Atmospheric Lengthscales for Global VSWIR Imaging Spectroscopy

David R. Thompson*, Niklas Bohn, Philip G. Brodrick, Nimrod Carmon, Michael L. Eastwood, Regina Eckert, Cédric G. Fichot, Joshua P. Harringmeyer, Hai M. Nguyen, Marc Simard, Andrew K. Thorpe

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review


Future global Visible Shortwave Infrared Imaging Spectrometers, such as the Surface Biology and Geology (SBG) mission, will regularly cover the Earth's entire terrestrial land area. These missions need high fidelity atmospheric correction to produce consistent maps of terrestrial and aquatic ecosystem traits. However, estimation of surface reflectance and atmospheric state is computationally challenging, and the terabyte data volumes of global missions will exceed available processing capacity. This article describes how missions can overcome this bottleneck using the spatial continuity of atmospheric fields. Contemporary imaging spectrometers oversample atmospheric spatial variability, so it is not necessary to invert every pixel. Spatially sparse solutions can train local linear emulators that provide fast, exact inversions in their vicinity. We find that estimating the atmosphere at 200 m scales can outperform traditional atmospheric correction, improving speed by one to two orders of magnitude with no measurable penalty to accuracy. We validate performance with an airborne field campaign, showing reflectance accuracies with RMSE of 1.1% or better compared to ground measurements of diverse targets. These errors are statistically consistent with retrieval uncertainty budgets. Local emulators can close the efficiency gap and make rigorous model inversion algorithms feasible for global missions such as SBG.

Original languageEnglish
Article numbere2021JG006711
JournalJournal of Geophysical Research: Biogeosciences
Issue number6
StatePublished - Jun 2022
Externally publishedYes


  • atmospheric correction
  • hyperspectral imagery
  • imaging spectroscopy


Dive into the research topics of 'Atmospheric Lengthscales for Global VSWIR Imaging Spectroscopy'. Together they form a unique fingerprint.

Cite this