Role of critical gas saturation in the interpretation of a field scale CO2 injection experiment