FEATURED GOLDSIM
APPLICATION
Hydrologic
Simulation Program – Fortran (HSPF) Water Balance Model in GoldSim
Peter Steinberg
GoldSim Technology Group
psteinberg@goldsim.com
Introduction
The Hydrologic Simulation Program- Fortran (HSPF) has become one of
the most widely used conceptual rainfall-runoff models since it was
developed as the Stanford Watershed Model in the 1960s. Though the
model has had many applications, it has a couple of critical
weaknesses. It is very difficult to use by non-experts, and it does
not support Monte Carlo simulations to
account for uncertainty in its input parameters. To develop a more
user-friendly runoff model supporting Monte Carlo
simulation, the water balance module of HSPF was rewritten in GoldSim.
The Model
The water balance model of HSPF is a conceptual empirical model that
is lumped at the catchment level. About twenty input parameters are
used to define runoff, which occurs as surface runoff, interflow, or
groundwater outflow, as a function of input precipitation and pan
evaporation time series. HSPF has been used at many scales. There
are many applications modeling runoff and stormwater
detention on small land development parcels. It has also been applied to
much larger scales, such as modeling the many catchments that make up the
Cheasapeake
Bay watershed in
the eastern US.
Despite thousands of applications, the model is still very difficult
to use. Input parameters are given in a fixed width text file and
input time series are given in an unwieldy binary file. The model
was rewritten in GoldSim as a user-friendly alternative with uncertainty
analysis features. With GoldSim’s support for dimensions and units,
rewriting the model was very efficient. A number of validation
tests showed excellent agreement between HSPF results and those of the
GoldSim implementation. The model is available in the model library at GoldSim Resource Center
(See Water
Balance for a Pervious Catchment).
Figure 1: Rain fall-runoff response
One of the main advantages of the GoldSim implementation is the
ability to track internal variables’ changes throughout a
simulation. This can really speed up parameterization, as one can plot
time histories of internal variables’ values and more directly see the
impact of input parameter changes than by looking at inputs and outputs
alone (as current HSPF users do).
Another advantage of the GoldSim implementation is the support for
mathematical optimization. In GoldSim, as an alternative to
manually adjusting parameters back and forth, optimization could be used
to calibrate a rainfall-runoff response to real world runoff and
precipitation data, and the optimized model could be used to predict
runoff under a different precipitation scenario.
Perhaps the most important advantage of the GoldSim implementation is
the support for Monte Carlo simulation,
so that error bars can be added to its runoff predictions. This is
especially important for HSPF, as many of its parameters are conceptual
and cannot be measured directly. In addition to parameter
uncertainty, uncertainty in input precipitation can also be
investigated. For example, a GoldSim HSPF model could be coupled to
another recent model library addition, WGEN, a stochastic weather
generator that generates daily precipitation, maximum and minimum
temperature, and solar radiation based on annual and monthly statistics.
© 2010 GoldSim Technology Group LLC. 300 NE Gilman Blvd., Suite 100, Issaquah,
WA 98027,
USA.
You can unsubscribe or subscribe to this Newsletter at our
website.
|
