1 Conditioning a surface - based model to wells 1 and thickness data

8 Surface-based models imitate a sequence of depositional events in time. By 9 considering sedimentation processes, these algorithms produce highly realistic 10 subsurface structures from a variety of environments. However, since depositional 11 events are forward-modeled, they cannot be directly conditioned to data. Instead, 12 conditioning requires inferring the right combinations of parameters that match data. 13 Solving such inverse problem is time-consuming and expensive because those models 14 are complex and highly parameterized. In this context, we propose an optimization 15 methodology that enables to fit data efficiently. The key idea behind this method is to 16 decrease the dimensionality of the problem. Three complementary approaches are 17 developed for this purpose. The first approach aims at identifying the leading 18 uncertainty. The second one is a sensitivity analysis on the input parameters. The 19 third one is a re-formulation of the full inverse problem. This methodology is general 20 in the sense that it can be used with any kind of data and environment of deposition. 21 In our work, the efficiency of the methodology is illustrated by using a forward model 22 reproducing lobe structures and a set of borehole data and thickness information 23 extracted from a realistic turbidite reservoir. 24 25