Automatic Calibration of a Simulator Applied to a Mountain River Employing Experimental Data of Rainfall and Level – Case Study: D’Antas Stream, RJ

ABSTRACT

This work couples an optimization technique to a precipitation-flow model, for the purpose, among others, of developing an automatic calibration tool. The strategy was tested in the estimation of some parameters required for modeling the water level in drainage systems, which were evaluated in a case study, involving the simulation of flood waves in the main course that drains the watershed of D’Antas Stream located in Nova Friburgo-RJ. Here, the direct problem consisted in calculating the water level variation at a given position on that course, implemented on the MOHID-Land platform. On the other hand, the inverse problem focused on the estimation of three hydraulic parameters, which are characteristic of both the basin and the course itself. By applying the method of Luus-Jaakola (LJ) values were estimated for these parameters, resulting in the best fit between simulation and observation. The assessment of the adopted methodology was based on results achieved for three different rainfall scenarios, including the one that led to the reported tragedy in the mountain region of Rio de Janeiro in January 2011. Generally speaking, the results were very satisfactory, both regarding the computational efficiency in the estimates and the ability of the model to fit the observed data.

Keywords:
Nova Friburgo; Watershed; Parameter estimation; MOHID platform; Floods