In 2012 we lost Jean Jacques Peters, former engineer of Flanders Hydraulics Research (1964 till 1979) and international expert in sediment transport, river hydraulics and morphology. As a tribute to him we have created potamology, a virtual memorial archive whose aim is to preserve and disseminate his way of thinking and his morphological approach to river problems all over the world.This archive provides four modules, each with its specific information set relevant to Peters’ work. Where available and if not confidential, there will also be access to the full text. In dialogue with Peters’ family we continue to make his life’s work accessible.
Comparison of the PAWN and Sobol’ sensitivity analysis methods for a highly-parameterized hydrological model using SWAT
Khorashadi Zadeh, F.; Sarrazin, F.; Nossent, J.; Pianosi, F.; Van Griensven, A.; Wagener, T.; Bauwens, W. (2015). Comparison of the PAWN and Sobol’ sensitivity analysis methods for a highly-parameterized hydrological model using SWAT, in: E-proceedings of the 36th IAHR World Congress 28 June – 3 July, 2015, The Hague, the Netherlands. pp. [1-4]
In: (2015). E-proceedings of the 36th IAHR World Congress 28 June – 3 July, 2015, The Hague, the Netherlands. IAHR: [s.l.].
The high number of parameters is a major problem for complex environmental models as it restricts their application. Therefore, sensitivity analysis (SA) methods, that aim to identify the influential and the non-influential parameters of a model, can be essential for an efficient calibration of these models. The SA indeed allows for a reduction of the number of parameters involved in a calibration procedure, by applying factor fixing (FF) and factor prioritization (FP). In this paper, a density-based Global Sensitivity Analysis (GSA) method -PAWN- is applied to the Soil and Water Assessment Tool (SWAT), a highly-parameterized hydrological simulator. The objective of this study is to compare the newly developed PAWN method with the Sobol’ method, which is a well-established and widely used variance-based SA method. The PAWN method considers the entire model output distribution to describe the output uncertainty while the Sobol’ method implicitly assumes that the variance is a sufficient indicator for this purpose. To this end, 26 water quantity related parameters of a SWAT model of the River Zenne (Belgium) are selected to be ranked, using both the PAWN and the Sobol’ methods. In addition, the two SA methods are evaluated and compared in terms of convergence, the related evolution of the parameter ranking results and required computation cost.