Sediment Transport Modeling: Calibration, Verification, and Evaluation

Abstract

Numerical modeling of sediment transport in fluvial and estuarine systems can be a reliable way of predicting sediment mobility. If approached naïvely, however, such modeling can produce results that do not have sufficient accuracy or reliability to be useful in decision making or design regarding a range of remediation or stabilization alternatives. It is important to recognize the numerical modeling process as merely one step toward a more complete and balanced understanding of the fluvial or estuarine system in question. Other steps include qualitative and quantitative geomorphic and engineering analyses used to evaluate the accuracy and reliability of numerical modeling as part of a three-level approach to analyze sediment mobility and overall channel behavior and trends. It must first be recognized that attempting to quantitatively analyze sediment mobility involves developing and applying simplified mathematical algorithms to the complexities of continually varying hydrodynamic and sediment transport processes through natural or modified bodies of water. Accuracy in sediment modeling can only be assessed by comparing measured data to model results with accuracy being defined as the model results matching the data within some acceptable band of uncertainty. Reliability of a sediment model is the concept of dependability in reproducing the processes one is attempting to model and implies that a model includes appropriate mathematical expressions that cover the pertinent physical processes of hydrodynamics and sediment mobility. The concept of reasonableness in sediment modeling is the evaluation of results, when compared with other independent analyses in the application of the three-level process, provide an acceptable level of consistency and consensus of conclusions. The importance of modeling software selection, data quality, model calibration, verification, validation, and reasonableness of results are discussed along with two case studies.