Script for real-time street flood prediction model using machine learning, Norfolk, VA

This is a python script used to train and test a Random Forest model built for real-time street flood prediction in Norfolk, VA, USA.. The Random Forest surrogate model approximates water depth on streets generated by a 1-D pipe/2-D overland flow hydrodynamic model TUFLOW. The inputs of the model are topographic features: topographic wetness index, depth to water and elevation, and environmental features such as hourly rainfall, cumulative rainfall in previous hours, hourly tide level, etc. The output of the model is hourly water depth on streets during storm events generated by the TUFLOW model.