Urban flooding is getting worse with climate change, especially in cities that don’t have enough flood sensors or ground data. This research introduces a new way to predict the severity of urban floods using satellite images and rainfall data, without needing traditional field observations. The method uses satellites to detect flooding, combines that with data on rain, terrain, and land cover, and then applies machine learning to estimate how severely each part of the city might flood. This helps us understand and map urban floods at a fine scale, even in places where no flood data exists. The approach is designed to be scalable, low-cost, and useful for decision-makers planning flood resilience and emergency response in vulnerable cities around the world.