Estimating the Uncertain Structure of a Water Balance Model via Bayesian Data Assimilation

Nataliya Bulygina and Hoshin Gupta

Department of Hydrology and Water Resources, the University of Arizona, Tucson, USA

HWR Department Seminar, Oct 3, 2007

 

ABSTRACT

To date, most ‘physically based’ models of hydrologic systems are based on an implicit up-scaling premise that the behavior at the model scale can be described by the small scale governing equations, by spatial averaging of the state variables and by use of ‘effective’ parameters.  Of course, the up-scaling assumption may be wrong, and the effective large scale governing equations for a heterogeneous system may be different in form, not just different in parameters, from the equations derived from small-scale physics.  In this work, we assume that the major hydrologic processes in the system and their interconnections have been identified a priori. We wish to construct the mathematical relationships in question via data assimilation, using measurements made on the system inputs and outputs. In this work, as opposed to assuming fixed mathematical structure, a model is perceived as a representation of both what we know and what we do not know about the structure and behavior of a system. The model is therefore seen as the ‘posterior’ joint probability density function for the relationships in question, constructed in such a way that data assimilation helps to correct ‘prior’ belief about the dependences.  In regions where no data are available ‘prior’ knowledge about the system can be incorporated and will dominate. The approach is applied to the problem of estimating a water mass balance model for the Leaf River catchment, Mississippi at various time scales: annual, monthly and weekly. The study considers incoming water as precipitation, and released water as evaporation, transpiration and discharge. On the annual and monthly time scales the estimated models performed well, while on weekly scale there were indications of the limitations in the assumed conceptual model structure.