2.2.6 Colored Dissolved Organic Matter
Interest among aquatic scientists has increased greatly over the past decade in ORS
applications to measure CDOM in surface waters. In part, this reflects a growing
interest in quantifying the role of lakes and other surface water bodies in the global
carbon cycle, along with the understanding that CDOM represents a large fraction
of the total DOC in many aquatic systems.
A wide range of approaches, including analytical, semi-analytical, matrix inversion, and empirical techniques, have been used to retrieve CDOM values for fresh
and marine waters by satellite imagery. The most successful algorithms for marine
conditions (including coastal waters) involve semi-analytical matrix inversion
methods (e.g., [59–61]); but such approaches have been used only a few times for
freshwaters (e.g., [11, 62]). The most common retrieval methods for lakes are
empirical reflectance-ratio equations that involve nonlinear (power) equations.
For example, the equation of Kutser et al. [4] uses the ratio of Advanced Land
Imager (ALI) band 2 (525–605 nm) to band 3 (630–690 nm): a 420 ¼ 5.13(ALI2/
ALI3)
2.67 . ALI bands 2 and 3 have approximately the same wavelength ranges as
Landsat TM and ETM+ bands 2 and 3 and Landsat 8 OLI bands 3 and 4. Menken
et al. [49] independently found a similar relationship using ground-based
hyperspectral reflectance data, a 440 ¼ 146.4(R 670 /R 550 )
2.08 , and Ficek et al. [63]
also used a similar equation.
Using in situ reflectance hyperspectra and associated water quality measurements on ~30 Minnesota and Wisconsin lakes with wide ranges of CDOM, chlorophyll, and TSS, Brezonik et al. [29] recently found that the best band ratio models
used similar wavelengths for Landsat 8 bands. With the larger selection of Sentinel2 and Sentinel-3 bands, a different ratio using ~500 nm:~750 nm worked best.
Fig. 4 Reflectance spectra for a eutrophic stretch of the Mississippi River downstream of St. Paul,
Minnesota. Data redrawn from Brezonik et al. [29]
124
L.G. Olmanson et al.
Interest among aquatic scientists has increased greatly over the past decade in ORS
applications to measure CDOM in surface waters. In part, this reflects a growing
interest in quantifying the role of lakes and other surface water bodies in the global
carbon cycle, along with the understanding that CDOM represents a large fraction
of the total DOC in many aquatic systems.
A wide range of approaches, including analytical, semi-analytical, matrix inversion, and empirical techniques, have been used to retrieve CDOM values for fresh
and marine waters by satellite imagery. The most successful algorithms for marine
conditions (including coastal waters) involve semi-analytical matrix inversion
methods (e.g., [59–61]); but such approaches have been used only a few times for
freshwaters (e.g., [11, 62]). The most common retrieval methods for lakes are
empirical reflectance-ratio equations that involve nonlinear (power) equations.
For example, the equation of Kutser et al. [4] uses the ratio of Advanced Land
Imager (ALI) band 2 (525–605 nm) to band 3 (630–690 nm): a 420 ¼ 5.13(ALI2/
ALI3)
2.67 . ALI bands 2 and 3 have approximately the same wavelength ranges as
Landsat TM and ETM+ bands 2 and 3 and Landsat 8 OLI bands 3 and 4. Menken
et al. [49] independently found a similar relationship using ground-based
hyperspectral reflectance data, a 440 ¼ 146.4(R 670 /R 550 )
2.08 , and Ficek et al. [63]
also used a similar equation.
Using in situ reflectance hyperspectra and associated water quality measurements on ~30 Minnesota and Wisconsin lakes with wide ranges of CDOM, chlorophyll, and TSS, Brezonik et al. [29] recently found that the best band ratio models
used similar wavelengths for Landsat 8 bands. With the larger selection of Sentinel2 and Sentinel-3 bands, a different ratio using ~500 nm:~750 nm worked best.
Fig. 4 Reflectance spectra for a eutrophic stretch of the Mississippi River downstream of St. Paul,
Minnesota. Data redrawn from Brezonik et al. [29]
124
L.G. Olmanson et al.