Time-lapse image registration using the local similarity attribute |
modl,rat
Figure 1. (a) 1-D synthetic velocity model before (solid line) and after (dashed line) reservoir production. (b) True (solid line) and estimated (dashed line) interval velocity ratio. |
---|
data,warp
Figure 2. 1-D synthetic seismic images and the time-lapse difference initially (a) and after image registration (b). |
---|
scan100
Figure 3. (a) Local similarity scan for detecting the warping function in the 1-D synthetic model. Red colors indicate large similarity. The black curve shows an automatically detected trend. |
---|
Figure 1a shows a simplistic five-layer velocity model, where we introduce a velocity increase in one of the layers to simulate a time-lapse effect. After generating synthetic image traces, we can observe, in Figure 2a, that the time-lapse difference contains changes not only at the reservoir itself but also at interfaces below the reservoir. Additionally, the image amplitude and the wavelet shape at the reservoir bottom are incorrect. These artifact differences are caused by time shifts resulting from the velocity change. After detecting the warping function from the local similarity scan, shown in Figure 3, and applying it to the time-lapse image, the difference correctly identifies changes in reflectivity only at the top and the bottom of the producing reservoir [Figure 2b]. To implement the local similarity scan, we use the relative stretch measure . When the two images are perfectly aligned, . Deviations of from one indicate possible misalignment. Finally, we apply equation 9 to estimate interval velocity changes in the reservoir and observe a reasonably good match with the exact synthetic model [Figure 1b].
Time-lapse image registration using the local similarity attribute |