Development and Application of Coupled Hurricane Wave and Surge Models for Southern Louisiana

DissertationCoastal Louisiana and Mississippi are especially prone to large hurricanes due to their geographic location in the north-central Gulf of Mexico. Several recent hurricanes have devastated the region, creating complicated environments of waves and storm surge. Katrina (2005) and Gustav (2008) made landfall in southeastern Louisiana, and their counter-clockwise winds pushed surge onto the Louisiana-Mississippi continental shelf, into the low-lying wetlands surrounding the Mississippi River, and over and through the levee system that protects metropolitan New Orleans. Rita (2005) and Ike (2008) passed farther to the west, moved across the Texas-Louisiana continental shelf, and created surge that flooded large portions of southwestern Louisiana.

These hurricanes demand detailed hindcasts that depict the evolution of waves and surge during these storm events. These hindcasts can be used to map the likely floodplains for insurance purposes, to understand how the current protection system responded during each storm, and to design a new protection system that will resist better the waves and surge. In addition, the resulting computational model can be used to forecast the system’s response to future storm events.

The work described herein represents a significant step forward in the modeling of hurricane waves and surge in complicated nearshore environments. The system is resolved with unprecedented levels of detail, including mesh sizes of 1km on the continental shelf, less than 200m in the wave breaking zones and inland, and down to 20-30m in the fine-scale rivers and channels. The resulting hindcasts are incredibly accurate, with close matches between the modeled results and the measured high-water marks and hydrograph data. They can be trusted to provide a faithful representation of the evolution of waves and surge during all four hurricanes.

This work also describes advancements in the coupling of wave and surge models. This coupling has been implemented typically with heterogeneous meshes, which is disadvantageous because it requires intra-model interpolation at the boundaries of the nested, structured wave meshes and inter-model interpolation between the wave and circulation meshes. The recent introduction of unstructured wave models makes nesting unnecessary. The unstructured-mesh SWAN wave and ADCIRC circulation models are coupled in this work so that they run on the same unstructured mesh. This identical, homogeneous mesh allows the physics of wave-circulation interactions to be resolved correctly in both models. The unstructured mesh can be applied on a large domain to follow seamlessly all energy from deep to shallow water. There is no nesting or overlapping of structured wave meshes, and there is no inter-model interpolation. Variables and forces reside at identical, vertex-based locations. Information can be passed without interpolation, thus reducing significantly the communication costs.

The coupled SWAN+ADCIRC model is highly scalable and integrates seamlessly the physics and numerics from deep ocean to shelf to floodplain. Waves, water levels and currents are allowed to interact in complex problems and in a way that is accurate and efficient to thousands of computational cores. The coupled model is validated against extensive measurements of waves and surge during the four recent Gulf hurricanes. Furthermore, the coupling paradigm employed by SWAN+ADCIRC does not interfere with the already-excellent scalability of the component models, and the coupled model maintains its scalability to 7,168 computational cores. SWAN+ADCIRC is well-suited for the simulation of hurricane waves and surge.

JC Dietrich (2010). “Development and Application of Coupled Hurricane Wave and Surge Models for Southern Louisiana,” University of Notre Dame.

Wind Drag Based on Storm Sectors

The animation below shows our latest hindcast of Katrina in southeastern Louisiana. This hindcast is on our new SL16 mesh, which has 5M vertices and 10M triangular elements. This increased resolution allows us to better resolve the generation of waves in the Gulf, the dissipation of waves as they interact with the changing bathymetry and bottom friction on the continental shelf, and the fine-scale features of the circulation in the channels and marshes of the region.

Contours of water levels (m) and vectors of wind velocities (m/s) during Katrina (2005).

Contours of water levels (m) and vectors of wind velocities (m/s) during Katrina (2005).

While finalizing our hindcasts of the four recent, major hurricanes to impact southern Louisiana (Katrina, Rita, Gustav and Ike), we implemented a variable wind drag coefficient that depends on wind speed and also location relative to storm movement. This work follows closely the azimuthal dependence of wind drag observed by Mark Powell (2006).

Continue reading

Swell Propagation on Fine Meshes

We encountered recently a problem that may be of interest to other SWAN+ADCIRC users. As we were validating our hindcast of Hurricane Ike, we noticed that the swell propagated slower as we increased the resolution in our meshes. This behavior was caused by the SWAN convergence criteria we had selected; we were not allowing the SWAN solution to converge sufficiently. By selecting criteria that are closer to the default settings recommended in the SWAN User Manual, we were able to fix our problem.

We first noticed the problem while validating the computed SWAN significant wave heights against the measured results at eight gauges deployed by Notre Dame assistant professor Andrew Kennedy. These gauges were deployed in the days before Ike’s landfall along the Texas coastline, as shown below:

Locations of Kennedy gauges during Ike (2008).

Locations of Kennedy gauges during Ike (2008).

Ike’s track is shown with a black line; note how the storm curved northwestward in the Gulf of Mexico and made landfall near Galveston and Houston. The eight gauges are shown with black dots, and are located very near the coastline in 10-20m of water depth. The information measured by these gauges is extremely valuable, especially with respect to wave validation, because the gauges are located nearshore, where the waves would feel the effects of bottom friction and depth-induced breaking.

Continue reading

Integral Coupling of Bottom Friction

Updated 2010/06/16: Revised figures from our Gustav MWR paper.
Updated 2011/08/26: Removed the Google Maps inset.
Updated 2012/09/10: Added the Google Maps inset and updated the equations in LaTeX.

One strength of the integral coupling of SWAN+ADCIRC is the increased accuracy resulting from the communication of the model components during the simulation. The computed wave solution is better because it includes the water levels and currents passed from ADCIRC, and the computed circulation solution is better because it includes the wave radiation stress gradients passed from SWAN. These benefits are described in a manuscript we submitted recently to Coastal Engineering (Dietrich et al., 2011).

We have applied SWAN+ADCIRC to the most recent hurricanes to impact southern Louisiana, including Gustav and Ike (2008). The following example shows the significant heights of the waves generated by Gustav as it moved through the Gulf:

gustav-hs-gomex

Note how the waves are generated in the deeper Gulf and then propagated onto the continental shelf, where they break due to changes in bathymetry and bottom roughness. The associated radiation stress gradients are passed from SWAN to ADCIRC and used to drive currents and set-up. Our new SL16 mesh contains significantly more resolution in all of the wave transformation zones, including mesh spacings of about 4km in deep water, 500m-1km on the entire shelf, and 100-200m in the breaking zones.

However, the water levels, currents and radiation stress gradients are not the only parameters that can be coupled integrally. This page describes how we have also coupled the bottom friction between the two model components. In SWAN, using the friction formulation of Madsen et al. (1988), we can vary the bottom rougness both spatially and temporally, by computing physical roughness lengths based on the Manning’s n coefficients used by ADCIRC.

Continue reading

How to Hot-Start SWAN+ADCIRC

Numerical models such as SWAN and ADCIRC can be started in different ways. The typical start is “cold,” i.e. with all computed quantities set to zero, and then the solution is allowed to develop naturally as the various forcings are applied. However, it is also possible to apply a “hot” start, in which the solution from a previous simulation is used as an initial condition. This page describes how to hot-start the tightly-coupled SWAN+ADCIRC.

As a point of discussion, please consider the following schematic of our hindcast validation of Hurricane Katrina:

Run-Schematic

In our hurricane hindcasts, such as the validations of Katrina and Rita described in papers to be published in Monthly Weather Review, the simulations are performed in two stages. First, a spin-up simulation is run for several days before the start of the hurricane winds, so that the tides and rivers can reach a dynamic equilibrium in the resonant basin of the Gulf of Mexico. Then the solution from that spin-up simulation is used as the initial condition for the hurricane hindcast simulation. For example, for Katrina, we employ an 18-day, tides/rivers spin-up simulation that starts on 07 August 2005, and then we run a 7-day, hurricane simulation that starts on 25 August 2005. The solution from the spin-up simulation is used to hot-start the hurricane simulation.

There are two ways in which the coupled SWAN+ADCIRC could be hot-started during this hindcast. First, it is always hot-started at the beginning of its simulation, at 2005/08/25/0000Z. Although SWAN is not run during the tides/rivers spin-up simulation, because its action is forced entirely by the hurricane winds, we do run ADCIRC during that stage. Thus, when the coupled SWAN+ADCIRC model is employed in the second stage, we must hot-start the ADCIRC half of the simulation. SWAN starts from scratch at the beginning of the hurricane simulation.

Second, we may need to hot-start at some time during the SWAN+ADCIRC stage if it ended abruptly, due to machine failure, user interruption, etc. Instead of re-starting that stage at its beginning, we would rather hot-start in the middle. Then both SWAN and ADCIRC would need to be hot-started, using an intermediate solution as an initial condition for the remainder of the simulation.

Thus, there are two distinct methods in which SWAN+ADCIRC might be hot-started. The first method, in which SWAN is cold-started and ADCIRC is hot-started, would occur in the transition between simulations in the Katrina hindcast. The second method would hot-start both SWAN and ADCIRC, such as at some intermediate time during the second stage of the Katrina hindcast. Instructions for both methods are included below.

Continue reading