Estimating time-lapse property changes of a subsurface volume
A backpropagation enabled model is trained for estimating time-lapse property changes of a subsurface volume. Synthetic models of the subsurface volume are generated, with pre-determined property changes before and after a time lapse. These models are used to compute baseline-monitor pairs of synthetic seismic traces before and after the time lapse, wherein the baseline synthetic traces are computed from the synthetic model before the time lapse and the monitor synthetic traces are computed from the synthetic model after the time lapse. A ground truth 4D attribute characterizing the time-lapse property changes in the synthetic models is defined, and a backpropagation enabled model is trained by feeding the baseline-monitor pairs of synthetic seismic traces and the corresponding ground truth 4D attribute. The thus obtained trained backpropagation enabled model can be used to estimate time-lapse property changes of the actual subsurface Earth volume from actual baseline-monitor pairs of seismic traces.
1. A computer-implemented method of estimating time-lapse property changes of a subsurface volume, comprising:
providing actual baseline-monitor pairs of baseline seismic traces of a physical subsurface Earth volume and monitor seismic traces of the physical subsurface Earth volume as obtained from physical seismic measurements in the physical subsurface Earth volume acquired at respectively a first time and second time which is later than the first time by a time lapse;
feeding the actual baseline-monitor pairs of seismic traces to a trained backpropagation enabled model, and obtaining as output estimates of the time-lapse property changes of the actual subsurface Earth volume,
wherein the trained backpropagation enabled model has been obtained by:
generating synthetic models of the subsurface volume with pre-determined property changes before and after a time lapse, said synthetic models comprising seismic velocities;
computing baseline-monitor pairs of synthetic seismic traces using the models generated before and after the time lapse wherein the baseline synthetic traces are computed from the synthetic model before the time lapse and the monitor synthetic traces are computed from the synthetic model after the time lapse;
deriving a ground truth 4D attribute characterizing the time-lapse property changes in the synthetic models;
training a backpropagation enabled model by feeding the baseline-monitor pairs of synthetic seismic traces and the corresponding ground truth 4D attribute, whereby obtaining the trained backpropagation enabled model.
2. The computer-implemented method of claim 1 , wherein random noise has been added to the synthetic seismic traces.
3. The computer-implemented method of claim 1 , wherein the backpropagation enabled model comprises a deep neural network.
4. The computer-implemented method of claim 1 , wherein the backpropagation enabled model comprises a U-Net structure.
5. The computer-implemented method of claim 1 , wherein employing a backpropagation enabled process for training the backpropagation enabled model, wherein the baseline-monitor pairs of synthetic seismic traces are provided in two distinguished input channels in said backpropagation enabled process.
6. The computer-implemented method of claim 1 , wherein said estimates of the time-lapse property changes comprises at least one selected from the group consisting of time shifts and time strains.
7. The computer-implemented method of claim 1 , wherein said estimates of the time-lapse property changes comprises rock property changes.
8. The computer-implemented method of claim 7 , wherein said rock property changes comprise at least one of the group consisting of: oil/gas/water saturation changes, reservoir compaction, overburden/underburden strain, velocity change, and impedance change.
9. A computer-implemented method of training a backpropagation enabled model for estimating time-lapse property changes of a subsurface volume, comprising:
generating synthetic models of the subsurface volume with pre-determined property changes before and after a time lapse, said synthetic models comprising seismic velocities;
computing baseline-monitor pairs of synthetic seismic traces using the models generated before and after the time lapse wherein the baseline synthetic traces are computed from the synthetic model before the time lapse and the monitor synthetic traces are computed from the synthetic model after the time lapse;
deriving a ground truth 4D attribute characterizing the time-lapse property changes in the synthetic models;
training a backpropagation enabled model by feeding the baseline-monitor pairs of synthetic seismic traces and the corresponding ground truth 4D attribute, whereby obtaining the trained backpropagation enabled model.