The researchers used a technique called ensemble Kalman filter to analyze data from a Doppler radar and improve storm forecasts. They found that the assimilation system accurately captured storm characteristics, reduced forecast errors, and improved forecast accuracy, especially for variables with larger errors. After 8 assimilation cycles, the radar observations and forecast fields were reliably correlated. The background error covariance was complex and varied with the flow of the storm. The ensemble mean forecast accurately represented the storm's details in the short term but errors increased quickly over time.