Diego G. Loyola R.
German Aerospace Center (DLR)
Mastering the Big Data from Atmospheric Composition Satellite Sensors
Air pollution has become by far the leading environmental health risk factor and satellite sensors are crucial for monitoring the air quality on a global scale. This talk presents machine learning techniques for mastering the expected Big Data from the new generation of European atmospheric composition Copernicus satellite missions (Sentinel-5 Precursor, Sentinel-4, and Sentinel-5) with increased spatial, temporal, and spectral resolution. We developed a systematic and comprehensive method for optimally handling regression tasks with very large high dimensional data. The proposed approach is based on smart sampling techniques for minimizing the number of samples to be generated by using an iterative algorithm that creates new sample sets until the input and output space of the function to be learned are optimally covered. This method was extended for the retrieval of atmospheric composition parameters from satellite sensors using a novel technique named full-physics inverse learning machines. These new algorithms are extremely fast and accurate and will be used for the operational near-real-time processing of the new generation of European atmospheric composition sensors, the first one (TROPOMI/Sentinel-5-Precursor) planned to be launched in summer 2017.