Localization in outdoor environments requires solutions for a robot to work safely in all conditions. GNSS is not sufficient as it can fail in particular conditions. The radar is a promising way to complement GNSS as it could be low cost and efficient under harsh conditions. Frequency modulated continuous wave panoramic radar with millimeter wavelength are promising solutions as they give long distance information at 360°. In order to use the radar in localization, a mapping process is needed. Due to the particular signal of the radar, the mapping is very dependent on the geometrical structure of the environment. Filtering steps are needed to extract a powerful and meaningful map. The choice of intensity mapping allows a denser map with minimal loss of information. In this article, radar data pre-processing, and the definition of new mapping processes are presented. The way the robot can be localized in the map is also presented. Experimental analyses are performed, and results show that the newly defined map carries information that was filtered out with previously existing methods. It is then shown that the obtained map improves robustness in the localization while facilitating interpretation. Fifty-centimeter accuracy with a third of a degree of precision can be reached while being thirty meters from the mapping path in an area densely populated with vegetation.