Storms and their associated phenomena have a high impact in a country’s social and economic environment. It is necessary to have tools that guarantee the safety of people at all times. The data provided by a network of meteorological radars is of utmost importance for storm tracking and identification. These radar networks are expensive and difficult to maintain. Not all countries have a radar network that covers all its extension and even so, they often have technical or connectivity problems that leave large areas without information about rainfall. In this work we present a technique to generate a synthetic radar reflectivity field using supervised deep learning techniques with cloud-to-ground lightning data. This product can be used as a cost-effective radar alternative in areas where there is no or poor coverage, or as a complement to meteorological radar images that are affected by attenuation, interference or other observation problems.