Under the support of the WoF and SENSR projects, we set up several ensemble data assimilation and forecast experiments using 36 members at 1 km grid spacing using data from the 31 May 2013 Oklahoma tornado and flash flood event. The Weather Research and Forecasting (WRF) model and Center for Analysis and Prediction of Storms (CAPS) 4D Ensemble Kalman Filter are used as the prediction model and data assimilation algorithm. An experiment assimilating PAR observations from the National Weather Radar Testbed research PAR is compared to an experiment assimilating WSR-88D observations. To address the rapid reduction of ensemble spread during the data assimilation period due to a large volume of data, covariance inflation is performed between subsequent radar volume scans. Several 1-hour forecasts are run to evaluate the effect of the different observations on the placement of heavy rainfall and supercell mesocyclones. Preliminary results show that the experiment assimilating PAR observations better capture the evolution of the mesocyclone than the experiment assimilating WSR-88D observations. Quantitative assessment of the impact of PAR data on the lead time for flash flood and/or tornado potential will be also presented.