HFAM Applications Guide and Screenshots

What can be done with HFAM? The following are examples of continuous hydrologic simulation applications for operation and design of water resource facilities, and for planning or environmental analysis of watershed behavior.

HFAM Screenshots illustrate commonly used results, and typical data requirements for operating HFAM are included. This summary is a 'quick tour' of HFAM. Images in these notes are from HFAM1.1.

Applications of continuous hydrologic simulation methods require time series of continuous hydrometeorologic data, and physical data on watersheds and facilities. To make handling these data as straightforward as possible, HFAM includes a comprehensive data base for hydro-meteorologic time series.

Calibration of HFAM, matching continuous simulated and observed streamflows and snow depths, is done for all applications. Calibration is essential. Hydrologic models are not reliable unless they represent continuously observed watershed behavior in the region where they are applied. HFAM includes a variety of tools for calibration of model parameters.

Design of Water Resource Facilities

Designs are investigated with HFAM Analysis runs. These are long-term simulations of 30 to 80 years, depending on the available historic meteorologic data in the watershed. Watershed behavior is modeled with and without new facilities. Analysis runs show continuous streamflow, snowpacks, generation, diversions, and reservoir levels or volume.

Statistical summaries of flood frequency, watershed soil moisture, snowpacks, streamflows, hydro generation, reservoir releases, and other time series are available.

Typical results screens for continuous time series from Analysis runs are;

Streamflows at different locations in a watershed, including outflow from Sapphire Reservoir



Snowpack in March 1979, in Ac-Ft of water equivalent for the watershed above Sapphire Reservoir

This plot shows the total snowpack water content above a reservoir during the spring snowmelt season. Snow water equivalent at a selected location and elevation can be plotted for comparison to a snow course or snow pillow measurement. Data required for snow accumulation and melt simulation are hourly air temperature (derived if necessary from max-min daily), wind movement, and solar radiation (or cloud cover).


Powerhouse Operations in May 1979 at Sapphire Reservoir

This display shows a powerhouse that is operating on a user defined weekly schedule. Between May 5th and May 11th HFAM defaults to full powerhouse capacity to avoid spill. Generation schedules can be set as time series or as annual patterns.


Reservoir Inflow and Outflow during a Probable Maximum Flood, Rangeley Lake , Maine

A common design problem is spillway capacity analysis, for new or existing dams. When HFAM is calibrated probable maximum meteorologic conditions can be modeled. HFAM long-term analysis runs made for the period of available historic hydrometeorologic records provide the most severe, or runoff producing, watershed initial conditions for the probable maximum event. Probable maximum meteorologic conditions vary seasonally, as do extreme runoff producing initial conditions. The probable maximum flood for a river is the largest flood found by combining probable maximum meteorologic conditions and watershed initial conditions in any season.

 Statistical summaries of watershed behavior and facilities operations are available from HFAM long-term Analysis runs. Results produced include flood frequency, and statistical summaries for soil moistures, snowpacks, generation, reservoir releases, and other time series. Typical results screens for statistical summaries of long-term HFAM Analysis runs are;


Flood Frequency for Release from Sapphire Reservoir, with current flood control policies, 1970 to 1994

Flood frequency for alternate flood control policies, or for the watershed prior to construction of the dam can be plotted.Flood frequency from long-term HFAM Analysis runs is compared to flood frequency at stream gages. HFAM will also create conditional, near-term flood frequency, e. g. flood frequency over the next six weeks.


Exceedance Probability of Annual Hydro Generation, by water-years for the period 1970 to 1994 at Sapphire Reservoir

These results are used in studies of reservoir or power plant expansions, or in studies of the long-term effects of new constraints for reservoir elevations or instream flows.


Probability Maps of Snowpack Exceedance Probability above
Sapphire Reservoir, plotted in Ac-Ft of water equivalent for the winter-spring season, November to June

Current year snowpack water equivalent for any date can be compared to these long-term exceedance probabilities based on historic records.


Operation of Water Resource Facilities

Operational Decisions depend on what will happen over relatively short future periods of time, from a few hours to one or two years. They are investigated with HFAM Forecast and Probabilistic runs. Watershed behavior over relatively short future period of time is conditional; it is dependent on current snowpacks, soil moistures and perched groundwater levels.

Typical results from Forecast runs are streamflows, or reservoir outflows and hydro generation;


Reservoir Outflows at Sapphire Reservoir, December 4 to 18, 1966.

Results are based on a meteorologic forecast for the December 4th to December 18th period. These results are used for operations over the next day or several days. The model runs are repeated as real-time data is collected and as weather forecasts change. Forecast runs use only one meteorologic forecast, and they typically show future conditions for only a few days. Probabilistic runs use many 'alternate' future meteorologic conditions based on historic weather at that time of year, and will show future conditions for weeks or months.

Typical results from Probabilistic runs are;

Probability map of reservoir volumes over the next 7 months, for a specified generation schedule

In this example the current date is August 17th, 1997. A fixed weather forecast is used between August 17th and Ausust 24th, 1997, so alternate weather samples do not begin until August 25th. The future release/generation schedule that is used in this model run is too large, and the reservoir has greater than 50% probability of reaching minimum pool by February 1st, 1998.


Conditional or Near-Term Flood Frequency for the period of a Probabilistic Run, Rio Nare at N. Viento, for a 7 month period starting August 17th, 1997

Conditional flood frequency can be significantly different than traditional or long-term flood frequency. Reservoir operations benefit by correctly anticipating the runoff and flood conditions for the current time, rather than using long-term averages.


Generation at San Lorenzo , August 17th to December 31st, 1997. Generation is based on the watershed and reservoir initial conditions of August 17th.

Generation forecasts are used to plan energy purchases or sales, and to schedule hydro energy in combination with energy from other sources.


Planning and Environmental Analysis

The hydrologic effects of changing land use in watersheds, often due to urbanization, forestry or agriculture, are investigated with HFAM Analysis runs. The hydrologic effects of changes to stream channels such as building levees, or introducing spreading basins to increase infiltration to groundwater aquifers, can be investigated. The HFAM model parameters that represent watershed characteristics can be adjusted to represent altered conditions.

Land use changes often increase or decrease runoff volumes and reservoir yields, and cause runoff flowpath changes. Changes to channel reaches, e.g. increased energy loss due to friction or increased cross-sections, will alter downstream peak flows and flood frequency.

Runoff on Hydrologic Flowpaths, in a Land Segment in Rio Nare, July, 1991


Hydrometeorological Data Management

Typical data management operations are reading or writing files of time series data, in commonly used hydrometeorologic data formats.

Interactive display and editing of hydrometeorological data.

HFAM includes three independent hydrometeorologic data bases for forecast, real-time and historic data. Each of these data bases handles all types of data series on hourly or daily time steps. Data bases are indexed using a 4-digit year to avoid Year 2000 ambiguity. Historic rainfall and potential evapotranspiration data (usually estimated from Pan records) are used in all watersheds. Additional data series (air temperature, wind movement, and solar radiation/cloud cover) are used where snow accumulation and melt are significant. Historic streamflows are used for HFAM calibration.


Simulation Model Calibration

Simulated and observed comparisons are made using HFAM Analysis runs for streamflows, snow depths and water equivalent, reservoir stage or volume, or other data as available. Simulated and observed flood frequency and flow duration curves are compared. HFAM includes tools to compare mean monthly simulated and observed flow volumes, and plots alternate measures of 'differences' between simulated and observed monthly flows.

Simulated and Observed Streamflow, Rio Nare at N. Viento , July 1991.

Additional HFAM screens (not included) show precipitation, rainfall or liquid water reaching the land surface, runoff from the land surface, and runoff in river channels in the same units (Ac-ft or Mm3).

Monthly, Annual, and Period of Analysis Comparisons for Simulated and Observed Flows, Rio Nare at N. Viento

HFAM model calibration is based on solving the water balance equation for a watershed continuously. All hydrologic flow paths that contribute to streamflow are modeled. Timing of runoff during the year is one measure of the model's ability to represent continuous soil moisture and perched groundwater processes in response to seasonally variable precipitation and potential evapotranspiration time series.


Flow Duration Curves, Simulated and Observed, for the period 1984 to 1996 in Rio Concepcion

Flow duration curves plotted over the period of record at a streamgage are useful for calibration. These plots are also done seasonally and monthly. Duration curves are sensitive to simulation model flowpath assignments; lower flows are largely base or groundwater flows. Note: The flow duration plot above was created with Excel, not HFAM 1.1. HFAM II has internal flow duration plots.


Peak Flow Frequency, Simulated and Observed, Diamond River nr. Wentworth, N. H., 152 sq. mi. for the period 1942 to 1993.

In HFAM calibration all aspects of watershed behavior; i. e., continuous hydrographs, runoff volumes, monthly and seasonal runoff distributions, flow duration curves and flood frequency, are simulated over many years. Model parameters are adjusted until the model and prototype, HFAM and the watershed being studied, behave very similarly. With comprehensive long-term continuous calibration HFAM will model watershed behavior for meteorologic sequences that were not included in the calibration period.

Return to Index

Hydrocomp, Inc.