Development of robust fully autonomous machine learning-based cloud processing for Infrared Cloud Imagers

Cloud cover is an important but poorly characterized component of current climate models. In particular, there is a need to detect and quantify cloud cover in a physically meaningful manner. This need has led to the development of infrared cloud imagers. This system uses long-wave infrared cameras to observe the thermal emission from the clouds and produces a measurement of spatially distributed cloud cover, cloud optical depth, and cloud radiative forcing independent of solar illumination. However, previous systems have been maintenance intensive, dependent on co-located instruments, and required integrated calibration sources. The dependence on co-located instruments comes from the need to model clear sky emission to separate cloud emission from sky emission. These problems have limited their utility as field-deployable instruments. The proposed project will make the infrared cloud imaging systems self-contained by adding onboard metrology instruments and incorporating artificial intelligence to build a hybrid artificial intelligence and radiometric processing system. These advancements will result in a robust and fully autonomous cloud imaging system that will provide the required radiometric cloud data. Furthermore, this approach will enable these instruments to deploy to any site increasing their utility and economic value. In Phase I, an atmospheric model based on global radiosonde launches will be built along with methods to modify this model in real-time by independently detecting clear sky. Work will also develop a suite of sensors required for model inputs, including precipitable water vapor derived from a global positioning system receiver. Artificial intelligence methods will remove the human from the loop by detecting and correcting for ground objects on the horizon and spots on the lens. The commercial potential for cloud imaging exists within climate science but also extends beyond. Any system significantly impacted by clouds are potential markets which include situational awareness at airports, solar power generation, and Earth to space optical communication.