Over the years, the city of Rio de Janeiro has been subject of several flooding events due to heavy rains. In particular, in early April 2010, the city endured one of the worst torrential rainstorms in decades with significant lost of lives and property damage. To assist in planning for such events in the future, a Center of Operations has designed and opened in December 2010. At present time, this center is responsible for the coordinated disaster and emergency management for the city. As part of this effort, IBM Research has an on-going project to develop mathematical models for flooding prediction with a lead time of 48 hours to help in the planning for emergency procedures. We have executed a feasibility study to understand the flooding conditions, and the quality and volume of data available. As part of this study, we had access to 1m LiDAR Digital Terrain Models (DTMs), and GIS maps of soil type, land occupation, and city structure (streets, lakes, rivers, etc). However, only limited digital drainage data was available and there were no detailed streamflow measurement or urban system drainage model accessible, which limits the creation of an accurate flooding model for the city. On the other hand, very good historical flooding data was available, and the city has catalogued at least 232 recurrent points of flooding, which are now being addressed by structural changes in the city drainage system. The information available is normally the location and the approximate peak time of flooding. In few sites, the peak flooding polygon has been mapped for particularly severe events. Using the limited historical data we developed a simplified high resolution analytical model for flooding prediction taking into account the geomorphological and the historical data. A mathematical model using shallow water equations coupled with a weather forecast model is used to analyze if a site which is historically prone to flood could receive a surface runoff flow which could cause a flooding event on these points. The system, known as PMAR, integrates the WRF-ARW community NWP model which was adapted for use in the Rio de Janeiro area. The estimated precipitation is used as input to our analytical model. At this point in time, we are calibrating the model for verification and improvements on the results.