Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Doctorat → Afrique du Sud → 2009 → Remote Sensing of Water Quality Parameters in Zeekoevlei, a Hypertrophic, Cyanobacteria-Dominated Lake, Cape Town, South Africa

University of Cape Town (2009)

Remote Sensing of Water Quality Parameters in Zeekoevlei, a Hypertrophic, Cyanobacteria-Dominated Lake, Cape Town, South Africa

Matthews, Mark William

Titre : Remote Sensing of Water Quality Parameters in Zeekoevlei, a Hypertrophic, Cyanobacteria-Dominated Lake, Cape Town, South Africa

Auteur : Matthews, Mark William

Université de soutenance : University of Cape Town

Grade : Doctor of Philosophy 2009

Résumé
Globally widespread eutrophication and harmful cyanobacterial algal blooms have degraded the condition of many of the earth‘s freshwater resources. In water scarce South Africa, many recreational and water supplying reservoirs are increasingly at risk from these threats. Zeekoevlei Lake, situated south of the City of Cape Town, is an archetype of a severely degraded, permanently hypertrophic and cyanobacteria dominated lake. This study aims to use satellite remote sensing to monitor and provide information on water quality parameters and cyanobacterial algal blooms in Zeekeovlei. Development of remote sensing tools will contribute towards understanding system processes and improving management of cyanobacterial algal blooms which may be toxic. The primary challenges for remote sensing in Zeekoevlei are the extremely high concentrations of covariant water constituents and its very small size (2.6Ha). This presents difficulties with algorithm development, especially differentiating between water constituents, and atmospheric correction. The study combines multispectral MERIS and Landsat 7 ETM+ data with in situ measurements of Chl a ; total suspended solids ; organic/inorganic suspended solids ; absorption by CDOM ; and Secchi Disk depth. Upwelling radiance and downwelling irradiance were measured in situ with a hyperspectral radiometer. Empirical algorithms for estimating the variation in water quality parameters generated from simultaneous field and remote measurements had significant correlation coefficients. The MERIS Level 2 Product and Eutrophic Lakes Processor Neural Network algorithms produced erroneous and invalid results. Maps produced from the satellite images reveal the synoptic variability and temporal trends of the parameters. The maps gave more accurate estimates of the mean spatial water quality parameters than the small number of in situ sample sites. The results indicate that MERIS, being designed for water applications, is far better suited than Landsat, and similar sensors, for frequent monitoring and change detection in Zeekoevlei. This is due to MERIS‘s higher acquisition frequency, free data availability, spectral arrangement, and higher signal-tonoise ratio. The study‘s success is not without consideration of the limitations of the small data set. Recommendations concerning future requirements for algorithm development, and the integration of real-time/operational remote water quality monitoring systems in southern Africa, are made.

Annonce

Version intégrale (3 Mb)

Page publiée le 28 novembre 2013, mise à jour le 26 juin 2017