NCAR's Runway Friction and Closure Prediction System (RFCPS)

It’s well known that adverse winter weather can significantly disrupt airport operations. Snow and ice buildup on the runways reduces the pavement friction and can cause airplanes to slide upon landing or takeoff. When the surface friction falls below certain levels, runways must be closed in order to keep planes safe. The safety and efficiency of airport and flight operations hinges on timely and accurate weather forecasts that can also give an indication as to when the runway friction is reduced to the point where runways must be closed. Therefore, having an accurate forecast of runway friction can help airport managers make better decisions on when and how long to close runways.

Until recently, the airport community has relied on conventional methods for acquiring and applying weather-related runway friction information in the runway closure decision process usually from multiple sources. Minneapolis–Saint Paul International Airport (MSP) experiences several winter storms each season where the runways must be closed due to a loss of runway friction. As a result, MSP contacted the National Center for Atmospheric Research (NCAR) for help in automating the procedure for recording and relating friction observations to runway closure times.

This paper will give an overview and discuss the basic components of the Runway Friction and Closure Prediction System (RFCPS) developed by NCAR for MSP airport. It will cover all the major components of the system and how they fit together to produce a friction forecast and a runway closure forecast. Items that will be covered include the back-end weather and road forecast engine, the different machine-learning models developed to predict runway friction, including an expert model for friction prediction, applying the friction models in real-time and the output products being created and delivered to MSP. Lastly, recommendations will be made for improving the system moving forward into the future.