Remote sensing of algae in inland southern African waters

Quibell, Gavin Edward (1991) Remote sensing of algae in inland southern African waters. Masters thesis, Rhodes University.

[img] Text
QUIBELL-MSc-TR92-56.pdf

19Mb

Abstract

Routine monitoring of algae in inland waters in southern Africa is a critical element in assessing the efficacy of eutrophication management options. Several authors have indicated that single point samples are not necessarily representative of conditions throughout the water body and some have suggested remote sensing as a means of overcoming this problem. Remote sensing of algae normally involves deriving the empirical relationship between radiance detected at a sensor, and contact sensed chlorophyll concentrations. Quantification of, or compensation for, contributions to the upwelling radiance other than that light reflected by the algae is critical for this approach. In southern Africa these contributions arise primarily from atmospheric effects and from scattering by sediments in the water. A review of the atmospheric correction models suggested that a cosine sun angle correction followed by dark pixel subtraction is the most feasible method to compensate for the former effects. Studies of the changes in upwelling radiance induced by addition of sediment to algal cultures indicated that subtraction of reflectance at ~665nm from that at ~700nm, may provide a means of compensating for the scattering by sediments. The disadvantage of this approach is that few sensor systems have narrow spectral bands centred at these wavelengths. Investigations of the nature of the reflectance from 5 algal species indicated that all had similar reflectance spectra, but the blue-green genera had a smaller peak at ~650nm. Chlorophyll absorption at ~665nm was evident by lower reflectance at this point, but the alga Microcystis sp. did not conform to the conceptual model of reflectance, in that reflectance at 665nm was higher at increased cell density . . Spectra of natural waters confirmed the results obtained in the laboratory. Reflectance at ~700nm showed the largest changes with increasing chlorophyll concentration and also had the highest correlations to chlorophyll concentrations. However, due to the strong absorption of these wavelengths by water, this reflectance peak only occurred when sufficient cells were found in the upper layers of water. Use of these wavelengths in remote sensing models should therefore be restricted to highly eutrophied waters. Although the reflectance spectra of different algae were similar, the amount of light scattered by each species (measured as turbidity) differed for any given chlorophyll concentration. This appeared to be due to the colonial nature of the cells and means that empirical models will be unique to the species on which they were developed. Comparisons of multispectral photography (MSP) and LANDSAT MSS imagery indicated the MSP data had higher correlations with chlorophyll concentrations than did the MSS data. Chlorophyll simulations from a test set of data using ordinary multiple regression showed that the MSP imagery had mean errors of 7.3M9/I, while that for the MSS imagery was 7.4M9/I. Similar tests using the canonical procedure produced larger mean errors of 9M9/I and 12M9/I for the MSP and MSS data respectively. This was due to the fact that the canonical procedure is not suitable for the spectral band widths of these sensors. In spite of similar simulation accuracies, the MSS imagery produced very patchy synoptic views. This was due to the lower variance (radiometric resolution) in the LANDSAT MSS data. This appears to be the most important criterion for accurate chlorophyll mapping in inland waters. Development of a single multidate algorithm for southern Africa is not yet feasible, and routine monitoring of chlorophyll using these techniques is impractical. However acceptable chlorophyll maps are possible if the model is recalibrated for each occasion and the sensor used has a high radiometric resolution.

Item Type:Thesis (Masters)
Uncontrolled Keywords:Algae, Remote sensing, Freshwater algae, South Africa
Subjects:Q Science > QL Zoology
Divisions:Faculty > Faculty of Science > Zoology & Entomology
Supervisors:O'Keeffe, J.
ID Code:4240
Deposited By: Mrs Judith Cornwell
Deposited On:11 Dec 2012 12:30
Last Modified:11 Dec 2012 12:30
0 full-text download(s) since 11 Dec 2012 12:30
0 full-text download(s) in the past 12 months
More statistics...

Repository Staff Only: item control page