Determining the onset and production of drizzle in shallow warm clouds is important for evaluating microphysical processes in numerical models, quantifying aerosol-cloud-precipitation interactions and for remote sensing applications. A large dataset of stratocumulus clouds from a continental and a maritime location are used to provide a robust description of the radar Doppler spectra moments as clouds transition from a drizzle-free state to drizzling. A one-dimensional steady-state model that accounts for condensation, evaporation, autoconversion and accretion is first evaluated using the long-term observations and subsequently is used to explain the radar observations. The study has two main outcomes: First, an objective algorithm for the detection of drizzle particles in low stratiform clouds is presented. The algorithm classifies radar observation above the cloud base in three areas: i) those where drizzle particles dominate the radar reflectivity, ii) those where drizzle particles are present but do not dominate the radar reflectivity and iii) those where either the radar observations are drizzle free or drizzle particles are below the detection limit of the proposed technique. Second, the radar Doppler spectra skewness is quantitatively related to drizzle properties such as drop size, water content and to microphysical processes (e.g. autoconversion to accretion rate). The results are compare to satellite-based observations of drizzling low warm clouds and the differences in sensitivity and definitions (e.g., probability of precipitation) are discussed. Finally, the use of radar reflectivity vs skewness relationship in low stratiform clouds for calibrating millimeter-wavelength radars is discussed.