Professional Documents
Culture Documents
Parameterization of The Inherent Optical
Parameterization of The Inherent Optical
Parameterization of The Inherent Optical
3Norwegian Institute for Water Research (NIVA), P.O. Box 173, Oslo, Norway
5Department of Physics and Engineering Physics, Stevens Institute of Technology, Hoboken, New Jersey 07030-5991, USA
Lake Victoria, Africa’s largest freshwater lake, suffers greatly from negative changes in biomass of
species of fish and also from severe eutrophication. The continuing deterioration of Lake Victoria’s
ecological functions has great long-term consequences for the ecosystem benefits it provides to the
countries bordering its shores. However, knowledge about temporal and spatial variations of optical
properties and how they relate to lake constituents is important for a number of reasons such as remote
sensing, modeling of underwater light fields, and long-term monitoring of lake waters. Based on statis-
tical analysis of data from optical measurements taken during half a year of weekly cruises in Murchison
Bay, Lake Victoria, we present a three-component model for the absorption and a two-component model
for the scattering of light in the UV and the visible regions of the solar spectrum along with tests of their
ranges of validity. The three-component input to the model for absorption is the chlorophyll-a (Chl-a),
total suspended materials concentrations, and yellow substance absorption, while the two-component
input to the model for scattering is the Chl-a concentration and total suspended materials. © 2007
Optical Society of America
OCIS codes: 010.4450, 290.5820, 300.1030.
4. Results and Discussion where x1 and x2 are two predictors [30] given by x1
In our parameterization we used measured values of ! pMC, x2 ! #MTSM " pMC$, and y ! a " ay " aw.
the absorption and scattering coefficients in the pho- Using a statistical multiple linear regression, we
tosynthetically active radiation (PAR) region and obtained the least-square estimates of '1 and '2
modeled values in the UV region. Because the mea- by minimizing the sum of the squared deviations
sured scattering coefficient (with contributions from 'i!1
n
#yi " '1xi1 " '2xi2$2. Here n is the number of data
suspended materials, which include phytoplankton, points, which was 14 in our case.
mineral particles, and undissolved organic matter) The square of the multiple correlation of y on the
was found to vary smoothly with wavelength [27], we predictors #x1, x2$ is given by
fitted b as a function of wavelength in the PAR spec-
tral region to the measured values according to the '1Sx1y % '2Sx2y
formula R2 ! , (10)
Sy2
n
where $o ! 550 nm, and where the average value of Sx1y ! ' #xi1 " x!1$#yi " ȳ$, (11)
n was 0.74. We used this formula to determine ex- i!1