Future weathercasters might add a traffic temperature to their forecasts. Georgia Tech civil engineer John Leonard is developing a traffic model of Atlanta that incorporates historical data and other variables, ranging from day of the week to special events and weather conditions, to estimate traffic conditions for the coming day. To make it simpler to interpret the model’s output, Leonard distills the results into a single congestion index, or “temperature.” Leonard’s model incorporates data from a network of video cameras and other detectors that monitor the city’s freeways; a byproduct is a real-time contour map of travel times from selected points. Over time, Leonard says, the model will adjust itself to reflect the way that people alter their driving patterns in response to the computer predictions. Full implementation of the model, which could be used elsewhere, remains several years away.