Localized heavy storm is one of the major causes of flood damage in urban regions. According to the recent disaster statistics in South Korea, the frequency of urban flood is increasing more frequently, and the scale is also increasing. However, localized heavy storm is difficult to predict, making it difficult for local government officials to deal with floods. This study aims to construct a Flood risk matrix (FRM) using ensemble weather prediction data and to assess its applicability as a means of reducing damage by securing time for such urban flood response. The FRM is a two-dimensional matrix of potential impacts (X-axis) representing flood risk and likelihood (Y-axis) representing the occurrence probability of dangerous weather events. To this end, a regional FRM was constructed using historical flood damage records and probability precipitation data for basic municipality in Busan and Daegu. Applicability of the regional FRMs was assessed by applying the LENS data of the Korea Meteorological Administration on past heavy rain events. As a result, it was analyzed that the flood risk could be predicted up to 3 days ago, and it would be helpful to reduce the damage by securing the flood response time in practice.