Abstract
Abstract. Visible–shortwave infrared imaging spectroscopy provides valuable remote measurements of Earth's surface and atmospheric properties. These measurements generally rely on inversions of computationally intensive radiative transfer models (RTMs). RTMs' computational expense makes them difficult to use with high-volume imaging spectrometers, and forces approximations such as lookup table interpolation and surface–atmosphere decoupling. These compromises limit the accuracy and flexibility of the remote retrieval; dramatic speed improvements in radiative transfer models could significantly improve the utility and interpretability of remote spectroscopy for Earth science. This study demonstrates that nonparametric function approximation with neural networks can replicate radiative transfer calculations and generate accurate radiance spectra at multiple wavelengths over a diverse range of surface and atmosphere state parameters. We also demonstrate such models can act as surrogate forward models for atmospheric correction procedures. Incorporating physical knowledge into the network structure provides improved interpretability and model efficiency. We evaluate the approach in atmospheric correction of data from the PRISM airborne imaging spectrometer, and demonstrate accurate emulation of radiative transfer calculations, which run several orders of magnitude faster than first-principles models. These results are particularly amenable to iterative spectrum fitting approaches, providing analytical benefits including statistically rigorous treatment of uncertainty and the potential to recover information on spectrally broad signals.
Highlights
Remote visible–shortwave infrared (VSWIR) imaging spectroscopy, known as hyperspectral imaging, is a powerful approach to map the composition, health, and biodiversity of Earth’s ecosystems (ESAS, 2018)
Inferring geophysical properties requires inverting the measurement with a physical model – typically one that accounts for both absorption and scattering by the atmosphere, and the fraction of light reflected from the surface at each wavelength (Schaepman-Strub et al, 2006)
This study demonstrates an accurate neural network model deployed as part of an iterative model inversion, showing that emulation is a practical solution for operational atmospheric correction of imaging spectroscopy data
Summary
Remote visible–shortwave infrared (VSWIR) imaging spectroscopy, known as hyperspectral imaging, is a powerful approach to map the composition, health, and biodiversity of Earth’s ecosystems (ESAS, 2018). To date, techniques designed to retrieve surface reflectance using learned RTM emulators have only been demonstrated on a small number of cases with limited surfaces and atmospheres (Verrelst et al, 2017; Martino et al, 2017; Brajard et al, 2006), and not across the VSWIR range with state vector flexibilities that would permit a functionally useful alternative for existing atmospheric correction routines (e.g., as a surrogate forward model). This study demonstrates an accurate neural network model deployed as part of an iterative model inversion, showing that emulation is a practical solution for operational atmospheric correction of imaging spectroscopy data This opens new possible avenues of research, for both the inversion algorithm itself (to explore further expansions of the state vector beyond the traditional retrieved variables) and downstream analyses (to exploit the benefits of new retrieval methods that do not require lookup tables). We describe paths for future development of neural network RTM emulation technology
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
