Radar is a fundamental tool for severe storm monitoring and nowcasting activities. Forecasters examine real-time NEXRAD observations from multiple radars, other remote sensing tools, severe weather detection algorithms, and use their considerable experience and situational awareness to issue severe storm warnings that help protect the public from hazardous weather events. However, escalating data flow rates from new sensors and applications will make it challenging for forecasters to make the best use of all the available data in warning operations in a timely manner. To overcome this difficulty, a real-time, weather-adaptive three-dimensional variational data assimilation (3DVAR) system has been developed recently for NOAA supported Warn-on-Forecast project (WoF) to incorporate all available traditional and radar observations within an analysis domain that could be hit by severe weather, including tornadoes, hails and strong damage winds. The unique features include: (1) The system has the ability to automatically detect and analyze severe local hazardous weather events at 1km horizontal resolution every 5 minutes. (2) The analysis can also be performed with on-demand capability in which end-users (or forecasters) set up the location of the analysis domain in real time based on the current weather situation. (3) The analysis product can help forecasters identify strong circulations imbedded in thunderstorms so that the accuracy of warnings for hazardous weather threats may be improved.. Although still in the early development stage, the system performed very well during the spring of 2010. Many severe weather events were all successfully detected and analyzed. Currently, we are working to make the analysis product available in "near realtime" (4-5 minutes delay) to the NWS forecasters as one of the official projects of the NOAA's HWT Experimental Warning Program. The objectivity of the procedure ensures that (i) all available information, including all nearby WSR-88Ds and NAM high resolution analysis and forecast products, are used, (ii) physically-consistent gridded data are provided to forecasters to help make their warning decisions in a timely manner, and (iii) the problem of subjectivity, inherent to some arbitrary criteria in some severe weather detection algorithms, is avoided. The performance of the system during the 2011 Spring season experiment will be reported during the conference.