RiverSMART—the RiverWare Study Manager and Research Tool—is an extensive set of integrated computational models and analysis tools for developing resilient management plans for river basins with deeply uncertain future conditions. Using the flexible graphical user interface centered on RiverWare, users can create, execute, and archive large and complex planning studies that explore and compare multiple scenarios. Uncertain hydrologic supplies are represented by ensembles that represent possible future hydrologic variabilities; and multiple ensembles can provide a range of scenarios, each produced with a different projection model and representing a different possible future variability.

The RiverSMART application provides a graphical tool to create combinations of future hydrologic supply ensembles, demand projections, and infrastructure and operating policy alternatives, with each combination of these defining a scenario. Different RiverWare model files can represent infrastructure alternatives, and different RiverWare rule sets can represent alternative operating policies. The policies may include signposts that point to approaching vulnerabilities that trigger dynamically executing changes in operations or other options to mitigate or avoid states of unacceptable performance.

Studies may consider many scenarios, each a combination of ensembles of inputs and other system characteristics; thus, full execution of a study sometimes leads to hundreds of thousands of simulations. RiverSMART sets up and manages the execution of all the simulations in parallel on multi-core hardware or cloud computing resources.

Results are typically model variable outputs and indicators of system performance. These can be output in many formats, including RiverWare native RDF, Microsoft Excel, CSV, and netCDF, and they can be post-processed using various statistical and visualization tools, including the Graphical Policy Analysis Tool, Tableau, or user-supplied R scripts. At the conclusion of a study, RiverSMART can serve as an archiving structure for saving all aspects of the entire study, including the input data, model files, configurations, policy sets, scenario selections, and results, making the study reproducible in the future.

RiverSMART was designed, developed, and is maintained by the Center for Advanced Decision Support for Water and Environmental Systems (CADSWES) at the University of Colorado Boulder to address the needs of complex modeling studies, such as the Basin Studies conducted by the Bureau of Reclamation, to explore options for meeting projected imbalances of future supplies and demands due to changing climate and water uses. The initial development was funded through the Bureau of Reclamation’s WaterSMART Grants to Develop Climate Analysis Tools and was first used by the Colorado River Basin Study; several features are directly influenced by needs identified by that study.

RiverSMART provides a graphical workspace for users to view, control, and fully understand all the steps involved in a RiverWare-based study. A sample RiverSMART workspace is shown in Figure 1.1. Users can add events and intermediate files to the workspace and connect them to show the flow of information. Each icon represents either a process—called an event—or a set of files produced by one or more processes. At the center of each study is the RiverWare event that is both the destination for all inputs and the source of all outputs. The arrows indicate the flow of information.

The following sections provide general descriptions of the features and uses of the available events.

The Hydrology Simulator is used to develop an ensemble of hydrology representing possible future supplies. The simulator uses non-parametric K-nearest-neighbor resampling techniques to generate ensembles of hydrologic traces based on historical data, optionally conditioned on long paleo-reconstructed data using various Markov Chain techniques, or conditioned on downscaled GCM projections to capture increased variability.

The hydrology disaggregators allow the results to be disaggregated spatially—from one to many locations—and temporally—for example, from annual data to monthly data using a pattern.

To create scenarios, users specify alternative inputs, and RiverSMART generates combinations of these inputs. Following are brief descriptions of the possible inputs that can be included as alternatives.

RiverSMART combines all combinations of alternative inputs to produce a list of possible scenarios. From these possible scenarios, users select the ones that are applicable or meaningful. Provided tools can be used to sort and filter the list of scenarios for ease of selection, and unneeded scenarios can be hidden from the displays.

RiverSMART then manages the execution of the simulations. Users can distribute RiverWare Multiple Run Management (MRM) simulations (concurrent runs) to different processors on one machine or create packages that can be distributed to many computers or cloud instances. The overall simulation processing time is thus reduced through parallel processing.

The results of the simulations are output files—typically text-based RiverWare Data Format (RDF) files and Comma Separated Values (CSV) files—that are ready for post-processing.

Once the scenarios have been simulated, users can post-process the results using provided tools or user-defined R scripts.

The built-in post-processing tools include tools that do the following:

In addition, scenarios can be grouped to form scenario sets for additional post-processing. Scenario sets are user-defined groups of scenarios that assist in analyzing and comparing results across scenarios. For example, you can create a set of scenarios that have the same hydrology but different demands and policies, or a set of scenarios with a common policy but differing demands and hydrologies. You can further post-process the scenario sets using GPAT and R script tools

RiverSMART provides additional tools to do the following: