20 Remote Sensing of African Lakes: A Review
413
chlorophyll estimates in the same order of magnitude as those measured. In the case
of the few available TSM values, the MERIS-estimated data were found to be in the
same range as those measured in the field.
Koponen et al. (2008) evaluated the performance of the chlorophyll-a MERIS
products in Lake Manzalah, Egypt, using the EUL processor both with and without
the use of the Improve Contrast between Ocean & Land (ICOL) technique.
While the correlation between estimated chlorophyll and measured chlorophyll
was linear, large errors were evident in slope and bias. Based on the observation
of the imagery products, it was clear that local water quality variation present in
Lake Manzalah exceeded the processor capability. Nearly all pixels were flagged
“out of scope” for estimated chlorophyll concentration product, largely due to failed
atmospheric correction techniques.
Kneubühler et al. (2007) used MERIS imagery to map chlorophyll-a in Lake Kivu
during the dry season over the course of three years. The images were atmospherically corrected using both the Atmospheric & Topographic Correction (ATCOR)
model, version 2 (Atmospheric Correction for Flat Terrain, ATCOR2), and Simplified Method for Atmospheric Correction (SMAC) processor. The study tested five
band radio algorithms using MERIS bands at 443, 490, 560, 620, 665, 709 and
754 nm, as well as, an algorithm using 443, 560 and 665 nm modified for Landsat
bands. Linear regression models were applied to seven MERIS images to investigate
spatial and inter-annual variability in the lake, particularly the increased chlorophyll
concentrations in bays and coastal areas. The authors note that CDOM variability
might have influenced these estimates.
Chavula et al. (2009b) used MODIS imagery products to estimate chlorophyll
in Lake Malawi. Lake water was collected at three lake sampling locations during
numerous sampling cruises and processed to determine chlorophyll-a concentration
(n = 136). Level 1 MODIS Terra imagery for the sampling dates was obtained from
the NASA Level 1 and Atmospheres Archive and Distribution System (LAADS)
system, while Level 2 and Level 3 MODIS Aqua imagery was obtained from the
NASA Ocean Color Group. Terra algorithms were found to be of limited use, a
situation attributed to the adverse effects of atmospheric effects on reflectance. The
most effective Aqua chlorophyll algorithm was R rs 443/R rs 551.
Majozi (2011) used a combination of field-collected spectral attenuation measurements with MERIS products to examine eutrophic depth distribution in Lake
Naivasha. Field measurements of downwelling irradiance at 0.1 and 0.6 m were used
to calculate attenuation (K d ) for three wavelength ranges (430–600, 600–800 and
800–930 nm). Lake-specific attenuation models were tested against MERIS products
[K d (490), K d (620)] and found to be fairly accurate [rmse = 0.044 m for K d (620)].
The attenuation model for euphotic depth was applied to MERIS images of the lake,
revealing the spatial variety and range of light penetration in Lake Naivasha.
Oberholster and Botha (2010) used Landsat 5 imagery in combination with molecular markers to successfully detect cyanobacteria in Lake Hartbeespoort. A red to
NIR algorithm was used to determine when cyanobacteria populations were present.
Lake sampling to determine the presence and density of cyanobacteria relied on
a combination of chlorophyll concentration, PCR analysis and microcystin toxin
detection using an Enzyme-Linked Immunosorbent Assay (ELISA).
413
chlorophyll estimates in the same order of magnitude as those measured. In the case
of the few available TSM values, the MERIS-estimated data were found to be in the
same range as those measured in the field.
Koponen et al. (2008) evaluated the performance of the chlorophyll-a MERIS
products in Lake Manzalah, Egypt, using the EUL processor both with and without
the use of the Improve Contrast between Ocean & Land (ICOL) technique.
While the correlation between estimated chlorophyll and measured chlorophyll
was linear, large errors were evident in slope and bias. Based on the observation
of the imagery products, it was clear that local water quality variation present in
Lake Manzalah exceeded the processor capability. Nearly all pixels were flagged
“out of scope” for estimated chlorophyll concentration product, largely due to failed
atmospheric correction techniques.
Kneubühler et al. (2007) used MERIS imagery to map chlorophyll-a in Lake Kivu
during the dry season over the course of three years. The images were atmospherically corrected using both the Atmospheric & Topographic Correction (ATCOR)
model, version 2 (Atmospheric Correction for Flat Terrain, ATCOR2), and Simplified Method for Atmospheric Correction (SMAC) processor. The study tested five
band radio algorithms using MERIS bands at 443, 490, 560, 620, 665, 709 and
754 nm, as well as, an algorithm using 443, 560 and 665 nm modified for Landsat
bands. Linear regression models were applied to seven MERIS images to investigate
spatial and inter-annual variability in the lake, particularly the increased chlorophyll
concentrations in bays and coastal areas. The authors note that CDOM variability
might have influenced these estimates.
Chavula et al. (2009b) used MODIS imagery products to estimate chlorophyll
in Lake Malawi. Lake water was collected at three lake sampling locations during
numerous sampling cruises and processed to determine chlorophyll-a concentration
(n = 136). Level 1 MODIS Terra imagery for the sampling dates was obtained from
the NASA Level 1 and Atmospheres Archive and Distribution System (LAADS)
system, while Level 2 and Level 3 MODIS Aqua imagery was obtained from the
NASA Ocean Color Group. Terra algorithms were found to be of limited use, a
situation attributed to the adverse effects of atmospheric effects on reflectance. The
most effective Aqua chlorophyll algorithm was R rs 443/R rs 551.
Majozi (2011) used a combination of field-collected spectral attenuation measurements with MERIS products to examine eutrophic depth distribution in Lake
Naivasha. Field measurements of downwelling irradiance at 0.1 and 0.6 m were used
to calculate attenuation (K d ) for three wavelength ranges (430–600, 600–800 and
800–930 nm). Lake-specific attenuation models were tested against MERIS products
[K d (490), K d (620)] and found to be fairly accurate [rmse = 0.044 m for K d (620)].
The attenuation model for euphotic depth was applied to MERIS images of the lake,
revealing the spatial variety and range of light penetration in Lake Naivasha.
Oberholster and Botha (2010) used Landsat 5 imagery in combination with molecular markers to successfully detect cyanobacteria in Lake Hartbeespoort. A red to
NIR algorithm was used to determine when cyanobacteria populations were present.
Lake sampling to determine the presence and density of cyanobacteria relied on
a combination of chlorophyll concentration, PCR analysis and microcystin toxin
detection using an Enzyme-Linked Immunosorbent Assay (ELISA).