Comparison of Two Models for Prediction of Seismic Streamer State Using the Ensemble Kalman Filter

Abstract

Towed marine seismic streamers are extensively utilized for petroleum exploration. With the increasing demand for efficiency, leading to longer and more densely spaced streamers, as well as four-dimensional (4D) surveys and more complicated survey configurations, the demand for optimal streamer steering has increased significantly. Accurate streamer state prediction is one important aspect of efficient streamer steering. In the present study, the ensemble Kalman filter (EnKF) has been used with two different models for data assimilation including parameter estimation followed by position prediction. The data used are processed position data for a seismic streamer at the very start of a survey line with particularly large cable movements due to currents. The first model is a partial differential equation (PDE) model reduced to two-dimensional (2D), solved using a finite difference method (FDM). The second model is based on a path-in-the-water (PIW) model and includes a drift angle. Prediction results using various settings are presented for both models. A variant of the PIW method gives the overall best results for the present data.