The University Center for Atmospheric Research (UCAR) has a good press release on improvements using short term nowcasting with Dual Pol 88D weather radar data to improve stream flooding models during the 2014 STEP (Short Term Explicit Prediction) experiment. It was done around the greater Denver, CO area during the summer of 2014 to improve the Weather Research and Forecasting (WRF) model and National Weather Service hydrological models. The WRF goes out to 12hrs and it is run with a Auto-nowcaster program to extrapolate in time future storm precipitation amounts.
Auto-nowcaster can predict storm growth from 10min to 1hr out. The stream models are done in a 100m grid. While it is still large it is improvement based on weather data acquisition and current atmospheric model capabilities/physics. As we learn more about weather physics and increase our weather sensor network, we can reduce the scale further below mesoscale for better modeling where GIS/Geomatics datasets can be better utilized for better public safety and protection of commerce.