LUFTBLICK OG (AT)
The OneSun project (a physics-constrained,self calibrating,data-driving system with a common solar spectrum for homogeneous trace gas retrieval network) introduces an innovative approach to trace gas retrievals, addressing critical challenges in the operation of fiducial reference networks for satellite validation: ensuring homogeneity in calibration, maintaining high data quality, and enabling scalable network operations.
Central to this development are two groundbreaking paradigms. First, the separation of instrumental and atmospheric features within raw spectral data is achieved through a deep learning-based, unsupervised training process. This approach eliminates the need for conventional laboratory calibrations, automating instrument calibration and enabling direct retrieval of trace gas total columns. Second, by utilizing a common solar reference spectrum and standardized gas cross sections, OneSun ensures consistent data quality across all instruments in the network.
At the heart of the proposed system lies a physics-driven artificial neural network (ANN) architecture. It includes an instrument model that processes raw spectra into calibrated data and an atmospheric model, based on the Beer-Lambert law, to retrieve trace gas quantities. This modular design allows simultaneous training of multiple instruments on a shared atmospheric model, ensuring intrinsic homogeneity across the network.
OneSun would represent a transformative step forward, automating processes that traditionally require intensive manual calibration and centralized processing. It leverages deep learning to process complex patterns in data, overcoming limitations of current methods and enabling a scalable, self-calibrating network of instruments.
The project offers immense potential for global trace gas monitoring, significantly reducing resource requirements for calibration and quality assurance. It is directly applicable to the Pandonia Global Network (PGN), enhancing operational efficiency and benefiting all stakeholders. By integrating state-of-the-art AI into environmental monitoring, OneSun paves the way for more efficient, scalable, and consistent atmospheric remote sensing.
