University of Twente Student Theses


Estimating chlorophyll-a absorption as a proxy to phytoplankton biomass from MERIS data lake Naivasha, Kenya

Ghirmai, Mussie (2011) Estimating chlorophyll-a absorption as a proxy to phytoplankton biomass from MERIS data lake Naivasha, Kenya.

[img] PDF
Abstract:This paper has investigated chlorophyll-a light absorption coefficient in lake Naivasha as a proxy to estimate the phytoplankton biomass using MERIS full resolution (FR) data (300 m) and empirical algorithms. An increase in the phytoplankton content of water bodies contributes to a profound degradation in water quality that can have adverse ecological and human health effects through production of toxins in the presence of cyanobacterias or mechanical damage to marine environment. Since such algal bloom events are sporadic in time and isolated in space, traditional field based measurement and monitoring methods for detecting blooms of phytoplankton are costly and time consuming, delaying management decisions. Remote sensing technique which utilizes the optical properties (absorption and reflectance) of blue-green algal pigments (chlorophyll a and phycocyanin) can provide rapid detection of algal blooms. The water leaving reflectance from MERIS FR products were derived using an image based processing (by comparing with the in-situ reflectance) for estimating and correcting atmospheric noise recorded by the sensor. Coupled with physical and bio-chemical data from lakes, remote sensing can provide an efficient method for tracking algal bloom occurrence and their distribution in time and space for long-term management strategies. A total of about 169 in-situ radiometric and about137 absorbance spectra were collected during one cruise from 17th of September 2010 to 03rd October 2010 in the two physically interconnected but optically different lakes; the Crescent Island lagoon and the main Lake Naivasha using a Trios RAMSESARC radiance sensor (7° field of view, 320–950 nm), and a Trios RAMSES-ACC-VIS irradiance sensors. Ground truth samples were analyzed for pigment absorption coefficients immediately after sampling using the RD2000 Spectrophotometer. An atmospheric correction was implemented on the MERIS imagery data and validated using in-situ hyperspectaral water leaving reflectance (WLR) spectra measured on Lake Naivasha. The R2 and RMSE values between the atmospherically corrected satellite derived WLR and in-situ measured values in the NIR was 0.746 and 0.117 respectively. Previously published empirical spectral algorithms for the detection of chl-a by Gitelson (2008), Shen et al. (2010), and FLH baseline methods by Gower et al. (2004), were applied to the in-situ phytoplankton absorption coefficient, Trios-reflectance and MERIS derived reflectance spectra divided into calibration and validation dataset. To account for weak FLH signal at around 685nm due to the higher chlorophyll content of lake Naivasha, a similar technique known as Maximum Chlorophyll index (MCI) Gower et al. (2005) was analyzed using the 709 nm band peak in the Trios and MERIS reflectance spectrum. Spectral measurements with concurrent analytical laboratory data were analyzed using statistical least square regressions for the determination of best fit coefficients. Algorithm accuracy was tested through a least squares regression by r2 and RMSE. Results show that three band algorithm of chl-a absorption coefficients prediction yielded coefficients of determination as high as 0.84, RMSE 0.058, for an aggregated dataset (n=42). The SCI and MCI empirical algorithms for the estimation of chl-a absorption coefficients resulted in a poor correlation with the in-situ absorption coefficients. A weak polynomial quadratic relationship was found between the SCI and MCI algorithms against the in-situ absorption coefficient (n=42) which is inadequate for characterizing the highly dynamic phytoplankton distribution of lake Naivasha. Although no improvement could be achieved for the poor correlation between the MCI index value and in-situ phytoplankton coefficient, the underperformance of the SCI algorithm showed an increase in the correlation coefficient for stations with SCI index and light absorption values of less than 0.0123sr-1 and 0.02m-1 respectively which is a lower value for lake Naivasha. Result obtained suggests that the three band algorithm is more robust and accurate and was used for mapping the phytoplankton absorption in lake Naivasha. Keywords: Phytoplankton Biomass; Chlorophyll-a absorption; remote sensing; MERIS; Lake Naivasha
Item Type:Essay (Master)
Faculty:ITC: Faculty of Geo-information Science and Earth Observation
Programme:Geoinformation Science and Earth Observation MSc (75014)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page