The purpose of this study was to evaluate the effect of instrument noise, atmospheric conditions, and reflectance retrieval method on multispectral reflectance and fractional cover estimates derived from spectral mixture analysis. Top-of-atmosphere (TOA) radiance was simulated for pixels with various ground (fractional cover of green vegetation (GV), nonphotosynthetic vegetation (NPV), and soil) and atmospheric/geometric (solar zenith angle, visibility, and water vapor) conditions for four operational spaceborne multispectral images, Landsat 5 Thematic Mapper (TM), Landsat 7 Enhance Thematic Mapper Plus (ETM. +), Landsat 8 Operational Land Imager (OLI), and the Moderate Resolution Imaging Spectroradiometer (MODIS). Instrument noise was added to the TOA radiance. Three reflectance retrieval methods were used to estimate surface reflectance, and fractional cover estimates of GV, NPV, and soil were made using spectral mixture analysis (SMA) and multiple endmember spectral mixture analysis (MESMA). Among atmospheric characteristics, we found that visibility and solar zenith angle made the greatest contributions to spectral and fractional cover error. In all cases, GV fraction was estimated with lower error than NPV or soil fraction. The reflectance retrieval method that provided the best reflectance estimates did not produce the fractional cover estimates with the lowest error due to variable bias across bands. Fractional cover estimates for all ground components were strongly impacted by instrument noise, but MESMA significantly improved fractional cover estimates for NPV and soil, regardless of noise level. This study suggests that there are practical constraints on the level of accuracy that can be attained in multispectral estimates of NPV and soil. A set of recommendations for the use of spectral unmixing of multispectral reflectance data is developed.
Okin, G. S., & Gu, J. (2015). The impact of atmospheric conditions and instrument noise on atmospheric correction and spectral mixture analysis of multispectral imagery. Remote Sensing of Environment, 164, 130–141. https://doi.org/10.1016/j.rse.2015.03.032