Saturday, 6 June 2015

Monitoring water quality in South African reservoirs from space.

Water quality is a major issue in South Africa, where efforts to expand clean water supplies to the whole population in the post-Apartheid era have faced severe challenges due to a limited supply of potable water combined with a large, fast growing and diffusely distributed population. One of the major obstacles to this objective has been widespread eutrophication of reservoirs, largely as a result of agricultural runoff. Eutrophication of water occurs when high levels of nutrients, such as nitrogen and phosphorus, in the water support growth of algae and other organisms which can prove detrimental to the ecosystem and the health of anybody drinking the water.

Ideally reservoir water should be oligotrophic prior to treatment (i.e. have very low levels of nutrients), but only 3% of South African surface waters are thought to fall into this category by the National Eutrophication Monitoring Programme of the Department of Water Affairs, while 37% are deemed to be mesotrophic (having intermediate levels of nutrients), 33% are eutrophic (having high levels of nutrients) and 28% are hypertrophic (having very high levels of nutrients). Waters in the latter two categories are prone to Algal Blooms, and in particular growths of Cyanobacteria (filament-forming photosynthetic Bacteria) which produce toxins harmful to other organisms, and Fish kills and deaths of Wildlife and Domestic Animals due to Cyanobacterial toxins are a common occurrence in South Africa.

Water quality in South African reservoirs is monitored by the National Eutrophication Monitoring Programme, however limited resources combined with a very large number of small reservoirs scattered over a very wide area make direct sampling of water a challenging task, and ways of supplementing this monitoring in order to better prioritise resources are actively being sought.

Hartbeespoort Dam in North West Province is notorious for high levels of eutrophication and frequent Cyanobacterial blooms. Igmar Grewar/African Lens.

In a paper published in the South African Journal of Science on 27 May 2015, Mark Matthews of CyanoLakes (Pty) Ltd. and the Department of Oceanography at the University of Cape Town and Stewart Bernard of the Department of Oceanography at the University of Cape Town and Earth Systems Earth Observation at the Council for Scientific and Industrial Research describe the results of an analysis of ten years of data collected by the European Space Agency's Medium Resolution Imaging Spectrometer (MERIS) satellite, which collected data between 2002 and 2012.

All molecules absorb energy in the form of electromagnetic radiation (light) across a broad spectrum, but can only store so much energy before it is re-emitted. Once this energy limit is exceeded the molecule must re-emit the energy, but does this at specific wavelengths corresponding to the chemical bonds present in the molecule. This means that every molecule has a distinctive electromagnetic spectrum, which can be used to detect specific molecules on Earth from orbiting satellites, or on other planets from telescopes on Earth.

Matthews and Bernard developed an algorithm for determining the difference between oligotrophic, mesotrophic, eutrophic and hypertrophic waters, as well as detecting differences between Cyanobacterial and other Algal blooms and detecting Cyanobacterial surface scums, which are considered to be a particular health hazard. This was applied to fifty of the largest water bodies in South Africa (measured by surface area), excluding waterways that were narrower than 600 m, highly seasonal in their extent or strongly affected by tidal movements.

In order to test the validity of this Matthews and Bernard looked for known instances of Cyanobacterial blooms within the data; that is to say blooms that were recorded from ground analysis that should also have appeared in the satellite data.

Blooms of the Cyanobacterium Ceratium hirundinella, were first detected at Albert Falls Dam in October 2006, with clearer conditions prevailing by January 2007. The satellite data was able to replicate this perfectly, and revealed that an earlier bloom had occurred in the spring and summer of 2004, which was not identified at the time.

A comparison between the spatial distribution and magnitude of chl-a determined from in-situ and satellite measurements for Ceratium hirundinella blooms in Albert Falls Dam. MERIS scenes acquired simultaneous to in-situ measurements are shown for (c) 13 October 2006 and (d) 22 January 2007. Matthews & Bernard (2015).

Cyanobacterial blooms were discovered to be occurring in the winter at Midmar Dam in 2005. This was surprising, as such blooms do not generally occur at temperatures of less than 20˚C, and has been linked to excess nitrogen in the water, generated by overwintering waterfowl, which enabled the Cyanobacteria to thrive at lower temperatures. Such winter blooms have subsequently been detected at a number of other South African reservoirs. Mathews and Bernard were able to detect these winter blooms at Midmar Dam using the MERIS data.

Chl-a maps for Midmar Dam showing two scenes acquired near or simultaneous to measurements made by in 2005 illustrating winter Cyanobacteria detection. The sample points in ground surveys are indicated at 1 and 2; the maximum chl-a value measured was 92 mg/m3. The arrows indicate pixels identified as cyanobacteria. Matthews & Bernard (2015).

Hartbeespoort Dam is noted for its extensive surface blooms of the Cyanobacterium Microcystis aeruginosa, which leads to the formation of toxic surface scums. These surface scums have previously been detected in Landsat data, and Matthews and Bernard were also able to detect them in the MERIS data.

Surface scum maps of Hartbeespoort Dam from simultaneously acquired MERIS and Landsat imagery. Dark-green pixels correspond to surface scum in the MERIS imagery. Matthews & Bernard (2015).

Of the 50 bodies included in the study, 36 were found to be hypertrophic, three eutrophic, four mesotrophic and seven oligotrophic. All of these bodies suffered Cyanobacterial blooms, with five determined to have extensive blooms (defined as blooms covering more than 30% of the surface of the water body), 18 had intermediate coverage (between 10% and 30% of the surface covered by blooms), 13 had little coverage (between 1% and 10% of the surface covered by blooms) and 14 had insignificant coverage (less than 1% of the surface of the water body covered by the blooms). Twenty six of the reservoirs also suffered Cyanobacterial surface scum formation, though of these 23 never had scums which covered more than 1% of the surface of the water body, two (Spitskop and Darlington Dams) occasionally produced scums which exceeded 1% surface cover, but never 5%, while only Hartbeespoort Dam produced scums which covered more than 10% of the surface area.

A small group of reservoirs were found to be highly turbid (i.e. have very low clarity due to extensive organic material suspended in the water), including Mthatha, Ncora, Erfenis, Krugersdrift, Ntshingwayo and Allemanskraal Dams. These are likely to have been highly eutrophic and prone to Cyanobacterial blooms, though the high level of turbidity makes this hard to assess, and Matthews and Bernard recommend that further investigations are carried out at these bodies.

Once these reservoirs were excluded the most eutrophic bodies were Grassridge, Klipvoor and Bloemhof, while the least eutrophic were Sterkfontein, Pongolapoort and Midmar. The bodies with the highest levels of Cyannobacteria were Barberspan, Hartbeespoort and Koppies, where the average coverage exceeded 45% of the surface area, while Lake Chrissiesmeer, Spitskop and Vaal Dams all had average coverage close to 30%. Most of these bodies suffered the worst blooms in March and April – the end of the summer season.

Over the course of the study Darlington, Heyshope, Kalkfontein, Lubisi, Rustfontein, Vanderkloof and Xonxa appeared to suffer from increasing eutrophication, while this fell at Bronkhorstspruit, Kwena, Ncora, Roodekoppies, Vaalkop and Voëlvlei (though Matthews and Bernard are cautious about interpreting two much from the limited data available). The levels of Cyanobacterial blooms appeared to be increasing at Klipvoor and Vaalkop Dams, but decreasing at Barberspan, Brandvlei, Darlington and Kuhlange (Kosi Lake). Instances of Cyanobacterial scums appeared to be increasing at Lake Chrissiesmeer, Loskop, Vaal and Roodekoppies Dams.

Finally all of the water bodies were given a score from zero to nine, based upon their levels of eutrophication, frequency of Cyanobacterial blooms and development of scums. This revealed Hartbeespoort Dam to be the worst affected body, and the only one which scored a nine on the scale used, with hypertrophic water, and frequent and extensive Cyanobacterial blooms and scum events. The next worst reservoir was Spitskop with a score of seven, which also suffered high levels of eutrophication, frequent Cyanobacterial blooms and scum events. Four bodies, Darlington, Barberspan, Koppies and Lake Chrissiesmeer, scored six on the scale, all of which show high levels of eutrophication and frequent Cyanobacterial blooms. Two of these bodies, Barberspan and Chrissiesmeer, are considered to be important resources for wildlife, particularly Birds, and their poor condition is considered to be a threat to South African biodiversity. Twenty five bodies scored four or five on the scale, indicating eutrophic waters with frequent or occasional Cyanobacterial blooms, in need of management to protect their water quality, and only twelve achieved scores of two or less, indicative of oligotrophic or mesotrophic waters and low levels of Cyanobacterial activity, though none of these were completely free of Cyanobacteria, suggesting that all shoulf be the subject of ongoing monitoring and management.

The MERIS satellite ceased functioning in 2012, so further monitoring of water quality with this satellite is not possible. However the Sentinel 3 satellite, due to be launched in 2015, should provide equivalent data from 2016 until at least 2026, so the methods developed with the MERIS data do have potential use for ongoing monitoring of South African water resources. A number of other such missions are also in the pipeline, some of which will provide higher resolution spectrographic data, with the potential to expand the use of remote sensing to monitor environmental conditions in Southern Africa.

See also…

The McMurdo Dry Valleys are a largely ice-free region of Antarctica on the Ross Sea coast, discovered by Robert Scott in the early twentieth century. The...

Free-living Amoebae are unicellular protozoan’s common in aquatic ecosystems, which are also known to colonize Human created water systems such as water supply systems and swimming pools. Whilst in some ways these organisms can be helpful...

South Africa has an arid climate and a rapidly rising Human population, which can place a strain on the country’s attempts to provide clean water for its entire population. Like many countries it has begun to...

Follow Sciency Thoughts on Facebook.


  1. other organisms which can prove detrimental to the ecosystem and the health of anybody drinking the water. ACTIVATED SPRINKLER SYSTEMS AND WATER DAMAGE