Development of a neural network model for cloud fraction detection using NASA-Aura OMI VIS radiance measurements

The discrimination of cloudy pixels is required in almost any estimate of a parameter retrieved from a satellite image in the ultraviolet (UV), visual (VIS) or infra-red (IR) parts of the electromagnetic spectrum. Also, the distincion of clouds within satellite imagery and the distribution of their micro-physical properties is essential to the understanding of radiative transfer through the atmosphere. This paper reports the development of neural network algorithms for cloud detection for the NASA-Aura Ozone Monitoring Instrument (OMI). We present and discuss the results obtained by training mathematical neural networks with simultaneous application to OMI and Aqua-MODerate Resolution Imaging Spectrometer (MODIS) data. The neural network delivers cloud fraction estimates in a fast and automated way. The developed neural network approach performs generally well in the training. Highly reflective surfaces, such as ice, snow, sun glint and desert, or atmospheric dust mislead the neural network to a wrong predicted cloud fraction.