Adaptive spatially-distributed water-quality modeling: an application to mechanistically simulate phosphorus conditions in the variable-density surface-waters of coastal Everglades wetlands

The Everglades region known as the Southern Inland and Coastal Systems is an important area that supports numerous endangere3d species and plays a crucial role in regulating water-quality conditions in Florida Bay. Taylor Slough is a major feature of this region and represents the primary surface-water pathway for freshwater inputs to Florida Bay. The slough is also subject to intensive flow management under the Comprehensive Everglades Restoration Plan, yet the consequences of such management for water-quality in these oligotrophic and sensitive wetlands are not well understood. A flexible phosphorus water-quality model was therefore developed and tested as an exploratory management tool for the region. Complex local hydrodynamics required that a spatially-distributed hydrodynamic model be used to simulate flow and transport and the USGS model FTLOADDS was selected for this. A user-definable biogeochemical reactive component (aRSE) was then coupled with the hydrodynamic model and the resulting FTaRSELOADDS model was tested against analytical solutions and field data.;Hydrodynamic field testing showed that depth-varying Manning's resistance was important for accurately capturing wet and dry conditions during the experimental period. Conceptual water-quality models of increasing complexity were tested against experimental phosphorus field data. Results revealed that a simple daily averaging method was the best approach for atmospheric deposition of phosphorus, which is a crucial but very uncertain water-quality input. A simple conservative transport model provided the best fit between modeled and total phosphorus concentration data. Similar results were also obtained with a more complex and mechanistically justifiable water-quality model. The adaptability of the biogeochemical component was used to study how additional model complexity affects model uncertainty, sensitivity and relevance by evaluating progressively more complex conceptual models using global sensitivity and uncertainty analyses. The framework applying these methods is suggested as a useful way of evaluating models in general, and deciding upon a relevant model structure when the freedom to dictate complexity exists.