Innovative Approach Based on Data Projection onto Ensemble Subspace for Improving Simultaneous Assimilation of Production and 4D Seismic Data

Summary Time-lapse (4D) seismic data are valuable information that can reveal spatial dynamic behavior such as fluid saturation and pressure changes caused by production and injection in a reservoir. They are complementary information to production and pressure data measured in wells. Well data are more frequent in time but poorly distributed in space, while 4D seismic data are well distributed in space and less frequent in time. Also, 4D seismic data are more susceptible to noise and data error than production data. Due to very different characteristics of both kinds of data, simultaneous assimilation of wells and 4D seismic data is challenging. One of the key issues is the definition of a proper 4D seismic measurement error to balance the weight of both data types in the data assimilation process. The objective of this paper is to propose an innovative approach based on 4D seismic projection onto ensemble subspace to properly calibrate the weight of 4D seismic data. We propose regional calibration of the standard deviation of 4D seismic measurement errors based on an automatic segmentation procedure to highlight the main 4D seismic anomalies. The proposed method was successfully applied in a sector of a complex real field from the Brazilian pre-salt under a water-alternating-gas (WAG) injection scheme. By comparing the proposed approach with existing methods, we demonstrate the superiority of our method, which offers clear contributions for improving simultaneous assimilation of production and 4D seismic data. The proposed method provides fine-tuning calibration of the 4D seismic anomalies and more accurate production forecast than the other tested methods.