Mesoscale models have become more prevalent in operational weather forecasting during recent times. We examine a long time series of data for evaluation of system performance for an operational NWP system over the northeastern United States. The system, dubbed 'Deep Thunder' is a mesoscale weather modeling and visualization system which integrates high-performance computing, visualization, and automation into a system that includes receiving and processing data, modelling, and post-processing analysis and dissemination. The forecast domain which is centered on the NYC metropolitan area extends from Virginia to Maine to the north and south, and western Pennsylvania to the coastal Atlantic waters to the west and east is a triple nested structure at 16, 4, and 1 km. Quantitative results are examined for several months of data using RMS and other statistical measures of system output versus observations for standard meteorological variables. For comparison the ETA model output is also compared to observed conditions. Qualitatively, model results are compared to observations and radar data for specific events.