Towards automated early detection of risks for a CO2 plume containment from permanent seismic monitoring data

This storage contains  the training data for neural networks proposed in a manuscript 'Towards automated early detection of risks for a CO2 plume containment from permanent seismic monitoring data'. The data consists of output from reservoir simulations of a small-scale CO2 injection at CO2CRC Otway Project Stage 2C (Victoria, Australia). The output is presented as a set of images, where each pixel in a portable network graphics is a plume thickness for a particular injection scenario at a particular day after the injection has commenced. The format is unsigned integer 16-bit. The data set contains images of two major types:

1. REALISTIC: plumes are obtained from reservoir simulations in a complex geological model that was calibrated on an extensive set of geophysical  surveys. File naming follows this convention 'plume_thick_real_scenario_%S_day_%N.png', where %S represents a string that encodes the injection scenario name and %N denotes day number after the injection started.

2. VANILLA:  plumes are obtained from reservoir simulations in a simple model of a reservoir that reflects only few typical features of the Otway injection interval. 'plume_thick_vanilla_scenario_%S_day_%N.png', where %S represents a string that encodes the injection scenario name and %N denotes day number after the injection started.