For several decades, National Weather Service (NWS) River Forecast Centers (RFCs) have used one-dimensional (1D) river hydraulic models to forecast stages on major rivers and analyze dam break scenarios. RFCs use either the Dynamic Wave Operational Model (DWOPER) or the Flood Wave Dynamic Model (FLDWAV) for operational forecasting. Projects are now underway to transition DWOPER and FLDWAV models into Hydrologic Engineering Center-River Analysis System (HEC-RAS) models and to develop specifications for including wind modeling capabilities in HEC-RAS. However, these projects do not address the inherent limitations of 1D models and the ability to obtain accurate estuary and ocean boundary conditions for 1D models.
In coastal zones where rivers widen and flow into estuaries, 1D modeling assumptions break down. Water levels near river mouths and in estuaries will vary in two dimensions depending on tides, winds, freshwater inflows, and atmospheric pressure (Gong et al. 2007; Walker 2001; Wang 1979). Observed and simulated data show that coastal waters are controlled by various meteorological phenomena such as thunderstorms, sea breeze, coastal fronts, tropical storms and hurricanes (Figure 1). These meteorological forcings work at scales ranging from 1 km for a single thunderstorm to over 100 km wide for hurricane rotational winds (USACE, 2002). Operational modeling of such a complex system requires a physically-based, 2D, or 3D model with the scientific basis to simulate responses to key forcing parameters. Although other NOAA offices run these types of models for estuaries and oceans, they are not always routine, operational runs and linkages between these models and RFC river forecast models are limited. The difficulty in linking to river and estuary models is one reason why the farthest downstream forecast points modeled by RFCs may be hundreds of miles away from the coastal areas (Figure 2). Figures 2-4 shows that large, heavily populated, coastal areas remain outside the domain of NWS routine river forecasts (Crossett et al., 2004).
Lower Mississippi River Forecast Center (LMRFC), Northeast River Forecast Center (NERFC), Mid-Atlantic River Forecast Center (MARFC), and Southeast River Forecast Center (SERFC) have indicated that current 1D model capabilities do not meet all of their forecasting requirements in coastal zones. Specifically, SERFC and LMRFC have attempted to use 1D models in coastal zones and have not been satisfied with the accuracy of their water level forecasts. A logical next step in improving the availability and accuracy of forecasts in coastal areas is to develop the capability to robustly couple 1D and 2D operational models. In doing so, 2D ocean and estuary models would receive freshwater inflows from 1D models, and 1D river models would receive stage boundary conditions from the 2D models (Figure 5). One of the challenges in doing this efficiently is to define where a 1D model should end and where a 2D model should begin.
In this project we intend to evaluate "state-of-the art" two-dimensional (2D) hydrodynamic models for possible use in an operational river-estuary-ocean (REO) forecast system. The evaluation will be done through a pilot study in which several models are implemented and validated over the same domain. Criteria considered in the evaluation will include:
- model accuracy
- computational requirements
- compatibility with NWS operational systems
- licensing costs
- availability of training and support
The study will define the costs and benefits of the 2D models in sufficient detail to aid decision makers on implementation related decisions. The study will not directly make recommendations on policy issues such as whether REO components should be run at RFCs, at an NWS national center, or within NOS, etc.; however, it will provide detailed information to aid in these types of decisions./p>
Candidate models for the evaluation include: 1) the NWS Sea, Lake and Overland Surges from Hurricanes (SLOSH) model; 2) the Advanced Circulation Model (ADCIRC) model 3) MIKE 21 software products developed by the Danish Hydraulic Institute; and 4) the SOBEK model developed by the WL¦Delft Hydraulics in the Netherlands. Researchers in the academia have developed several other forecast models that may also be evaluated.