Use of Prior Information for Calibration of River Roughness

Abstract

River roughness calibration usually is an underdetermined model for lack of hydrological information, and its calibrated results may deviate from the true solution seriously. Using prior information is an effective way to regularize inverse problems, which can make the calibrated results reasonable and improve the stability of numerical calculation. A robust model, which takes account of river roughness smooth matrix and error minimization, is developed on the basis of the prior information that the spatial distribution of river roughness is smooth. In the computing example of a river network, cases are tested about the uniqueness of solution, numerical stability, and reliability. The results show that the present model can calibrate river roughness effectively with high numerical stability and reliable results.