Assessing effects of data limitations on salinity forecasting in Barataria basin, Louisiana, with a Bayesian analysis

Reliable forecasts of salinity changes are essential for restoring and sustaining natural resources of estuaries and coastal ecosystems. Because of the physical complexity of such ecosystems, information on uncertainty associated with salinity forecasts should be assessed and incorporated into management and restoration decisions. The objective of this study was to investigate uncertainty in salinity forecasts imposed by limitations on data available to calibrate and apply a mass balance salinity model in the Barataria basin, Louisiana. The basin is an estuarine wetland-dominated ecosystem located directly west of the Mississippi Delta complex. The basin has been experiencing significant losses of wetland at a rate of nearly 23 km(2)/y. A Bayesian-based methodology was applied to study the effect of data-related uncertainty on both the retrieval of model parameters and the subsequent model predictions. We focused on uncertainty caused by limited sampling and coverage of salinity calibration data and by sparse rain gauge data within the basin. The results indicated that data limitations lead to significant uncertainty in the identification of model parameters, causing moderate to large systematic and random errors in model results. The most significant effect was related to lack of accurate information on rainfall, a major source of fresh water in the basin. The approach and results of this study can be used to identify necessary improvements in monitoring of complex estuarine systems that can decrease forecast uncertainty and allow managers greater accuracy in planning restoration of coastal resources.