Disclosed are methods and systems for predicting time varying loudness in a geographic region. Training data, including noise information, weather information, and traffic information is collected from a plurality of sensors located in a plurality of geographic regions. The information is collected during multiple time periods. The noise information includes time varying loudness. Static features of the geographic regions are also defined and included in the training data. The static and time varying dynamic features train a model. The model is used predict time varying loudness within a different region and at a time later than times the training data is collected. The predicted loudness levels are utilized, in some aspects, to determine a route for an aircraft.