SYNTHETIC EXPERIMENTS OF WATER LEVEL OBSERVATION ASSIMILATION INTO THE RRI MODEL WITH THE ENSEMBLE OPTIMAL INTERPOLATION SCHEME

Abstract

 This paper presents a study on the performance of the ensemble optimal interpolation (EnOI) scheme with stationary covariance matrices for assimilation of synthetic water level observations into the rainfall-runoff-inundation (RRI) model. Five state and observation error covariance matrices are picked from the ensemble Kalman filter (EnKF) simulation of a flood event and then used during the update stage for Kalman gain calculation for (i) state and (ii) parameter estimation with the EnOI scheme. The stationary EnOI improves state estimation at randomly selected validation locations (along with most of the river grids) for all five covariance matrices, but improvement throughout the basin is not guaranteed as performance is degraded at some unobserved locations. The method also has the potential to be used in real applications as the results improve not only for the same flood, but also for a different event from which the matrices are not extracted. Model parameters can also be successfully estimated with the EnOI scheme, but the estimation may be compromised if the parameters interfere with each other. Independently updating, or only updating the most sensitive parameter (Manning’s roughness coefficient for river in this case), leads to better estimation.