loading page

The Spectral Mixture Residual: A Source of Low-Variance Information to Enhance the Explainability and Accuracy of Surface Biology and Geology Retrievals
  • +4
  • Daniel Sousa,
  • Philip Gregory Brodrick,
  • Kerry Cawse-Nicholson,
  • Joshua B Fisher,
  • Ryan Pavlick,
  • Christopher Small,
  • David Ray Thompson
Daniel Sousa
San Diego State University

Corresponding Author:dsousa1@gmail.com

Author Profile
Philip Gregory Brodrick
Jet Propulsion Laboratory, California Institute of Technology
Author Profile
Kerry Cawse-Nicholson
Jet Propulsion Laboratory, California Institute of Technology
Author Profile
Joshua B Fisher
Chapman University
Author Profile
Ryan Pavlick
California Institute of Technology
Author Profile
Christopher Small
Columbia University
Author Profile
David Ray Thompson
Jet Propulsion Laboratory, California Institute of Technology
Author Profile


Mixed pixels are the rule, not the exception, in decameter terrestrial imaging. By definition, the reflectance spectrum of a mixed pixel is a function of more than one generative process. Physically-based surface biology or geology retrievals must therefore isolate the component of interest from a myriad of unrelated processes, heterogenously distributed across hundreds of square meters. Foliar traits, for example, must be isolated from canopy structure and substrate composition which can dominate overall variance of spatially integrated reflectance. We propose a new approach to isolate low-variance spectral signatures. The reflectance of each pixel is modeled assuming linear geographic mixing due to a small library of generic endmembers. The difference between the modeled and observed spectra is deemed the Mixture Residual (MR). The MR, a residual reflectance spectrum that is presumed to carry the subtler and variable signals of interest, is then leveraged as a source of signal. We illustrate the approach using three datasets: synthetic composites computed from field reflectance spectra, NEON AOP airborne image compilations, and DESIS satellite data. The MR discriminates between land cover versus plant trait signals and accentuates subtle absorption features. Mean band-to-band correlations within the visible, NIR, and SWIR wavebands decrease from 0.97, 0.94, and 0.97 to 0.95, 0.04 and 0.31. The number of dimensions required to explain 99% of image variance increases from 4 to 13. We focus on vegetation as an illustrative example, but note that the concept can be extended to other applications and used as an input to other algorithms.
Feb 2022Published in Journal of Geophysical Research: Biogeosciences volume 127 issue 2. 10.1029/2021JG006672