Proceedings PaperUse of statistical modeling for setup of full-light-attenuation spectra from measurements in three wavelengths
|Format||Member Price||Non-Member Price|
The light attenuation coefficient c()) is a key parameter which gives information about basic optical properties of marine environment. However, quick and frequent measurement of full light attenuation spectra is not always possible. Usually good correlation between the light attenuation coefficients of different parts of spectra makes possible to derive a full attenuation spectra based on its two or three values, typically in blue, green and red range. The current presentation will describe one of such a method based on analysis of covariance matrix of the data set. The matrix consists of [c(,2), c(,)] elements, and is build upon a sufficient amount of field measurements in certain area. As it is known, set of characteristic values of that matrix makes possible to estimate in optimal way any realization of analyzed random function. Any value of c() may be expressed as a sum of its mean value and linear combination of certain coefficients and characteristic vectors of covariance matrix. The method was tested on light attenuation data sets taken on the Baltic Sea. The estimated error of derived spectra was less than 15 %. The conditions and range of application is discussed.