In this work, we describe an efficient approach for wind retrieval from dual Doppler radar data. The approach produces a gridded field that not only satisfies the observations, but also satisfies the anelastic mass continuity equation. <br><br> The method is based on the so-called three-dimensional variational approach to the retrieval of wind fields from radar data. The novelty consists in separating the task into steps that reduce the amount of data processed by the global minimization algorithm, while keeping the most relevant information from the radar observations. The method is flexible enough to incorporate observations from several radars, accommodate complex sampling geometries, and readily include dropsonde or sounding observations in the analysis. <br><br> We demonstrate the usefulness of our method by analyzing a real case with data collected during the TPARC/TCS-08 field campaign.