Toward Discharge Estimation for Water Resources Management with a Semidistributed Model and Local Ensemble Kalman Filter Data Assimilation

Abstract

Estimating discharges is a major challenge in water resources management, and techniques such as data assimilation (DA) can be used to improve these estimates. This study assessed application of the local ensemble Kalman filter (LEnKF) DA scheme within a large-scale hydrological-hydrodynamic model to improve discharge estimates. Different scenarios with assimilation and validation gauges were performed to obtain an optimal setup of localization, ensemble size, assimilation parameters, and observation type (discharge or logarithm of discharge). These parameters were used to analyze the estimated discharge series and reference discharge values for maximum, minimum, and average flows. Results showed that joint perturbation of precipitation and groundwater reservoir volume leads to the most sensitive model responses. Furthermore, using the adequate setup of LEnKF parameters and assimilating the logarithm of discharge, better estimates of reference discharge values were obtained with the DA scheme when compared to a simple regionalization method, leading to reduction of errors of more than double in many cases.