Premium
Evaluation and optimization of bio‐optical inversion algorithms for remote sensing of Lake Superior's optical properties
Author(s) -
Mouw Colleen B.,
Chen Haidi,
McKinley Galen A.,
Effler Steven,
O'Donnell David,
Perkins Mary Gail,
Strait Chris
Publication year - 2013
Publication title -
journal of geophysical research: oceans
Language(s) - English
Resource type - Journals
eISSN - 2169-9291
pISSN - 2169-9275
DOI - 10.1002/jgrc.20139
Subject(s) - colored dissolved organic matter , remote sensing , inversion (geology) , environmental science , ocean color , satellite , absorption (acoustics) , satellite imagery , phytoplankton , algorithm , computer science , geology , optics , physics , chemistry , paleontology , structural basin , organic chemistry , astronomy , nutrient
Satellite remote sensing offers one of the best spatial and temporal observational approaches. However, well‐validated satellite imagery has remained elusive for Lake Superior. Lake Superior's optical properties are highly influenced by colored dissolved organic matter (CDOM), which has hindered the retrieval of chlorophyll concentration through band‐ratio algorithms. This study evaluated seven existing inversion algorithms. The top‐performing inversion algorithm was tuned to a Lake Superior optical data set and applied to satellite imagery. The retrieval of chlorophyll concentration via inversion algorithms was not possible due to errors in derived CDOM absorption being greater than phytoplankton absorption values and the very small contribution of phytoplankton absorption to the overall absorption budget. However, the retrieval of absorption due to CDOM from satellite imagery was encouraging. To ensure that the best satellite remotely sensed reflectance estimates were used in the retrieval of absorption due to CDOM, several atmospheric correction schemes were evaluated. The absorption due to CDOM was greatest in the western arm of Lake Superior and near river mouths and decreased with distance offshore. The absorption due to CDOM had a bimodal distribution over the annual cycle with the greatest peak in fall and a smaller peak in spring.