Image-Based Ground Visibility for Aviation: Is What You See What You Get? - (Pilot Study)

Including a crowd in the weather-sensing loop has the potential for improving the availability of weather observations to Alaska’s widely dispersed airfields where essential weather data sets are currently not available. One method of virtually expanding the existing weather-sensing infrastructure, at least in part, would be to pair the images taken by the large network of aviation weather cameras installed in Alaska with crowdsourced estimates of ground visibility (FAA 14 CFR 1.1 defines ground visibility as the prevailing horizontal visibility near the earth’s surface as reported by the United States National Weather Service or an accredited observer [1]) derived from those images. In 2018, the Civil Aerospace Medical Institute (CAMI) conducted a pilot study of image-based ground visibility utilizing CAMI’s cloud-based research platform at https://cbtopsatcami.faa.gov. The goal of this exploratory research was twofold. First, make observations about the behavior of the different image-based and non-image-based visibility models across different weather conditions during daytime at airfields where a traditional weather sensor is collocated with an aviation weather-camera installation and on-staff expert human weather-observers. Second, survey the viability of deriving ground visibility from Alaska’s weather camera network via crowdsourcing in applied settings. The models’ behavior was examined using daily time-series plots. The recommendations for future research are founded on the observations of the models’ behavior, participation rates and feedback from the pilot and expert human observer’s communities.