Streamflow forecasts are fundamental to the effective operation of flood control reservoirs and levee systems. Streamflow forecasts provide hydrographs (time of flood flows and volume) and peak flow estimates to inform reservoir operation. Forecasts may also support emergency operations by providing estimates of the timing and extent of expected hazardous or damaging flood conditions in the form of inundation mapping. Streamflow forecasts can be developed using a precipitation-runoff model of the watershed. The forecast simulations are based on recent meteorologic and hydrologic conditions in the watershed (including snow pack and soil moisture conditions). The forecast simulations may also incorporate predicted future meteorologic conditions. HEC-HMS includes a forecast alternative simulation type with features that were specifically added for flow forecasting. The forecast alternative simulation type differs from the basic simulation run in that it is meant for real-time forecasting with quick calibration methods built in. Forecast uncertainty is assessed by changing parameter initial states and meteorologic boundary conditions.

General Process

The objective of precipitation-runoff modeling within a flood forecast is to provide reliable hydrograph estimates in a time-sensitive manner. The modeling effort must produce accurate results, which are used to provide information about flooding magnitude and are used to inform future operation of reservoirs and other flood risk management projects. The main components of either developing an HEC-HMS precipitation-runoff model or applying the HEC-HMS model to a flood forecast are listed below.

  1. Develop an HEC-HMS model for the watershed and calibrate the model to a range of precipitation-runoff events. Model development includes delineation of subbasins and reaches, selection of a meteorological model type that will be used for model calibration, validation, and forecast alternatives (gridded versus rain gage approaches), and initial parameterization of the model. The calibration events should include the type of events that the model will be applied to when used in forecasting mode. Additional events should be simulated to verify the model and selected parameter sets. Finally, document model development, model parameter information, and modeling assumptions that would be useful for others when applying the model to a flood forecast.
  2. Once the basic HEC-HMS model has been configured, then it can be customized for performing forecast simulations. Part of the process for configuring a forecast alternative simulation is to create zone configurations and select basin model elements within each zone for parameter adjustments. During the development of a forecast alternative, it is common to group hydrologically similar subbasins or reaches together so that parameter adjustments can be applied uniformly across all elements in the group. HEC-HMS includes zone configurations to group similar features together. A parameter adjustment can be entered once and applied to all of the subbasins/reaches within the zone. A local element override can be used if a subbasin/reach within a zone diverges in behavior from the rest of the subbasins/reaches in that same zone. More information on creating zones is described here.
  3. Choose the relevant basin model (e.g. dry, average, wet, summer, winter, etc.) and meteorologic model (e.g. rainfall only, rainfall with snowmelt, no future precipitation, with future precipitation, etc.) to be used in the forecast alternative. A forecast alternative is constructed using the same initial conditions and parameters found in the selected basin model. Parameters can be adjusted using zonal editors, blending, and baseflow and reservoir initialization editors.
  4. Choose a time interval to compute the forecast alternative. The time interval must adequately capture the hydrograph peak, and in some cases the total runoff volume is needed for reservoir operations.
  5. Select appropriate zonal configurations for loss rate parameters, transform parameters, baseflow parameters, and/or routing parameters. Zones may be configured based on existing stream gage locations, soil types, etc. that group elements which will have similar parameters.
  6. Determine a time window for the forecast alternative. The time window includes a start date and time, an end date and time, and a time of forecast. Typically, the time of forecast represents the last available time for meteorologic observations (future meteorologic conditions can be included within the forecast alternative). The simulation can be started hours, days, or weeks before the time of forecast, depending on the watershed and/or meteorologic conditions. The difference between the start time and the time of forecast is called the look back period. The look back period can be used to compare observed data with the computed results to help with real-time model calibration. Model calibration is where model parameters (including initial conditions) are adjusted to improve model results. The end date is typically only a few days into the future. The length of the look back period and forecast period depends on the response time of the watershed. Small watersheds may only need a two to three day look back period for a one day forecast. A larger watershed may need a seven day look back period for a three to four day forecast period. Some special simulations may use a forecast period exceeding a week in duration, like a snowmelt simulations in larger watersheds.
  7. Set boundary conditions for a flood forecast. The meteorologic model defines the boundary conditions for the simulation. Boundary conditions include precipitation, evapotranspiration, and snowmelt. Observed flow data can also act as a boundary condition if blending is used, or used as flow at a source element or at a reservoir element with the specified release option turned on.
  8. Initialize conditions for a flood forecast, including baseflow and reservoir storage. Information on using the Forecast Initial Baseflow and the Forecast Reservoirs editors can be found here.

Baseflow usually dominates flow composition at the beginning of a forecast simulation. Observed flow provides one good source for calculating the baseflow initial condition. When there are multiple subbasins above the observed flow location, a ratio can be applied to the observed flow to calculate the baseflow initial condition for each subbasin. The Forecast Initial Baseflow editor includes methods for initializing subbasin flow contributions, and it uses the baseflow zone configurations to assist in organizing data entry.

Pool storage is a key factor in reservoir discharge. The initial condition must be specified for each reservoir at the start of the forecast simulation. When using forecast alternatives, initial elevation can be selected as the initial condition for each reservoir in the basin model. The Forecast Reservoirs editor provides a choice of methods to estimate discharge data for the forecast period. Reservoir storage can also be initialized using observed elevation data with this editor. The Forecast Reservoirs editor also contains an option for updating the reservoir elevation at the time of forecast using an observed elevation.

  1. Calibrate the real-time model efficiently - Subbasins and reach parameters are initialized to default values using parameter values from the existing basin model. Parameter values in the elements may be adjusted for the forecast without altering the original parameter values in the basin model using the features described below. The order of parameter adjustment within a forecast alternative is 1) parameter values are taken from the basin model, 2) adjustments are made based on information defined within the zonal editors, 3) individual element overrides are applied, and 4) initialization from observed data (initial baseflow and reservoir stage) are applied. Information about forecast parameter adjustments, slider adjustments, and blending can be found here.
  2. Forecast Parameter Adjustments - Zone parameter adjustments are entered using the parameter editors for the forecast alternative. A separate parameter editor is available for each loss rate, transform, baseflow, and routing method used in the basin model. The following figure shows the Forecast Deficit and Constant Loss parameter adjustment editor. Adjustments are set for each zone using the upper half of the editor, while individual element parameter overrides can be applied using the bottom half of the editor.
    Forecast Deficit and Constant Loss Parameter Adjustment Editor
  3. Slider Adjustments - Slider adjustments are an alternate way to view and change a zone parameter adjustment or an element parameter override. Instead of typing a new value for the adjustment, a slider can be moved to quickly obtain a new adjustment factor.
  4. Blending - Blending provides a method for systematically using observed flow to adjust computed streamflow. If blending is used, observed flow replaces computed streamflow for the look back period, and a transition from observed flow to computed streamflow is applied starting at the forecast time. Three different options are available for performing the transition between observed flow and computed streamflow: step, taper, or none. More information on the forecast blending options can be found here.
  5. Computing to a Computation Point - Any hydrologic element in a basin model can be designated as a computation point. A forecast alternative can be computed to a computation point rather than computing the entire model each time a change is made. One approach to calibration is to start at the most-upstream element with observed flow; after adjusting parameters above the upstream point, calibration efforts are focused on the drainage between the upstream point and the next point downstream. The computing to a point feature in HEC-HMS supports this calibration approach.
  6. Model Computational Efficiency - The HEC-HMS modeling software is designed to be computationally efficient. Only components with data changes since the last compute will be recomputed. This design provides greater efficiency as changes are made to elements further downstream in the basin model, since changes made to upstream elements affect results in each element downstream. The forecaster may choose to force all components to be recomputed if it is desired (holding down the control key when pressing the compute button forces all elements to be computed).
  7. View and analyze results; extract results for further analysis. The output from HEC-HMS gives the forecaster stage and flow hydrographs from each hydrologic element. These results can be imported into additional modeling software such as HEC-ResSim for reservoir operations and HEC-RAS for inundation mapping.

Authority and Procedural Guidance

The following are USACE guidance on flood forecasting:

  • ER 1110-2-240 (USACE, 2016) states that flood forecasts may be used for planning future operations; however, actual releases should be determined following operation in the water control manual and based on observations, to the extent practical. Flow forecast alternatives can be used to determine impacts of different scenarios on reservoir operations and streamflow.
  • EM 1110-2-3600 (USACE, 1987) discusses methods of hydrologic analysis which are applicable to management of water control systems. Forecasts "simulate the continuous natural response of hydrologic and river systems, combined with the effects of project regulation on conditions of streamflow and river stages". Forecasts can evaluate different alternatives for streamflow and project regulation, and forecasts can reduce damage in unprotected areas, making them an important element in the management of water during floods.

Forecasting Procedures

To meet the objectives of a flood forecast analysis, peak flow, total runoff volume, hydrograph timing, and peak reservoir stage are often required. These values are calculated for a look back period and then a forecast period (typically 3-5 days into the future). In general, the modeling process includes two parts: model development and flood forecasting. Model development typically utilizes the simulation run compute type where the time window is set to a historic flood event and the model is calibrated to observations from the historic event. A separate basin model is used for each calibration event (each basin model has a unique set of parameters). Flood forecasting uses the forecast alternative compute type which includes forecast specific parameter editors that efficiently estimate initial condition and allows for quick adjustments to the base model. The base model is used as a starting point for real-time calibration.

Model development steps include:

  1. Select appropriate methods to represent watershed,
  2. Collect watershed data and characteristics,
  3. Utilize regional studies and equations to estimate parameter values,
  4. Calibrate the model using historical data,
  5. Validate the model using historical event data, and
  6. Develop model parameter sets for different watershed conditions (such as dry, average, and wet conditions). These parameter sets can be used to initialize the flood forecasting model and will then be refined for the flood event being modeled.

Steps for flood forecasting include:

  1. Configure the forecast alternative,
  2. Calibrate the model to observed data in the look back period by adjusting model parameters,
  3. Import future precipitation, and
  4. Run the calibrated model to develop a flow forecast.
  5. Analyze and view the results. Model results can be extracted for additional model applications such as HEC-ResSim and HEC-RAS.

A case study is presented to illustrate the steps for developing an HEC-HMS forecast alternative and applying the forecast alternative to generate multiple forecast scenarios. The case study includes a description of the decisions required when configuring the model and the types of watershed information needed to build the model. Selection of methods for modeling hydrologic processes is included as well as model calibration and validation. Finally, the case study shows how to create a forecast alternative simulation and steps for applying the forecast alternative for hydrologic simulation.