A new numerical nowcasting technique to predict the radar reflectivity field at very short term, up to few hours, is presented. The method is based on the spatial segmentation of the reflectivity field and estimated advection field to produce radar reflectivity forecasts and, for this reason, is named Spatially-adaptive Precipitation Advective Radar Estimator (SPARE). A large data set coming from the Italian radar network mosaic (spatial domain size of about 1200 1200 km2) is used to test the overall performance of SPARE against the simplest method of radar map temporal persistence. An original approach to estimate the radar field motion, based on the phase cross-correlation principle, is formulated in this paper. Results are given either in terms of skill scores of predicted radar maps or in terms of predicted uncertainty. The latter provides a new methodology to evaluate the expected performance of SPARE predictions.