Modeling Freshwater Flow Diversions for Coastal Restoration: An Example from Coastal Louisiana

In response to the catastrophic events of Hurricanes Katrina and Rita, the U.S. Congress directed the U.S. Army Corps of Engineers to “conduct a comprehensive hurricane protection analysis and design…to develop and present a full range of flood control, coastal restoration, and hurricane protection measures” (i.e., Louisiana Coastal Protection and Restoration, LACPR). This resulted in interagency efforts to assess and weigh benefits of coastal restoration via freshwater flow diversion.

What are freshwater flow diversions?
Fresh water flow diversions release river water into marshes to simulate flooding of a river onto its floodplain and thus increase hydrologic connectivity. Freshwater flow diversions offer significant nutrient and sediment inputs to marshes that promote both organic and inorganic accumulation of soil.

Although flow diversions provide a tool for combating coastal land loss, the optimization of flow diversion locations and operation has been difficult due to the complexity of data needs for a coupled ecological and hydrodynamic model. These complexities encouraged the development of a simple, screening-level model that includes the effects of vegetation growth and senescence as well as sediment dynamics. This would allow for straightforward examination and optimization of flow diversion feasibility and operational benefits.

Model Selection & Adaptation
The Nutrient and Sediment model (NSED1, Boustany 2010) was developed to compare long term relative benefits of many flow diversion locations and was implemented with an annual time step to provide quick estimates of the potential benefits of diversions. The NSED1 is sufficient for quick estimation of flow diversion benefits and initial screening of alternatives, but the Louisiana Coastal Protection and Restoration program required greater temporal resolution in order to assess not only the relative benefits of diversion locations, but also the effects of diversion structure type, diversion operational regimes, and hydrologic variability. Therefore, the NSED1 was adapted in two primary ways to create the Sediment and Nutrient Diversion tool (SANDv1.0):

• Refinement of the temporal resolution of the sediment model to include daily impacts of the diversion on the marsh. (This was prompted by high temporal variability in sediment processes.)

• The addition of calculation capability for many of the parameters of nutrient and sediment processes to include temporal variance. This calculation capability reduced data requirements and minimized potential input errors.

Components of the Model
Nutrient Benefits – examine the annual nutrient requirements of the marsh relative to the nutrients loaded to the marsh (total nitrogen and phosphorous) to determine the percent of vegetated area strengthened from nutrient addition.

Sediment Inputs – combine the concentration of sediment from source water (typically the Mississippi River) with known diversion rates to determine the amount of diverted sediment. (The modification incorporated daily variation of both diversion and river discharges and increased the temporal resolution to examine diversion structure operation.)

Sediment Retention – identify the fraction of diverted sediments retained within the coastal wetland as a function of wetland geometry, sediment size and flocculation rates, and flow hydrodynamics.

Verification and Application: Caernarvon Diversion and Breton Sound Estuary
The Caernarvon Diversion (operated regularly since 1991) provided an excellent test case for the new model due to the variable discharge inputs and extensive knowledge of current system processes. Typical diversion and river hydrographs for Caernarvon were applied over the period of analysis. Additionally, uncertainty associated with input parameters was examined through a Monte Carlo type analysis.

Figure 2 presents the evolution of land area within Upper Breton Sound from before the diversion was opened (1 November 1990) until the end of 2006 (31 December 2006). The estimated future without project (FWOP) is presented to provide the magnitude of marsh area benefit the Caernarvon Diversion is providing Breton Sound. Vertical lines indicate the beginning of diversion operation and hurricanes making landfall in Louisiana. It is clear that hurricanes create significant disturbances to the system; however, for the purpose of this screening level model, hurricanes are assumed to create no net import or export of sediment over a long planning horizon.


Figure 2. Marsh area prediction for the Caernarvon Diversion from 1990-2006 with observed acreages, model predictions with parameter uncertainty bounds, as well as the Boustany Model predictions

Diversion Optimization
The Caernarvon Diversion discharges Mississippi River water to Upper Breton Sound through five 15-ft box culverts with vertical lift gates which can be used to control diversion discharges to the marsh. When applied to the Caernarvon Diversion, the land evolution model assessed the land gain benefits of six operational and five structural scenarios with near equal annual discharge volumes. This application demonstrates the ability of the model to provide relative benefits of different operational and structural conditions.

Figure 3. Land evolution predictions for multiple operational scenarios at the Caernarvon Diversion

Figure 4. Land evolution predictions for various structure types at the Caernarvon Diversion

Limits and Future Application
The model is limited by some of its assumptions (e.g., net zero effect of hurricanes). Further refinement of model processes and algorithms was undertaken in later projects (Donaldsonville to the Gulf, SANDv.2.0), but further adaptation is encouraged for future applications.

With these limitations in mind, SANDv1.0 served as an appropriate screening tool to screen the number and location of diversion sites for LACPR. The modifications to the model were crucial for examining different operational and structural scenarios and alternatives.

EBA Resources

• Modeling & Forecasting