Physics and Chemistry of the Earth 66 (2013) 38–44

Contents lists available at ScienceDirect

Physics and Chemistry of the Earth

journal homepage: www.elsevier .com/locate /pce

Remote sensing based water quality monitoring in Chiveroand Manyame lakes of Zimbabwe

1474-7065/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.pce.2013.09.003

⇑ Corresponding author. Tel.: +263 772 680 954.E-mail address: tmarychawira@gmail.com (M. Chawira).

M. Chawira a,⇑, T. Dube b, W. Gumindoga c

a University of Zimbabwe, Department of Biological Sciences, Box MP 167, Harare, Zimbabweb University of Zimbabwe, Department of Geography and Environmental Science, Box MP 167, Harare, Zimbabwec University of Zimbabwe, Department of Civil Engineering, Box MP 167, Harare, Zimbabwe

a r t i c l e i n f o

Article history:Available online 25 September 2013

Keywords:Chl_aEutrophicationRemote sensingTSMBlue-green algaeMERIS

a b s t r a c t

Lakes Chivero and Manyame are amongst Zimbabwe’s most polluted inland water bodies. MEdium Res-olution Imaging Spectrometry level 1b full resolution imagery for 2011 and 2012 were used to derivechlorophyll-a (chl_a) and phycocyanin (blue-green algae) concentrations using a semi-empirical bandratio model; total suspended matter (TSM) concentrations were derived from the MERIS processor.In-situ measured chl_a was used to validate the remotely sensed values. Results indicate that remotesensing measurements are comparable with in situ measurements. A strong positive correlation(R2 = 0.91; MAE = 2.75 mg/m3 (8.5%)) and p < 0.01 (highly significant)) between measured and modeledchl_a concentrations was obtained. Relationships between optically active water constituents wereassessed. Measured chl_a correlated well with MERIS modeled phycocyanin (PC) concentration(R2 = 0.9458; p < 0.01 (highly significant)) whilst chl_a and TSM gave (R2 = 0.7344; p < 0.05 (significant)).Modeled TSM and PC concentrations manifested a good relationship with each other (R2 = 9047; p < 0.001(very highly significant)). We conclude that remote sensing data allow simultaneous retrieval of differentwater quality parameters as well as providing near real time and space results that can be used by watermanagers and policy makers to monitor water bodies.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Optical remote sensing techniques provides rapid, temporal andsynoptic information on the state of the water quality variableswhich includes the hydro-physical, biological and biochemicalwith no interpretive problems that are associated with under sam-pling (Kutser et al., 2012; Koponen, 2006). Using satellite based re-mote sensing, surface reflectance is measured from a water body orwater leaving radiance of concentrations of different water constit-uents such as coloured dissolved organic matter (CDOM), chloro-phyll_a (chl_a), total suspended matter (TSM) and many others(Babin and Stramski, 2002; Hu et al., 2010; Kutser et al., 2012). Re-mote sensing of water quality monitoring came into existence in1978 when Coastal Zone Colour Scanner (CZCS) sensor waslaunched although the first satellite was launched in 1961. Thespatial resolution (pixel size) of CZSC was not suitable for monitor-ing small and moderately sized lakes. As a result, it was largelyused for monitoring oceans. Thereafter, other satellites which in-clude Sea-viewing Wide Field-of-view Sensor (SeaWiFS) and theModerate Resolution Imaging Spectroradiometer (MODIS) came

into existence with spatial resolution ranging from 250 m to1000 m. The Land Satellite Thematic Mapper (Landsat TM), a moreadvanced version observing the earth with TM band 1 developedfor bathymetric and mapping coastal areas with a pixel size of30 m by 30 m came into existence in 1999 Robert et al. (2004).

In the African continent, a handful studies has been carried outin countries such as Kenya (Lake Naivasha), South Africa (Vaaldam, Bunguela upwelling zone, Zeekoevlei in Cape Town andNatal coast) and in Nigeria Lake Chad. Studies for monitoring al-gal blooms and TSM in South Africa have been carried out byMatthews et al., 2010; Bernard et al. (2001) in Zeekoevlei eutro-phic lake and Bunguela zone respectively whilst studies on phy-toplankton primary production in South African lakes have beencarried out by Harding (1997). In the studies by Horion et al.,2010, daily bio-optical time series were derived from MODIS/Aqua imagery for Lake Tanganyika to monitor the lakes’ waterquality. In Lake Naivasha, remote sensing has been applied to re-trieve diffuse attenuation coefficient and to map euphotic depthin studies carried by Majozi et al., 2012. In the Zimbabwean situ-ation, no optical remote sensing water quality studies have beenconducted to the heterogeneous optical water quality constitu-ents. Thus this paper’s main focus is directed towards testingMEdium Resolution Imaging Spectrometer (MERIS), a sensor onboard the Environmental Satellite (EVISAT) to quantify water

M. Chawira et al. / Physics and Chemistry of the Earth 66 (2013) 38–44 39

quality parameters (chl_a, TSM and blue-green algae) in theeutrophic lakes of Manyame and Chivero.

The deterioration of water quality in Lakes Chivero and Manyamedue to eutrophication is a major public concern. These two inlandwater bodies stand as the most polluted sources of water in metro-politan Harare. The major sources of pollution consist of point andnon-point sources. The point sources consist of raw sewage fromillegal industrial and domestic waste disposal as well as poorgarbage collection by responsible city authorities (Magadza, 1997).The non-point sources consist of agricultural and partially from min-ing activities among others. The Marimba and Mukuvisi rivers (5and 8 km respectively upstream Lake Chivero) are the major effluentreceiving rivers and their water quality has been deteriorating overthe years (Marshall, 1997; Nhapi et al., 2006). Lake Manyame situ-ated immediately downstream of Lake Chivero, maintained its waterquality for some time with nitrogen rather than phosphorus beingthe main limiting nutrient but recent work suggests that situationhas changed and the lake is now showing evidence of eutrophica-tion. This reflects both the deterioration of Lake Chivero, whichsupplies most of its water, and the growth of the town of Norton,located on the southern shore of the lake.

These two lakes are the primary source of water for Harare andChitungwiza and, apart from the difficulty and cost of water puri-fication there are concerns about the effects of algal toxins in thewater. This issue was first raised in the 1960s (Zilberg, 1966) andother authors have determined the concentration of these toxinsin the lake (Ndebele and Magadza, 2006). It is essential, therefore,that the concentration of chlorophyll_a is monitored in these lakesbut conventional methods are expensive and labour-intensive, eco-logically destructive and also cannot capture the spatial heteroge-neity in the Lakes within the lake (Dube, 2012). Previous studies ofLake Chivero, used remote sensing techniques based on Normal-ized Difference Vegetation Index (NDVI) to assess the water qualityand found that NDVI was correlated with the chl_a and total sus-pended matter (TSM) (Shekede et al., 2008). This investigationdid not provide full and direct measures of optical water qualityparameters, thus leading to an incomplete water quality assess-ment in the two lakes.

These parameters are spatially heterogeneous and, therefore,point based measurements are not adequate to provide informa-tion on their spatial coverage. Therefore, the objective of this studyis to combine in situ chl_a measurements and remote sensing tech-niques to quantify chl_a, and phycocyanin (PC) pigment as a proxyfor blue-green algae and TSM concentrations from MERIS image.

2. Methods and materials

2.1. Study area

Lake Chivero formerly known as Lake Mcllwaine is a reservoirconstructed in 1952 on the Manyame River 35 km SW of Harare,Zimbabwe. It has a surface area of 26 km2, a volume of 250 � 106

m3and and maximum depth of 27 m. The widest point stretchesfor about 8 km and shoreline of the entire water body coversapproximately 48 km. Lake Manyame formerly Darwendale damwas constructed between 1973 and 1976. It is located some76 km west Harare downstream of Lake Chivero. Its water coversa surface area of 8100 hectares with a maximum depth of depthof 23.6 m and is much larger than Lake Chivero.

Lake Chivero has, in the past, experienced major outbreaks ofwater hyacinth Eichhornia crassipes but it has been reducedthrough biological control (Chikwenhere and Phiri, 1999). Itseutrophic state has stimulated the growth of other macrophytesand the shoreline supports a dense growth of alien plants suchas Myriopyllum, Hydrocotyle and Limnobium.

2.1.1. Field measurements and sample analysesChlorophyll_a was measured from in samples taken from six

locations in Lake Chivero on 23 March, 2012 (Fig. 1) using the lab-oratory analysis method. Chlorophyll_a was not determined inLake Manyame and, instead, chl_a, TSM and PC concentrationswere directly derived from MERIS imagery in both lakes fromwhich concentration and distribution maps were derived. Mea-sured chl_a concentrations from Lake Chivero for 23 March 2012MERIS image were used to validate modelled chl_a concentrations(Table 1). The coordinates of the sampling stations in the lake(Fig. 1) were matched with the satellite images. The locations ofthe sampled sites are indicated in (Fig. 1). Water samples were col-lected in 1-l brown bottles from a depth of no more than 1metre,and stored in cooler box with ice pack to avoid degradation ofthe chlorophyll by sunlight as recommended in (Bartram andBallance, 1999). The samples were pre-processed within a periodof 24 h. Samples were first filtered using paper fibre filters(Whatman GF/C) and chl_a was extracted using 90% acetone inan agate mortar, which was then centrifuged at 3000 r.p.m for10 min. Colour absorbance determination was derived at 665,643, 630 and 750 nm as indicated in (APHA et al., 2001; WorldHealth Organisation, 2003).

The relationship between measured chl_a for 23 March 2012was assessed with other optically active constituents (TSM andPC) derived from the MERIS of same day. It is usually assumed that,if weaker relationships among pigment concentrations for exam-ple, chl_a and PC and residents such as TSM exist, it would suggesthigher amounts of non-algal constituents in a water column andvice versa (Randolph et al., 2008).

2.2. Remote sensing data acquisition and pre-processing

MERIS full resolution (FR) level 1b imagery with a high spec-tral resolution and an on-spot re-visitation period of 2–3 dayswas used to determine the concentrations of chl_a, PC, andTSM. MERIS spectrometer was found to be well suited to viewLakes Manyame and Chivero as these lakes are relatively largerin size, more than 1 km2 in total area. The sensors’s narrow spec-tral bands are also well suited for deriving water quality constit-uents (Matthews et al., 2010). An atmospheric correction (AC)was done using the Case 2 regional processor (C2R), which is atool developed by Doerffer and Schiller, 2007. Path radianceand transmittance of bands between 412 nm �708 nm, derivedfrom the top of atmosphere (TOA) directional radiance reflec-tance in the NIR spectral range band of 708–870 nm, using aneural network (NN) algorithm (Guanter et al., 2010). The AC isa pre-requisite for quantitative remote sensing methods that re-quire images to be calibrated for surface reflectance. Literaturepoints out that the removal of adjacency effects (an effect whichresults when photons are reflected and scattered towards thesensor and where a substantial contrast exits between the darktarget/water and its surroundings) is crucial as it can cause inac-curacies in the estimation of aerosol optical thickness (AOT) atthe wavelengths used for the estimation of water quality param-eters (Brockmann Consult, 2010; Guanter et al., 2010; Kratzerand Vinterhav, 2010). Fortunately, correction schemes such asICOL take this effect into account, otherwise the increased radi-ance is mistakenly associated with other physical processes,and the derived geophysical quantities have an increased error.ICOL is a prototype ocean and land processor which correctsfor stray land reflection in the dark object pixels in relation tothe sun angle, taking the coupling between Rayleigh scatteringand Fresnel reflection into account Santer et al. (2007). Thus,the Improved Contrast over Ocean and Land (ICOL) methodologywas considered in this study.

Table 2Characteristics of Lake Chivero MERIS Satellite Derived Water Quality Parameters from se

Fig. 2. Reflectance spectra and absorption troughs of PC and chl_a pi

Fig. 1. Study areas showing sampling points in Lake Chivero.

Table 1MERIS match-up sites in Lake Chivero.

Date Sensing start time Lake Chivero

Longitude Latitude

21/12/2011 07:54:34 30�4904900E 17�5405200S24/12/2011 07:44:33 30�4703500E 17�5305400S12/03/2012 07:50:38 30�4605600E 17�5301600S23/03/2012* 07:47:19 30�4800300E 17�5405200S

* Chl_a data used for validation.

40 M. Chawira et al. / Physics and Chemistry of the Earth 66 (2013) 38–44

2.3. Blue-green algae and chl_a

The semi empirical band ratio model developed for opticallycomplex waters (Simis et al. 2005) was used to estimate PC andchl_a concentrations from the 620 nm and 665 nm bands, respec-tively. The algorithm uses the band configuration of the MERISchannels 6, 7, 9 and 12 centered at 620, 665, 709 and 778.75 nmrespectively. The 620 nm band is dominated by both phycocyaninpigment (PC) and chl_a, and the 665 nm band by chl_a alone,

lected pixels.

gments derived from MERIS over (a) Chivero and (b) Manyame.

Table 3Lake Manyame MERIS Derived Water Quality Parameters.

Date Point Chl_a (mg/m3) PC (mg/m3) TSM (g/m3)

23/03/2012 5 0.0789 29.6 0.046 37.9 103.3 10.57 19.7 55.4 28 26.1 83.9 11.79 23 82 12

10 0.11 30.8 0.05

M. Chawira et al. / Physics and Chemistry of the Earth 66 (2013) 38–44 41

whilst the 709 nm and 778.75 nm bands are dominated by wateralone (Ruiz-Verdú et al., 2008). The absorption by coloured dissolvedorganic matter (CDOM) and suspended particulate matter (SPM) isconsidered negligible, which may result in an overestimation of(absorption by water that is not necessarily correlated with pigmentconcentrations (Gons et al., 2005; Simis et al., 2007). Thus, PC andchl_a concentrations derived from MERIS images are expected tobe slightly higher than the concentrations measured in the field.

The 620 nm band is dominated by both PC and chl_a, and the665 nm band by chl_a alone, whilst the 709 nm and 778.75 nmbands are dominated by water alone (Ruiz-Verdú et al., 2008).The absorption by coloured dissolved organic matter (CDOM) andsuspended particulate matter (SPM) is considered negligible,which may result in an overestimation of absorption by water thatis not necessarily correlated with pigment concentrations (Gonset al., 2005; Simis et al., 2007). Thus, PC and chl_a concentrationsderived from MERIS images are expected to be slightly higher thanthe concentrations measured in the field.

2.4. Total suspended matter determination

Total suspended matter is one of the major apparent opticalproperties (AOPs) of water bodies (Song et al., 2011). In this study,

R2 = 0.7344P value = 0.0292

Slope = 3.291+- 0.9896

TSM (g/m3 )

Chl

_a (m

g/m

3 )

0 5 10 150

10

20

30

40

50 (c)

3

R2 = 0.9111P value = 0.0031

Slope = 0.5784+- 1.464

Measured(ch_a (mg/m3)

Pred

icte

d Ch

l_a

(mg/

m3 )

0 10 20 30 400

10

20

30

40

50 (a)

Fig. 3. Relationships for image dated 23/03/2012; (a) between in situ and predicted chl_aconcentration and (d) predicted PC and TSM.

the TSM data were based on the MERIS eutrophic optical modelprocessor and was derived using a scattering coefficient in the re-gion between 690 nm and 730 nm (Dekker et al., 2002).

3. Results

3.1. Reflectance spectra of Lakes Chivero and Manyame

The reflectance spectra for Lakes Chivero and Manyame were allsimilar in their pattern and shape (Fig. 2). The conspicuous absorp-tion dips at 620 nm and 665 nm is can be attributed to highabsorption by the blue-green algae (PC) and chl_a pigmentsrespectively. The reflectance of Lake Chivero (Fig. 2(a)) was gener-ally lower than that of Lake Manyame (Fig. 2(b)) indicating that itspigment concentration was much higher.

3.2. Validation of MERIS water quality parameters

Three water quality variables from Lake Chivero (chl_a, PC andTSM concentrations) were derived from MERIS imagery (Table 2)on 21 and 24 December 2011 as well as 12 and 23 March 2012.Generally, optical conditions exhibited a highly (>20 mg/m3)eutrophic state with very high concentrations of pigments, wellabove those of World Health Organization (WHO, 2003) recrea-tional guidelines. In situ measured chl_a for 23 March 2012 fromLake Chivero was used to validate the MERIS modeled chl_aconcentrations. For comparison purposes, MERIS derived LakeManyame water quality parameters results for 23 March 2012,were also included (Table 3).

3.3. Relationships between optically active water constituents

The relationship assessed on the water quality parameters werebased on MERIS data for 23 March 2013. There was a linear

R2 = 0.9458P value = 0.0011

Slope = 0.4750 +- 0.05682

PC (mg/m3)

Chl

_a (m

g/m

3 )

0 50 100 1500

10

20

30

40

50

R2 = 0.9047P value = 0.0035

Slope = 0.210+- 0.01963

TSM (g/m3)

PC (m

g/m

)

0 5 10 150

50

100

150 (d)

(b)

; (b) between modeled TSM and PC concentration; (c) between in situ chl_a and TSM

Fig. 4. Maps showing concentration and distribution patterns of TSM in lakes Manyame (a) and Chivero (b).

Fig. 5. Maps showing concentration and distribution patterns of chl_a pigment inlakes Manyame (a) and Chivero (b) from MERIS image 2012.

42 M. Chawira et al. / Physics and Chemistry of the Earth 66 (2013) 38–44

relationship between chl_a and PC concentration in Lake Chivero(Fig. 3(a)) with a close correlation between measured and pre-dicted chl_a concentrations in the samples obtained on 23 March2012 (R2 = 0.9111, P < 0.01 (highly significant) and MAE of2.75 mg/m3 with 8.5% error). There was also a significant correla-tion (Fig. 3(c)) and between TSM and PC (Fig. 3(d)) with(R2 = 0.9047, and a significant P < 0.01) whilst TSM had a somewhatweaker correlation with chl_a (R2 = 0.7344, P < 0.05). However, themodel presumably over-predicted chl_a where it reached satura-tion point (no further increase in living biomass) with TSM contin-uing to increase after chl_a saturation point. This suggests thatvery high levels of TSM could have originated from sediment, min-eral as well as from dead material of living organisms which in-cludes chl_a, thus an indication of ecological dynamics. The PCvalues (Fig. 3(b)) estimated by using the semi empirical algorithmwere strongly correlated with the measured chl_a concentrations(R2 = 0.9458, P < 0.001 (very highly significant)).

3.4. TSM distribution maps

The TSM varied from 0.04 g/m3 to 17.6 g/m3 on the dates indi-cated in (Fig. 4). As typical Case-2 water bodies, Lakes Manyameand Chivero had high TSM concentrations due to soil erosion, mi-cro pollutants (Nhapi and Tirivarombo 2004) and non-point sourcepollution. TSM concentrations were low on 21 and 24 Decemberbut increased remarkably between 12 and 23 March 2012 due tohigh sediment loads feeding into the lakes as a result of the rainyseason (from November onwards).

3.5. Chl_a distribution maps

The concentrations of chl_a ranged from 0.05 mg/m3 to39.4 mg/m3 with lower concentrations being noted in the areaalong the shoreline. This could be attributed to wind effects whichsweep algae towards the central parts of the water bodies wherethey become concentrated and lead to huge blooms. A glance atLake Chivero and Manyame maps shows that concentrations ofchl_a also were in the higher range (14–39 mg/m3) at this time(Fig. 5).

3.6. Blue-green algae distribution maps

The concentrations of PC (Fig. 6) were highest in Lake Chiverowhile the distribution patterns were almost uniform in Lake Many-ame. However, there was a large gradient in PC concentrations be-tween 21 and 24 December in Lake Manyame. The massive bloomon 21 December 2012 could have been triggered by a number offactors, including, high water temperatures, nutrient concentra-tions and the buoyancy regulating mechanism of blue-green algae.Furthermore, abundances of blue-green algae are known to vary onshort time scales, as in hours or days, which also could result in thebig difference which was noticed between 21 and 24 Decembermaps. The concentrations of blue-green algae (Fig. 6) ranged be-tween 27 and 121 (mg/m3).

4. Discussion

The remote sensing measurements obtained during this investi-gation were comparable to in situ measurements suggesting that

Fig. 6. Concentration and distribution patterns of PC pigment in lakes (a) Manyame and (b) Chivero.

M. Chawira et al. / Physics and Chemistry of the Earth 66 (2013) 38–44 43

satellite-based methods may be useful for assessing the trophicstatus of Zimbabwean water bodies. The distinct reflectance peakin the two lakes at 685 nm appears to be an effect of sun-inducedchl_a fluorescence (Xu et al., 2009). The two absorption troughs at620 nm and 665 nm are an indication similar to those obtained inLake Taihu, China, and the Geist reservoir, Indiana, with respect tothe absorption peaks of the two pigments (Dingtian and Pan, 2006;Randolph et al., 2008).

These water quality parameters variables indicate that LakesManyame and Chivero are highly eutrophic and variables werepositively correlated with each other. The correlation betweenTSM and chl_a indicated that TSM included the backscatter causedby the inorganic parts of the phytoplankton (Dekker et al., 2002;Song et al., 2011). The concentrations of TSM (Fig. 3) were higherin Lake Manyame than in Lake Chivero therefore, the reservoir’sapparent reflectance which indicated a high TSM backscatteringin the infrared and NIR bands.

5. Conclusions

In this study three conclusions have been drawn:

i. Water quality parameters of the two lakes (Chivero andManyame) were able to be successfully derived from remotesensing data. Thus MERIS is suitable for real-time monitor-ing applications and observing changes happening overshort time scales.

ii. The semi empirical band ratio based model proved to be acrucial utility in detecting PC and chl_a pigment concentra-tions and their spatial distributions.

iii. Remote sensing techniques demonstrated that the two lakesare typical Case 2 water bodies dominated by chl_a, blue-green algae and total suspended matter, which ultimatelydefine the lakes as eutrophic.

Acknowledgements

The authors would like to thank Stefan Simis who developedthe algorithm that is used in this paper. We would also like tothank the European Space Agency (ESA) for providing full-timeresolution MERIS images over the two lakes.

References

APHA, AWWA, WEF, 2001. In: APHA. (Ed.), Standard methods for the examination ofwater and wastewater. Washington DC, USA.

Babin, M., Stramski, D., 2002. Light absorption by aquatic particles in the near-infrared spectral region. Limnology and Oceanography 47, 911–915.

Bartram, J., Ballance, R., 1999. Water quality monitoring: a practical guide to thedesign and implementation of freshwater quality studies and monitoringprogrammes (Retrieved from info@routledge.com).

Bernard, S., Probyn, T.A., Barlow, R.G., 2001. Measured and modelled opticalproperties of particulate matter in the southern Benguela. South African Journalof Science 97, 410–420.

Brockmann Consult, 2010. ICOL+ Software User Manual.Chikwenhere, G.P., Phiri, G., 1999. History of water hyacinth and its control efforts

on Lake Chivero in Zimbabwe. In: Hill, M.P., Julien, M.H., Center, T.D. (Eds.),Proceedings of the First IOBC Global Working Group Meeting for the Biologicaland Integrated Control of Water Hyacinth, 16–19 November, Harare, Zimbabwe.

Dekker, A.G., Vos, R.J., Peters, S.W.M., 2002. Analytical algorithms for lake waterTSM estimation for retrospective analyses of TM and SPOT sensor data.International Journal of Remote Sensing 23 (1), 15–35.

Dingtian, Y., Pan, D., 2006. Hyperspectral retrieval model of phycocyanin in case IIwaters. Chinese Science Bulletin 51, 149–153.

Doerffer, R., Schiller, H., 2007. The Medium Resolution Imaging Spectrometer.International Journal of Remote Sensing 28 (3–4), 517–535.

Dube, T., 2012. Primary Productivity of Intertidal mudflats in the Wadden Sea: ARemote Sensing Method. University Twente-ITC.

Gons, H., Rijkeboer, M., Ruddick, K.G., 2005. Effect of a waveband shift onchlorophyll retrieval from MERIS imagery of inland and coastal waters.Plankton Research 27, 125–127.

Guanter, L., Ruiz-Verdú, A., Odermatt, D., Giardino, C., Simis, S., Estellés, V., Heege, T.,Domínguez-Gómez, J.A., Moreno, J., 2010. Atmospheric correction of ENVISAT/MERIS data over inland waters: validation for European lakes. Remote Sensingof Environment 114 (3), 467–480.

Harding, W.R., 1997. Phytoplankton primary production in a shallow, well-mixed,hypertrophic South African lake. Hydrobiologia 344, 87–102.

Horion, S., Bergamino, N., Stenuite, S., Descy, J., Plisnier, P., Loiselle, S.A., Cornet, Y.,2010. Optimized extraction of daily bio-optical time series derived fromMODIS/Aqua imagery for Lake Tanganyika, Africa. Remote Sensing ofEnvironment 114, 781–791.

Hu, C., Lee, Z., Ma, R., Yu, K.L., Dshang, S., 2010. Moderate Resolution ImagingSpectroradiometer (MODIS) observations of cyanobacteria blooms in TaihuLake, China. Journal of Geophysical Research 115, 1–22.

Koponen, S., 2006. Remote sensing of water quality: the development and use ofwater processors available in BEAM. Finland, Helsinki: Helsinki University ofTechnology TKK, Department of Radio Science and Engineering, pp. 1–58.

Kratzer, S., Vinterhav, C., 2010. Improvement of MERIS level 2 products in Baltic Seacoastal areas by applyingz the Improved Contrast between Ocean and Landprocessor (ICOL) – data analysis and validation. Journal of Basic Research inMarine Sciences with Emphasis on European Seas 52, 211–236.

Kutser, T., Paavel, B.C., Verpoorter, C., Kauer, T., Vahtmäe, E., 2012. Remote sensingof water quality in optically complex lakes. International Archives of thePhotogrammetry, Remote Sensing and Spatial, Information Sciences, vol.XXXIX-B8.

Magadza, C.H.D., 1997. Water pollution and catchment management in LakeChivero. In: Moyo, N.A.G. (Ed.), Lake Chivero: a Polluted Lake. University ofZimbabwe Publications/IUCN, Harare.

44 M. Chawira et al. / Physics and Chemistry of the Earth 66 (2013) 38–44

Majozi, N.P., Salama, S.M., Bernard, S., Harper, D.M., 2012. Water quality and remotesensing: a case study of Lake Naivasha, Kenya. CSIR Conference Presentation, 1–13.

Marshall, B.E., 1997. Lake Chivero after forty years: the impact of Eutrophication. In:Lake Chivero a Polluted Lake, pp. 1–12 (Chapter 1).

Matthews, M.W., Bernard, S., Winter, K., 2010. Remote sensing of cyanobacteria-dominant algal blooms and water quality parameters in Zeekoevlei, a smallhypertrophic lake, using MERIS. Remote Sensing of Environment 114, 2070–2087.

Ndebele, M.R., Magadza, C.H.D., 2006. The occurrence of microcystin-LR in LakeChivero, Zimbabwe. Lakes and Reservoirs Research and Management, 57–62.

Nhapi, I., Tirivarombo, S., 2004. Sewage discharges and nutrient levels in MarimbaRiver, Zimbabwe. Water SA 30 (1).

Nhapi, I., Siebel, M.A., Gijzen, H.J., 2006. A proposal for managing wastewater inHarare, Zimbabwe. Water and Environment Journal 20, 101–108.

Randolph, K., Wilson, J., Tedesco, L., Li, L., Pascual, D.L., Soyeux, E., 2008.Hyperspectral remote sensing of cyanobacteria in turbid productive waterusing optically active pigments, chlorophyll a and phycocyanin. Remote Sensingof Environment 112 (11), 4009–4019.

Robert, K.V., Xiaoming, Q., Michael, R., McKayb, L., Minerb, J., Czajkowskic, K.,Savinod, J., Bridgeman, T., 2004. Phycocyanin detection from LANDSAT TM datafor mapping cyanobacterial blooms in Lake Erie. Journal: Remote Sensing ofEnvironment 89, 381–392.

Ruiz-Verdú, A., Simis, S.G.H., de Hoyos, C., Gons, H.J., Peña-Martínez, R., 2008. Anevaluation of algorithms for the remote sensing of cyanobacterial biomass.Remote Sensing of Environment 112 (11), 3996–4008.

Shekede, M.D., Kusangaya, Schmidt, K., 2008. Spatio-temporal variations of aquaticweeds abundance and coverage in Lake Chivero, Zimbabwe. Physics andChemistry of the Earth 33 (8), 8–13.

Simis, S.G.H., Peters, S.W.M., Gons, H.J., 2005. Remote sensing of the cyanobacterialpigment phycocyanin in turbid inland water. Limnology and Oceanography,237–245.

Simis, S.G.H., Ruiz-Verdú, A., Domínguez-Gómez, J.A., Peña-Martinez, R., Peters,S.W.M., Gons, H.J., 2007. Influence of phytoplankton pigment composition onremote sensing of cyanobacterial biomass. Remote Sensing of Environment 106(4), 414–427.

WHO, 2003. Guidelines for safe recreational water environments, vol. 1, Geneva, WHO.