For ANFAS-IISAS internal use only! Original site:
The info-digest from original www pages follows:
November 28, 1998
Hydrologic Research Laboratory
Office of Hydrology
National Weather Service (NWS), NOAA
Silver Spring, Maryland 20910
The NWS FLDWAV model is a combination of the NWS DAMBRK and DWOPER models.
It is based on the four-point implicit finite-difference numerical
solution of the complete Saint-Venant equations of
one dimensional unsteady flow along with appropriate internal boundary
equations representing downstream dams, ridges, weirs, waterfalls, and other
man-made/natural flow controls
- real-time flood forecasting of dam-break floods and/or natural floods
- dam-breach flood analysis for sunny-day piping or overtopping
Computes outflow flood wave hydrograph from a dam due to spillway,
overtopping and/or dam breach outflows
FLDWAV features (Windows 95 or Windows 98):
(a) the flood may occur in a system of interconnected rivers such as the
main-stem river and its tributaries
(b) levee-overtopping/crevasse flows into and through levee-protected
(c) automatic calibration of Manning roughness coefficients for historical
(d) user selected routing techniques (implicit dynamic wave, explicit dynamic
wave, implicit diffusion wave, level pool) throughout the river system
(e) color graphic displays of model output (discharge/stage hydrographs, peak
discharge/stage profiles along the river, cross sections, stage-discharge
(f) variable dimensioning of arrays which eliminates the problem of exceeding
number of time steps and cross sections associated with DAMBRK applications
Unavailable (next release):
(a) menu-driven interactive data input utility
(b) modeling flow through bifurcated channels (islands) and bypasses
(c) Muskingum-Cunge routing
(d) routing non-Newtonian (mud-debris) unsteady flows
(e) modeling landslide-generated waves in reservoirs
Scheduled on May 4, 2000: New capabilities in thisversion include the Mudflow
option and the Network option
The beta release of FLDINP has been delayed until late May 12, 2000
Version 1-0-0 date: November 28, 1998
Requirements: i486, 4MB RAM, 5-20MB HD, MS-Win95/98
Example (Ex1) Teton dam run time (Pentium 166 MHz): 10 seconds
CURRENT FLDWAV CAPABILITIES:
- Variable Dimensioning - The input data structure has been
arranged in a manner so that array sizes are determined internally based on
the river system. This eliminates the problem of running out of number of
time steps or number of cross sections.
- Multiple Rivers - FLDWAV can model river systems that have a
dendritic structure (first order tributaries). Second order
tributaries may sometimes be accommodated by reordering the system, i.e.,
selecting another branch of the system as the main stem.
- Dam and Bridge/Embankment Failure Analysis -All of the capabilities
associated with dams and bridges have been retained.
- Levee Option - Flows which overtop levees located along either or
both sides of a main-stem river and/or its principal tributaries may be
simulated within FLDWAV. For a detailed description of this option, refer to
the previously mentioned papers.
- Simultaneous Method of Computation - FLDWAV can route
unsteady flows occurring simultaneously in a system of
interconnected rivers. Any of the rivers may have one or more
structures (dams, bridges, levees, etc.) which control the flow and which
may breach if failure conditions are reached.
- Flow Regime - FLDWAV can handle subcritical, supercritical, or
a combination of each, varying in space and time from one to another.
A new computational scheme (LPI) has been developed to model mixed flow
(see New Enhancements section).
- Boundary Conditions - The upstream boundary may be either a
stage or discharge hydrograph for each river. The downstream
boundary choices remain the same as those for DAMBRK and DWOPER. Although the
downstream boundary on tributaries is a generated stage hydrograph, the KD
parameter must be set to zero for these rivers. The KD=1 option is being
reserved for an observed stage hydrograph which will allow diverging channels
(e.g., branches of a river delta) to be modelled more directly. Currently
these channels are modelled by labelling the downstream end of the channel
as the upstream boundary condition and negating the inflow hydrograph which
forces it to become outflow (Q=-Q).
- Initial Conditions - The initial conditions include the initial
water surface elevations (WSEL) and discharges at each of the
read-in cross section locations. FLDWAV can start up in either a
steady-state (not changing temporally) or an unsteady-state
- Computational Time Step - Currently the initial computational time
step must be read in. This value will be used throughout the run period until
a dam breach failure mode is activated. The model will use the smallest value
between failure time step(s) and the initial time step.
- Roughness Coefficients - A Manning n table is defined for
each channel reach bounded by gaging stations and is specified
as a function of either WSEL (h) or discharge (Q) according to a piece-wise
linear relation with both n and the independent variable (h or Q) read in to
FLDWAV in tabular form. Linear interpolation is used to obtain n for values
of h or Q intermediate to the tabular values. Unlike DWOPER, the Manning n
reaches are defined by their upstream-most section rather than their
downstream-most section. To allow FLDWAV to function like DAMBRK, Manning n
tables are duplicated internally such that there is a table at each reach
between cross sections.
- Automatic Calibration - This option allows the automatic
determination of the Manning n so that the difference between
computed WSELs (stage hydrographs) and observed hydrographs is minimized. In
areas where detailed cross sections may not be available, there is an
option (Fread and Lewis, 1986) that will automatically adjust average
sections obtained from topographic maps in addition to the Manning n.
- Printer Output - Although the format may be slightly different,
FLDWAV will display the same data (e.g., echo print of the input data,
hydraulic information, summary of peak data, etc.) as the DAMBRK model.
- Other Options - The following options are in FLDWAV and have not
been altered from the original definitions in DAMBRK or DWOPER. For additional
information, the user is referred to the DAMBRK/DWOPER documentation.
- Low flow filter
- Pressurized flow
- Cross section interpolation
- Floodplain option (sinuosity)
- Metric option
- Off-channel (dead) storage
- Robust computational features
- Local losses
- Wind effects
- Hydraulic radius option
- Lateral inflow/outflow
- Graphical Output Display - A utility (FLDGRF) has been developed to
display output data generated by the FLDWAV model. FLDGRF is a user friendly,
menu-driven, DOS application which is written in C. The following
information is displayed: peak profiles, hydrographs, cross sections, rating
curves, and external boundary conditions. WSELs and/or discharges may be
displayed at any interpolated cross section. Multiple profiles and hydrographs
may also be displayed. Actual cross sections may be displayed showing the peak
conditions. Unlike FLDWAV, this utility is not portable to the
- LPI Mixed-Flow Algorithm - In the Local Partial Inertial (LPI)
mixed-flow algorithm, the first two terms (inertial terms) in the momentum
equation are multiplied by the factor ß, where ß =(1-Fk) in which F is the
Froude number and k ranges from 1 to 10. The value of ß determines
the type of routing that will be used (0 for diffusion, 1 for
dynamic). The diffusion routing technique tends to be more stable
than the dynamic routing technique for certain mixed flows, particularly
those in the near critical range of Froude number. When routing
supercritical flow (ß=0), the error between using dynamic and
diffusion techniques is approximately 1 percent. For subcritical flow,
the power k is used to control the portion of the inertial terms utilized.
When k=10, essentially all of the inertial terms are utilized until the
Froude number exceeds 0.60. (At F=0.8, 0.9, and 0.95, 90%, 65%, and 50% of
the inertial terms are utilized respectively). When k=1, the ß vs. F
relationship is linear where essentially all of the inertial terms are
utilized at Froude numbers near zero, and none of the terms are utilized at
Froude numbers near one. Although a high k value is desired to maintain
accuracy, a smaller value may be needed sometimes to obtain stability. ß is
a local parameter since the Froude number used is for each computational
point. Therefore, portions of the routing reach with low Froude numbers will
be modelled with the inertial terms essentially included, while those
portions with high Froude numbers will be modelled with little or no
inertial terms included.
- Explicit Dynamic Routing - A characteristics-based upwind explicit
scheme has been added to FLDWAV to model instantaneous dam failures
and very rapidly occurring failures with a time of failure less than
3 minutes. This scheme is also applicable to the complicated flows in the
- Multiple Routing - FLDWAV has the capability of using multiple
routing techniques in a river system. Currently, there are four
routing techniques available: dynamic implicit, dynamic explicit, level
pool (storage), and diffusion. Each reach between adjacent
cross sections is assigned a routing technique by the user via the
KRCH parameter. The LPI computational scheme may also be applied to specific
- Kalman Filter - A real-time Kalman filter estimator has been
added to FLDWAV. If a river has stage observations for more than two gaging
stations, the Kalman filter may be turned on to update the predictions
for each time step using observations. This option is applicable for
real-time forecasting only.
- Time Interval Time Step - This option allows the user to specify
multiple computational time steps throughout the temporal range of the inflow
General Release - The following enhancements are expected to be completed in
time for the general release of FLDWAV (approx. Fall, 1995):
- FLDINP Utility - The interactive input program (written in C) is a user
friendly, menu-driven, Windows application that will allow the user to
generate the data file required by the FLDWAV model. FLDINP will have graphics
capabilities to allow the user to display hydrographs, cross sections, and
- NETWORK Option - The multiple channel option (NETWORK) that is currently
in DWOPER will be incorporated into FLDWAV. This option allows the user to
model nth order tributaries as well as channel bifurcations (islands).
- Other Options - The following options are also expected to be added to
- Muskingum-Cunge routing
- Routing flow thru culverts
- Mudflow/Debris flow
- Landslide generated waves
- Multiple movable gates
- Routing losses
- Automatic time step increase
- Future Releases - Additional capabilities that are being developed and
added to FLDWAV in the future include sediment transport and pollutant
In order to produce an acceptable forecast using FLDWAV the model must first
be calibrated by adjusting the roughness coefficients until the computed and
observed stages match at each gage.
The following steps represent the calibration procedure:
- Select an appropriate reach to calibrate, one that is bounded by gaging stations.
- Select the observed floods to be used in the calibration.
- Determine the lateral flows.
- Gather appropriate data (boundary hydrographs, observed stage hydrographs at gages, topographic maps, cross sections, dam information, etc.).
- Plot profiles of channel invert, minimum and maximum water-surface elevations, and flood stage.
- Prepare the cross sections in the FLDWAV format.
- Prepare FLDWAV data set.
- Calibrate the FLDWAV data set (observed stages at gaging stations) for the minimum flood using the automatic calibration procedure.
- Manually adjust (fine tune) the n values.
- Using the best set of n values, use the FLDWAV model to simulate the flows at locations where observed flows are available; and compare the simulated and observed flows.
- Repeat steps 6-10 for the maximum observed flood.
- Simulate an intermediate flood and compare stages and flows.
- Determine the minimum (smallest) flow that the model will allow.
- Increase the maximum flood hydrograph by 50%-100% and simulate to insure cross sections and n tables are adequately defined beyond the flood of record.
- Calibration is done! Add as an operation to NWSRFS.
Q5. My question is regarding the downstream boundary conditionin the FLDWAV
model. How can I simulate critical flow such as occurring at the waterfall
or rapids located at the most downstream cross section?
A5. A single valued rating curve based on the critical flow equation must be
generated and then read into FLDWAV as an emperical rating curve (KD(1)=3)
at the downstream boundary.
Q6. Should I use DAMBRK OR FLDWAV to model debris flows?
A6.The DAMBRK model does handle mud flows. We are in the process of adding
this capability to FLDWAV and it should be available over the Internet by
1. Single Dam with Dynamic Routing (Teton dam failure) (dynamic =Saint-Venant
2. Single Dam with Level-Pool Routing (tok v nadrzi-jazere aj v kanali-odtoku)
3. Unsteady Flow Simulation (breached dam & flooded bridge)
4. Level-Pool Routing, Average Movable Gate, Conveyance
5. Subcritical/Supercritical (5ft/mile / 20ft/mile)
6. Free-Surface/Pressurized Flow (undergroung pipe)
7. Multiple Rivers, Levees (river + 1 tributary, 3 levees/ponds)
8. as 2. with metric option
9. Supercritical Flow Downstream of Dam (upstream=subcritical,
10. Two Dams