Improving Flood Prediction with Hydraulic River Routing Techniques Len Wright, Ph.D., P.E., D.WRE Tom Hopson, Ph.D.

Outline Benefits of hydraulic routing to improve accuracy of flood prediction Theoretical background Software model survey Recommendations

Reasons for a Flood Routing Model To extend usefulness of any rainfall/runoff estimate (i.e. historical, future scenario, near-term prediction) Estimate timing, attenuation and height of peak flood levels Accurate addition of hydrographs at confluences – the problem of coincident peaks vs. off-setting peaks

Ancillary Reasons for Models Estimating the impacts of change resulting from planned, or unplanned, future conditions Evaluating engineering decisions on planning, design and operation of structural and non structural flood mitigation strategies Water quality and sediment transport modeling

Problems with Attempting to Predict Flood Levels Difficult technical questions to be answered: –WHEN: How long after the rain will the flood arrive? –HOW MUCH? Magnitude of the flow? How deep will it be? Answer is a complex physical interaction between watershed hydrology, channel geometry and network topology: –Watershed response to precipitation: hydrology –Watershed and channel geometry –River network topology: tributary connectivity afftects watershed response and subsequent downstream flooding

Theoretical Background River flow – assumed one-dimensional Two-dimensional models available for estuaries (provides downstream boundary) Governing equations – St. Venant Data requirements – channel and flood plain geometry, roughness, upstream spatial and temporal boundary, downstream boundary (for subcritical flow)

Governing Equations “St. Venant’s” Conservation of mass, energy and momentum May be derived from full “Navier-Stokes” equations Continuity and Energy: Coupled pair of hyperbolic partial differential equations No closed-solution (except for simple cases), variety of numerical solutions exist.

St. Venant Equations Conservation of Mass Conservation of Momentum Steady, uniform Steady, nonuniform Unsteady, non-uniform

Review of Energy Terms

Derivation of St. Venant

Effect of Storage in River Important for flood routing – wedge routing provides opportunity for wave attenuation Non-uniform unsteady flow affects rating curve – flow is not a one-to-one relationship with depth

Wedge Storage Effects on Real Flood Hydrographs Dashed line rating curve shows results of uniform flow assumption Loop is result of acceleration on leading edge of flood hydrograph and decceleration on the falling edge

Simplifications of the Momentum Equation

Internal Watershed Features Man-made drainage improvements affect the way a watershed responds to precipitation – should be accounted for.

Hydraulic Floodplain Impacts Storage and routing on floodplain is critical to accurate prediction of attenuation and timing of flood peak. Results show spatial extent of innundation.

Floodplain Structures Structures and features in the floodplain must be correctly represented. Note landfill and levee in this floodplain.

Cross-sectional View of Simulated Flood

Data Requirements High quality cross section data – elevation versus distance Roughness Longitudinal channel distance Over bank detail – roughness and curvature Structures – bridges, dams, levees, etc

Partial Model Survey HEC-RAS: US Army Corp of Engineers Hydrologic Engineering Center River Analysis System SWMM: US EPA Stormwater Management Model FLDWAV: US National Weather Service Mike-11: Danish Hydraulic Institute

HEC-RAS Both steady state and unsteady solutions available Widely used in the US for floodplain management studies Free Good capability of simulating structures and for doing sensitivity and “waht-if” analyses.

Typical HEC-RAS Cross section Data Window

3-D Graphical Interface

USEPA SWMM 5 Includes rainfall runoff processes as well as routing Full Dynamic Wave routing Structures and control available Designed for urban environmental model Long-term continuous simulation supported

US EPA SWMM Key strength: simulation of surcharged pipes and culverts Performs hydrologic and hydraulic simulation Continuous, long-term simulation or event- based

NWS FLDWAV Good for areas of low- quality or non-existent data Precedence in flood warning systems in North America

DHI MIKE-11 Sophisticated, state of the art modeling environment Dynamic wave, handles many types of hydraulic structures and control rules Licenses are relatively expensive

MIKE 11 is a professional engineering software package for modelling of rivers, reservoirs, flood plains and canal systems It is used for simulation of: Hydrodynamics Water quality & ecology Sediment transport HOMERAINFALL-RUNOFFWATER QUALITY MIKE 11 PRODUCT

KEY APPLICATION AREAS flood analysis and mapping design of flood alleviation systems dam break analysis real-time flood, inflow and WQ forecasting analysis and design of hydraulic structures drainage and irrigation studies optimization of river and reservoir operations water quality issues integrated groundwater/surface water analysis

Flow routing (Muskingum, linear reservoir) Steady state Hydrodynamic modeling Stratified flow Combined 1-D and 2-D simulation FLOODING & HYDRAULICS SIMULATION OPTIONS

1-dimensional model Rainfall-runoff Hydraulics and structures Sediment and water quality Flooding DHI product integration Forecasting MIKE 11 - FEATURES

Databases Topographical Data Time Series Data Water Quality Sediment Transport Hydrodynamics Rainfall-Runoff Advection-Dispersion Flood Forecasting Modules and Databases that Interact Dynamically A MODULAR STRUCTURE

DEVELOP A MODEL – THE BASIC DATA

MIKE 11 GIS is an ArcGIS desktop extension for: facilitating model setup and river network editing in MIKE 11 pollutant load estimations from non-point sources, and time series data analysis MIKE 11 GIS

RIVER NETWORK EDITING IN GIS Generate Cross-section lines Auto generate based on desired distances and river width –Import surveyed lines –Extend existing X-sec lines onto flood plain Generate Actual cross-sections –Extract from DEM –Import surveyed data –View/edit in cross-section editor

City of Canberra, Australia PSEUDO 2D FLOODPLAIN MAPPING

Rainfall – Runoff modelling is the process of transforming rainfall into catchment runoff Runoff in MIKE 11 includes all the lateral inflow from the small streams not represented in MIKE 11 Catchment yield estimation is important in water resources operation, management and planning Forecasting floods/droughts Reservoir inflow Water resources management … and many other applications RAINFALL-RUNOFF MODELLING IN MIKE 11

…while maintaining lumped, conceptual approach where we lack data e.g. groundwater Distributed, physically based where we have detailed info e.g. climate, vegetation and soils Dynamically coupled to MIKE 11 Auto-calibration (AutoCal) MIKE 11 - DISTRIBUTED RAINFALL-RUNOFF MODELLING

It may be worth spending time & money to investigate how reservoir operation can be improved Small changes can make large difference in earnings and welfare RESERVOIR OPTIMISATION

THE SIMULATION-OPTIMISATION FRAMEWORK Optimize  Flood potential / risk  Water deficits (hydropower, irrigation etc.)  Water quality conditions Control variables Simulation model Optimisation algorithm Objective functions Model outputOptimisation targets MIKE BASIN MIKE 11 MIKE SHE AUTOCAL

Simulation of the flow through a dam breach defined as: 1) Predefined in time 2) Calculated using sediment transport formula. t = t 1 t = t 2 t = t 3 Hydraulic Structures Weirs Culverts Bridges Regulation Control Structures Dambreak Failures MIKE 11 – STRUCTURES & DAMBREAK

Model Selection Model selection is a very important step in the process. Fully examine all the possible uses the model could be used for. Collaborative efforts across agencies and organizations may provide greater room for greater project resources. Model should be selected that best handles all purposes (flood warning, water quality, sediment transport, planning and engineering)

Recommendations Data requirements are an important consideration – do project resources warrant a high-level of sophisticated data? For flooding: time-scale of warnings required Economics of data collection should be considered in this evaluation, as is software cost Will the model be used for engineering decision making? If so, HEC-RAS and MIKE-11 (expensive) are better suited for these purposes than FLDWAV. SWMM is better at pipe network simulation if that is important anywhere.

Ancillary Issues Are there other analyses (besides flood warning) that the model could be used? If so include these benefits when making a model selection Model selection is long-term decision Will “What-IF?” analyses be needed? The ability of the model to handle the important scenarios should be evaluated (ie real-time control, structural improvements, etc). Will continuous simulation be used to evaluate sediment transport (e.g. bridge scour)? This will impact the model selection process.

QUESTIONS?

Disaster Management in US is a Growing Field

Hurricane Katrina in US

Flood Warning System