The detection of narrow zones of low-level convergence (or boundaries) critically important for thunderstorm initiation, development and evolution are primary objectives of automated nowcasting systems. The Space-Time Meso Analysis System (STMAS) has been developed at the NOAA Earth System Research Laboratory to take full advantage of the nearly 32,000 high temporal and spatial resolution mesonetwork data being made available for operational nowcasting through the Meteorological Assimilation Data Ingest System (MADIS). Despite the great number of stations, many gaps in coverage remain over the U.S., a feature that we refer to as data “deserts” interspersed with “oases”, which typically occur in and surrounding large urban areas. STMAS variational analyses are produced every 15 min on a 5-km grid over CONUS using a 90-min temporal window to incorporate appropriately weighted off-maptime observations for better time-space coherence, plus a sequential multigrid technique to provide maximum detail in the oases, while not introducing undesirable noise in the data deserts. The cases to be presented at the conference demonstrate how gravity waves, thunderstorm gust fronts, mesoscale pressure systems, lake breezes, and other important mesoscale features can be obtained with unprecedented time-space coherence, while still being able to detect significant changes that can occur in less than 30 minutes.