Conventional time-lapse imaging methods rely on the repeatability of the acquisition geometries to extract subsurface variations. The usual approach is to record a baseline survey at the beginning of the monitoring stage and repeated surveys every couple of years or so.  A problem with the conventional approach is that although it is assumed that the properties in each location may change with time, property distributions at each time are not statistically independent.

It would then not be appropriate to consider each reconstruction separately, since if the successive reconstructions are carried out independently, we can lose much of the time-sequential information.

We are developing techniques to use time-sequential information to better constrain time-lapse inversion. One approach is the joint inversion of multiple time-lapse data with temporal regularization. In this case data recorded at different times are inverted simultaneously. A more practical approach which inverts the time-lapse data sequentially as it is recorded is being developed.