• It is well established in rock physics that when the fluid filling the pore spaces of a reservoir is primarily gas or light oil there is a significant change in the bulk modulus but little to no changes in the shear modulus when compared to fully brine-saturated reservoirs. This in turn leads to a lower velocity ratio in gas or light oil reservoirs. This concept underlies and has been extensively used in the fluid identification of conventional siliciclastic reservoirs in seismic interpretation. Yet not so much in the use of higher-resolution sonic logs. Here we ask the question, can we use sonic logs to not just identify but quantify hydrocarbon saturation in gas reservoirs? Using sonic logs, can we accurately estimate water saturation with uncertainties that are similar to those obtainable from conventional saturation models? If possible, this would have significant implications in the evaluation of shaly sand reservoirs as well as low resistivity low contrast reservoirs, where conventional saturation models may not be as effective.

• Given P-wave and S-wave velocities computed from measured compressional and shear sonic logs, a new empirically derived saturation model for estimating water saturation is presented. One that does not require formation resistivity or brine salinity. The new model is based on the lower bulk modulus or velocity ratio observed in gas reservoirs. Assuming the reservoir is fully brine saturated, the deviation of the measured P-wave and S-wave velocities from the fully brine saturated velocities is used to invert for water saturation. The new model is shown to be consistent with theoretical rock physics models.

• We apply the new model to compressional and shear sonic well log data acquired in 10 different wells and 8 different formations. The reservoirs range from an organic gas shale reservoir to a gas sand reservoir offshore the Gulf of Mexico. In one of the organic shale examples, the Haynesville shale, average water saturation from the new model was 33% compared to 37% from the conventional saturation model that has been calibrated to tight rock core analysis. In the Gulf of Mexico gas sand example, the model yields 24% average water saturation compared to 21% average water saturation from core calibrated Archie saturation model. In a low resistivity low contrast shaly sand reservoir where conventional saturation models indicated the reservoir is wet, the new model yields results that are consistent with gas production from the reservoir. We also present the results of the application of the new model to a gas condensate reservoir in the North Slope of Alaska and an oil reservoir offshore the Gulf of Mexico.