Prediction and validation of gas hydrate saturation distribution in the eastern Nankai Trough, Japan: Geostatistical approach integrating well-log and 3D seismic data

Accurate reservoir potential evaluation requires reliable 3D reservoir models. Geostatistical simulation techniques can reproduce the heterogeneity and quantify the uncertainty in a reservoir. We have applied sequential Gaussian simulation with collocated cokriging to generate the spatial distribution of gas hydrate (GH) saturation around a gas production test site in the eastern Nankai Trough. The simulation was performed using well-log data obtained from the exploration and production tests as a primary variable and inversion-derived seismic impedance data as a secondary variable under the good correlations between two variables. The integrated model adequately described the reservoir heterogeneity and effectively interpolated the seismic trend with respect to the well data. To confirm the usability of the seismic data for the accurate representation of the GH saturation distribution, we ran two model simulations: one using well data only and the other using well and seismic data. Each model was validated using the well-log data obtained at the production test site that were not included during the simulation. The model generated using well and seismic data appropriately reproduced the trend of well-log data at the production test site, especially for the low-GH-saturation unit within the reservoir. However, the model generated using well data only was insufficient to predict the trend of the well data. The results demonstrated that the seismic data were effective for the prediction of the GH saturation distribution, and integration of the well and seismic data could improve the accuracy of the reservoir model.