storage node above it, this has been changed to allow splitters to operate as a regulated structure and receive orders.
Hot start rainfall-runoff models
When running Rainfall-runoff models GR4J or Sacramento, you can take a Snapshot of their internal state variables, they can then be saved to a file and applied at the start of a model run. This removes the need to do a warmup period for running these rainfall-runoff models. The snapshot files now include catchment name and functional unit names as well as an additional metadata file, e.g:
- rsnap
- rsnap.metadata
Safe Release Capacity
The Save Release Capacity setting on Storage Outlet Paths was previously a static value. This has been changed to allow a function to be set. We changed the internal model processing so that the release algorithm checks the input parameter for every calculation:
Ability to model individual components of distribution loss
You can now model 9 different individual components of Distribution Loss at a Supply Point. This is on a new, optional, separate tab:
River Operations Improvements
The Operations Tabular Editor has new options including Uncheck All for working with the active recorders:
From within View Overrides you can now Clear Historical Overrides:
Operations models now have a summary of Forecast models across the model
Summary of minor changes
- Order Crawler now includes constraint factors. They are specified per owner, so when there are multiple owners in the system, we have included the average across owners. We also resolved a bug when there were multiple owners, the total wasn't always being shown and it was reporting the results for the first owner it found.
- Improvements to the Hydroclimate data transformation tool plugin (version 2.2), details and the new manual are available here: Hydroclimate Data Transformation Tool
- Scenario Input Sets now supports values that can go over one line. To use {MultiLineValue} include it after the = sign, and end it with {/MultiLineValue}. Example: MultiLineValue.txt
- Issues with date formats changing on different computers have been resolved by moving to export based on yyyy-MM-dd HH:mm:ss. We will continue to import data from a variety of formats, however, tools such as Scenario Input Sets, and the Export Summary will now be using this format regardless of the user's computer settings.
- An oversight in the algorithm to determine ownership at the splitter node has been resolved. Ownership at splitters could become negative in rare cases when sharing unallocated flows, it was previously assumed the allocated flow was always less than the actual flow.
- Confluences now handle constraints for Unregulated sections of networks. We introduced changes to allow splitters to be considered regulated in 5.6 - Beta Release (February 2021). For orders to be passed up to the splitter, constraints needed to be propagated down confluences. This improvement makes Confluences sum together the forecasts from each of its two branches when they are both unregulated.
- When using the Shuffled Complex Evolution optimisation function, you can now run multiple instances of Source to improve the speed of the calibration: This can be run through the User interface or through the commandline.
- We no longer support node or link names with trailing white space. For example, nodes and links with names such as "Storage " won't be allowed to open in this version of Source. You will need to fix these names in a previous version of Source. These trailing white spaces were causing errors to occur in the results manager.
- Improvements to the Data Sources Overview
- Auto export results now use the dimension Display Units
- The Simulation log file has more content for Data Sources
- New recorder on confluences: Constrained Historical Order For Long Branch
- New Scenario creator Single Catchment Scenario Builder
- Other minor bug fixes.
Community Plugin Changes
We have moved to .net 4.8.0 which requires Visual Studio 2019. We are planning on moving to .NET 5. The transition to .NET 5.0 won't be a quick process as some components we currently use will no longer be supported, however it does provide a potential pathway to running natively on Linux. Source can currently be run on Linux under Wine, but only natively on Windows.
Results and Configuration Changes
Some results have changed between Source 5.0 and Source 5.4. Changes include how marker constituent processing works in Weirs. Details are available here: 5.10 Results and Configuration Changes from 5.4