A deep-learning-based surrogate model for data assimilation in dynamic subsurface flow problems

•Deep-learning-based surrogate model for dynamic subsurface flow is developed.•Method uses a residual U-net and convolutional LSTM recurrent network.•Surrogate capable of predicting states and well rates in channelized geomodels.•Data assimilation accomplished by combining surrogate with CNN-PCA par...

Ausführliche Beschreibung