Ensembles are sets of time series data where each member represents a different possible outcome for a specific location and parameter. Instead of running multiple independent forecasts to represent each scenario, an ensemble forecast run groups these runs under a single ensemble identity. Each ensemble forecast run includes a set of individual runs—one for each scenario—handled collectively. Since most programs in the HEC-RTS modeling suite don't natively support ensemble runs, HEC-RTS manages the decomposition of ensembles into individual runs. This approach shifts the burden from the user or model to HEC-RTS, allowing it to apply its own efficiency strategies regardless of model-specific capabilities.
Collections
In HEC-DSS, data sets are grouped using a concept called collections. A collection consists of time series data that share a common location (A & B Parts), parameter (C Part), and time interval (E Part). While any set of matching records can form a collection, ensemble records must also share a common time window to be grouped. As a result, ensembles in HEC-DSS are stored as collections. Currently, this feature is only available for regular or irregular interval, point-based time series—it does not support gridded data. Ensemble time series can currently be accessed only as point-based precipitation data for HEC-HMS and streamflow data for HEC-ResSim.
Collection Naming in HEC-DSS
The collection structure is implemented through a specific naming convention in the F Part of the pathname. Each record in a collection has a unique collection ID prefixed to the standard version label in the F Part. This ID is formatted as C:######|, where:
- C: indicates the start of a collection ID
- ###### is a unique 6-character alphanumeric identifier for the collection member
- | is a delimiter separating the ID from the version label
The portion after the pipe (|)—the version string—is the same for all records in the collection. This ensures consistency while allowing each member to remain uniquely identifiable within the ensemble.
The Ensemble Forecast Processor (EFP) enables you to calculate and apply metrics from ensemble flow data at specific watershed locations. These metrics can then be used to influence other model alternatives within the forecast. The EFP plugin is highly flexible and can be integrated at various points within your forecast workflow to support a wide range of applications.
The Ensemble Viewer allows you to explore ensemble time series data. It also provides tools to display ensemble metrics in different formats—such as a summary time series, individual values per ensemble member, or a single aggregated value.