Abstract
Ensemble Kalman filter (EnKF) is a recursive data process algorithm that uses continuous updating. It has been proven that EnKF is an efficient method for data assimilation, uncertainty assessment, and large scale problems in many engineering fields. However, there are two common limitations-filter divergence and overshooting/undershooting. These are due to reduction of cross-covariance between model parameters and measurements. We propose a streamline-assisted ensemble Kalman filter (SL EnKF) that uses covariance localization according to the types of well and measurement data. This method enables selective updates of permeability, therefore, providing more reliable permeability field estimations than the standard EnKF without overshooting/undershooting or filter divergence. In addition, it gives efficient uncertainty evaluations by considering the performances of each ensemble member.