Reducing Flood Risks in Belgrade Area through AI Solutions

Among the many capitals around the world, Belgrade had great luck being placed on the delta of two wonderful rivers,the Sava and Danube, with 200 km length of the waterfrontss. Both river’s streams stretch to many countries in Europe, connecting Belgrade with them. In Belgrade’s area, there are 14 river’sIcelandd, some of which are protected by the law as stationeries of protected animal’s species. Some parts of the waterfronts are popular spots for the fisherman ,while some of them are arranged as beaches, during the summer they have over 100000 visitors. All this brings a lot of benefits to the city, but also there is a risk from floods. In recent decade, Belgrade was affected with manyfloods, where one from 2014 was devastating. Obrenovac, one of Belgradee municipalities, was totally flooded, over 30000 people were relocated, around 10000 houses were devastated.

Even though the government made some effort to improve the protection from floods, there are still areas, which are not protected. Such areas depend on the right time action of the local community, to set up mobile fortifications in vulnerable places.

With our project, we can provide an AI predictive system, which can give a possibility for floods, based on the current levels of the water, precipitation, and other parameters. For the training, we’ll use historical hydrometeorology data, focusing on time fragments of recent floods in the last 100 years. 

With such a solution, the preparedness of the local community for potential floods will be on a high level, which means less devastation.

The main goal should be the creation of a supervised ML model, based on historical weather and hydrological data for Danube, Sava and other small rivers in Belgrade area, which will bring a prediction of possible floods in the future. It can be later integrated into the web and mobile app.

• Datasets with historical hydrological data.
• Machine learning process.
• API for a predictive service for floods.

- Data preparation. - Data Mining. - Machine Learning Model's techniques.