Abstract The architectural complexity of a paleovalley ~350 m deep has been revealed by acquisition and conventional processing of a high-resolution seismic-reflection survey in northern Alberta, Canada. However, processing degraded much of the high quality of the original raw data, particularly with respect to near-surface features such as commercial methane deposits, and that motivated use of additional processing algorithms to improve the quality of the final images. The additional processing includes development of a velocity model, via tomographic inversion, as the input for prestack depth migration (PSDM); application of a variety of noise-suppression techniques; and time-variant band-pass filtering. The resulting PSDM image is of poorer quality than the newly processed time-reflection profile, thus emphasizing the importance of a good velocity function for migration. However, the tomographic velocity model highlights the ability to distinguish the materials that constitute the paleovalley from the other surrounding rock bodies. Likewise, the reprocessed seismic-reflection data offer enhanced spatial and vertical resolution of the reflection data, and they image shallow features that are newly apparent and that suggest the presence of gas. This gas is not apparent in the conventionally processed section. Consequently, this underscores the importance of (1) ensuring that primarily high-frequency signals are kept during the processing of near-surface reflection data and (2) experimenting with different noise-suppression and elimination procedures throughout the processing flow.