Use of Citizen Science Data for Validating Geostationary Satellite Cloud Detection Capabilities During the Solar Terminator

Detecting clouds using satellite imager data can be difficult under some conditions including areas within the solar terminator during sunrise and sunset. This study explores the use of Citizen Science observations to validate geostationary satellite cloud detection capabilities devised for the NASA SatCORPS (Satellite Cloud Observations and Radiative Property Retrieval System) operated at NASA Langley Research Center (LaRC). To validate the satellites cloud detection capabilities during the solar terminator, NASAs GLOBE (Global Learning and Observations to Benefit the Environment) Clouds team also at LaRC, launched a program where citizen scientists could record observations near sunrise and sunset. Over 600 terminator observations were recorded from February 2021-July 2021 through the GLOBE Programs GLOBE Observer app. This data was used alongside NWS (National Weather Service) ceilometer data from over 2,000 ceilometer stations in the continental United States. In addition, any GLOBE Cloud observation recorded through the programs app within the terminator hours were also used. This expanded our dataset of citizen observations during the solar terminator by over 5,000. Comparisons of citizen scientist and ceilometer data with satellite analyses revealed several cases where the traditional satellite cloud detection algorithm was over-reporting cloud coverage, particularly near sunset. A new machine learning cloud detection algorithm recently devised by the SatCORPS team was then applied to these cases and found to significantly improve cloud detection accuracies in the solar terminator.