The ExoMars 2022 rover mission has been designed to search for traces of life, past or present, on Mars and more specifically below its hostile surface. This will be the first mission able to collect samples down to 2 meters in the Martian subsurface, where organic molecules and bio-signatures may be preserved. The images of the subsurface (or radargrams) provided by the ground penetrating radar WISDOM onboard the ExoMars rover will be key to select the safest and scientifically most relevant locations for drilling. However, the vertical resolution of WISDOM radargrams is limited (to ∼5.5 cm in a typical soil of dielectric constant 4) by the bandwidth of the instrument when applying classical processing techniques. Here we propose to enhance WISDOM radargrams vertical resolution by implementing the Bandwidth Extrapolation (BWE) technique. We demonstrate that this technique yields an improvement of WISDOM vertical resolution by a factor of 3 while well preserving the time of arrival and amplitude of the detected echoes. We validate the BWE technique on synthetic data before applying it to experimental observations acquired both in controlled and natural environments. The resulting super-resolved radargrams greatly improve our understanding of the investigated terrains with a better separation of echoes from underground reflectors, and demonstrate the value of implementing the BWE technique in the WISDOM data processing pipeline which will ultimately provide images of the Martian subsurface down to a few meters depth with a vertical resolution of ∼2 cm in typical soils.