Abstract

Frequent and continuous water quality monitoring of Olushandja Dam in Namibia is needed to inform timely decision making. This study was carried out from November 2014 to June 2015 with Landsat 8 reflectance values and field measured water quality data that were used to develop regression-analysis-based retrieval algorithms. Water quality parameters considered included turbidity, total suspended solids (TSS), nitrates, ammonia, total nitrogen (TN), total phosphorus (TP) and total algae counts. Results show that turbidity levels exceeded the recommended limits for raw water for potable water treatment while TN and TP values are within acceptable values. Turbidity, TN, and TP and total algae count showed a medium to strong positive linear relationship between Landsat predicted and measured water quality data while TSS showed a weak linear relationship. The regression coefficients between predicted and measured values were: turbidity (R2 = 0.767); TN (R2 = 0.798,); TP (R2 = 0.907); TSS (R2 = 0.284,) and total algae count (R2 = 0.851). Prediction algorithms are generally the best fit to derive water quality parameters. Remote sensing is recommended for frequent and continuous monitoring of Olushandja Dam as it has the ability to provide rapid information on the spatio-temporal variability of surface water quality.

HIGHLIGHTS

  • Over past years, frequent and continuous water quality monitoring has been problematic in Namibia.

  • A linear regression can now be used to develop algorithms for retrieving water quality data.

  • Good prediction accuracy for turbidity, TN, TP and total algae count.

  • More sampling points needed to further improve regression model accuracy.

  • Remote sensing provides rapid information on water quality spatio-temporal variability.

Graphical Abstract

Graphical Abstract
Graphical Abstract

INTRODUCTION

Fresh water is a finite resource that is essential for human existence (Bhuyar et al. 2019a, 2019b; Bhuyar et al. 2020). Without freshwater of adequate quantity and quality, sustainable development will not be possible (UN-Water 2011; McMillan et al. 2017). Effective management of this scarce resource ensures that present generations' needs do not deprive future generations the same privileges of access in both quantity and quality as envisioned in the Sustainable Development Goals (SDGS) (Bain et al. 2020). Therefore, Integrated Water Resources Management (IWRM) is an essential approach as it emphasises effective management of water resources within the basin.

In the Southern Africa Region, pollution of both surface and ground water is on the increase, particularly from mining, agricultural and industrial activities. Man-made reservoirs such as dams and lakes are threatened by nutrient enrichment and heavy metals and their water quality is continuously degraded as a result (Oberholster & Ashton 2008; Nhapi 2009; Lehmann 2010). For example, the water quality in Von Bach and Swakoppoort Dam in the Central Area of Namibia has been reported to be poor due to treated and at times partially treated wastewater from the City of Windhoek and Okahandja town (Lehmann 2010; NamWater 2012).

In the north-central regions of Namibia, poor sanitation practices are notably high (IWRMPJVN 2010), with about 67% of the population having no access to improved sanitation facilities. North-central Namibia is more vulnerable to effects of climate change and variability. Combined effects of environmental degradation, social vulnerability to poverty and a changing climate will compromise water and sanitation provision (Angula & Kaundjua 2016). Flash floods are further predicted to impact overall sanitation and human health conditions. Therefore, during high rainfall events, cholera outbreak has been reported due to wash-away of water and wastewater aided by poor and inadequate sanitation facilities in the areas (UN 2011).

Shuuya & Hoko (2014) found that the quality of water in the Calueque–Oshakati canal is deteriorating from upstream to downstream due to human activities along the canal such as agriculture, settlements and mining. All these activities tend to affect the quantity and quality of water leading to an increase in the cost of purifying water for human consumption. IWRMPJVN (2010) noted that lack of data, continuous monitoring and poor data management exist in the Namibian water sector. Water quality monitoring in many southern African countries, including Namibia, has been noted to be hindered by lack of data, capacity and resources. Water quality monitoring in Namibia has been noted to be one of the key issues that requires urgent attention by researchers and water managers.

Few studies that are available for the Kunene River and Cuvelai Basins in the north-central regions of Namibia focus on ad-hoc monitoring. Studies have focused on the quality of water in the canals that bring water into the Olushandja Dam and to the treatment plants (Shuuya & Hoko 2014) and transport water from the dam (SDP10 2001). These studies mainly employ traditional in-situ methods of determining water quality. According to IWRMPJVN (2010), determining water quality using traditional techniques is costly and may be one of the major contributors to poor monitoring frameworks not only in Namibia but also in many developing countries in the world. Ritchie et al. (2003) stated that traditional methods for assessing and monitoring water quality are expensive and time-consuming. In addition, they do not give the spatial or temporal views of water quality needed for accurate assessment of water bodies, therefore there is a need for more robust techniques which have a spatial and a temporal dimension (Kallio 2000).

Thus, this study explored the applicability of remote sensing in combination with in-situ observed measurements to develop algorithms and prediction of selected water quality parameters at Olushandja Dam. Remote-sensing-based water quality assessment is an economical way to monitor water quality, since it allows routine monitoring of large areas in a short time and on a repetitive basis (Hellweger et al. 2004; Somvanshi et al. 2012). According to Ritchie et al. (2003), the use of remote sensing in water quality dates back to the 1970s. Namibia is one of the countries with the greatest number of sunshine days or cloud-free skies, which make it easy for optical remote sensing application in the visible and infrared regions of the electromagnetic spectrum. Information obtained from this study will help the institutions responsible for management of Olushandja Dam such as NamWater and Namibian community-based water management (CBWM; Kelbert 2016) to carry out continuous monitoring of the quality of water in an economical way for decision-making.

However the major constraints would be the lack of reliable retrieval algorithms, cost of satellite data and equipment for in-situ, on-site and laboratory measurements of water quality parameters (Bauer et al. 2007). For the retrieval of water quality parameters from different satellite sensors, a number of methods have been used which range from empirical, semi-empirical to analytical techniques (Schalles et al. 1998; Wang & Ma 2001; Dekker et al. 2002; Brando & Dekker 2003; Vignolo et al. 2006; Chen et al. 2007; He et al. 2008; Maillard & Pinheiro Santos 2008; Salama et al. 2009; Olet 2010; Chawira et al. 2013; Kibena et al. 2014). The advantage of satellite remote sensing techniques lies in their capacity to allow managers to detect and control the pollutants before they reach alarming levels. The water quality parameters that have been assessed include chlorophyll-a, suspended matter and turbidity as they are most likely to change the water colour (Schalles et al. 1998; Li 2009; Salama et al. 2009; Olet 2010; Kibena et al. 2014). In Belbok Spruit, South Africa, du Plessis et al. (2014) predicted water quality from landcover changes using partial least squares regression analysis. Yang & Jin (2010) predicted NO3 using the spatial regression method for Lowa River in the United States and obtained an acceptable R2 of 0.8. Most researchers in sub-Saharan Africa have also applied earth observation techniques in the retrieval of water quality parameters mainly for lakes and dams (Chawira et al. 2013; Kibena et al. 2014). There is still need to explore in greater depth the application of earth observation techniques for predicting water quality parameters, especially in developing countries where land use changes are pertinent (du Plessis et al. 2014; Chapra 2015).

A few studies have attempted to monitor and model nutrients (such as total nitrogen, phosphorus, nitrate etc.) and proven the ability of remote sensing for these predictions (Alparslan et al. 2007; He et al. 2008; Chen & Quan 2012). Most of the nutrient prediction models developed so far are based on statistical regression approaches. Predicting water quality characteristics from remote sensing requires ground-truthing and validation data (Schaeffer et al. 2013). Therefore, assessment and monitoring of water quality using the combination of remote sensing and in-situ measurements plays a significant role in providing reasonable and accurate optical constituents of water (Salama et al. 2009).

Remote-sensing-based water quality assessment and monitoring may use the same method of retrieval or predicting water quality parameters, but the available sensor types may differ (Tomppo et al. 2002; USGS 2013). These include Moderate Resolution Imaging Spectroradiometer (MODIS), Medium Resolution Imaging Spectrometer (MERIS), SPOT, and Landsat, which differ in spatial, temporal, and number of spectral bands and pixels. Landsat, used in this study, has a fair revisit time (16 days) and better resolution (30 m) than most medium-resolution sensors. In addition, Landsat imagery is available free of charge from an archive hosted by the USGS Earth Resources Observation and Science (EROS) Center. The Landsat mission represents the longest continuous satellite record of the Earth, beginning in 1972 with Landsat 1 and currently with the operation of Landsat 8 (USGS 2013). The different Landsat satellite mission images have been used in a number of water quality studies (Zuccari Fernandes Braga et al. 1993; Wang et al. 2004; Vignolo et al. 2006; He et al. 2008; Olet 2010; Somvanshi et al. 2012; Waxter 2014; Bonansea et al. 2015). The best correlations between remote sensing signal and in-situ observed water quality parameters have been found mainly in the visible (blue, green, red) and near-infrared spectral range.

The objectives of this study were: (i) to characterize the status of the quality of water in Olushandja Dam through in-situ observed measurements, (ii) develop algorithms for predicting selected water quality parameters through satellite and in-situ observed measurements and (iii) to predict selected water quality parameters from remote sensing as a framework for continuous monitoring of water quality in Olushandja Dam.

STUDY AREA

Olushandja Dam is located in the upper western part of Omusati Region in the Olushandja Sub-Basin of the Cuvelai-Etosha Basin in north central Namibia (Figure 1). The Namibian part of the Cuvelai Basin covers an area between the Okavango and Kunene Rivers and ends up in the Etosha Pan. The basin covers an area of approximately 250,053 km2 from the Namibia–Angola border terminating in the Etasha Pan (Kolberg 2002). The Cuvelai lies within a relatively small depression along the western margins of the vast Kalahari Basin that covers much of south-central Africa. The basin is made up of shallow pans known as ‘Iishana’, which form a network of shallow ephemeral river systems.

Figure 1

The study area in relation to Cuvelai-Etosha Basin.

Figure 1

The study area in relation to Cuvelai-Etosha Basin.

The Cuvelai Basin in north-central Namibia comprises a unique system of seasonal wetlands and constitutes some of the areas having the highest human population densities in Namibia (50–100 persons/km2). The region carries about 40% of Namibia's population of about 2.2 million people. The Cuvelai-Etosha Basin and Namibia as a whole are characterized by a semi-arid to arid climate and desert environment (Mendelsohn et al. 2000, 2013). The average temperatures vary between 20 °C and greater than 22 °C in most areas. This translates to potential evaporation that ranges between 2,600 and 3,200 mm per year, much higher than the yearly amount of rainfall (CuveWaters 2014). The geomorphology of the Cuvelai Basin is complex. Mendelsohn et al. (2016) has provided a comprehensive description of the Cuvelai Basin landscapes and how the characteristics of these landscapes have influenced human settlement patterns.

The Olushandja Dam lies between 1,100 and 1,150 metres above sea level (CuveWaters 2014). Olushandja Dam has a surface area of about 29.0 km2 and capacity of around 42.331 Mm3 when full (NamWater 2013). The dam is about 20 km long and varies between 200 m and 2,000 m in width (Mendelsohn et al. 2000). The dam was built in 1973 during the country's liberation struggle in the old Etaka Channel to act as a balancing reservoir that stores water during excess flows of the Kunene River. In addition, the dam was built to provide a strategic reserve in the event of supplies from Calueque Dam in Angola being interrupted. In 1990 the system was renovated and now raw surface water is transferred from Calueque Dam in Angola through a concrete open canal into the dam and also to four treatment plants in the region. In recent years, the dam has been held at 50% capacity and since its construction, more than 100 households have been built in the dam flood plain (Mendelsohn et al. 2000; NamWater 2013).

MATERIALS AND METHODS

Field data collection

Sampling points (Figure 2) in the dam were selected systematically to reflect on the impacts of major tributaries' landcover activities on the dam water quality. Parameters were also selected taking into account the relationship between pollution sources and different water quality parameters including their potential impacts on human, aquatic life and environmental health. In addition, the selection considers water quality parameters that are known to be successfully retrieved from satellite data. Turbidity, algae and nutrients are known to have been successfully retrieved from satellite data (Zuccari Fernandes Braga et al. 1993; Hellweger et al. 2004; Wang et al. 2004; He et al. 2008; Li 2009; Olet 2010; Irenosen et al. 2012; Chawira et al. 2013; Waxter 2014; Bonansea et al. 2015).

Figure 2

Sampling point locations on a geometrically corrected Landsat 8 OLI image of 2015.

Figure 2

Sampling point locations on a geometrically corrected Landsat 8 OLI image of 2015.

Representative water samples were collected from six systematically selected locations in the Olushandja Dam following the Landsat imagery capturing calendar obtained from the USGS (earthexplorer.usgs.gov). At each sampling point, grab water samples were taken at about 10 cm depth below the surface using three 500 ml bottles with screw-caps. The sampling campaigns were done twice a month between January and April 2015 from 08:30 am to 14:00 pm. On the same days that satellite images were taken, the in-situ observed measurements were done because a time difference between in-situ observed measurements and satellite measurements is an important aspect (Hellweger et al. 2004). The water samples were collected, preserved, transported and analysed according to the Standard Methods for Water and Wastewater described by Wait et al. (2020).

Satellite data retrieval

Landsat 8 images of Path 180, Row 072, which covers Olushandja Dam, were downloaded from the Landsat imagery archive hosted by the USGS Earth Resources Observation and Science (EROS) Centre using the USGS Globalization Visualization Viewer (Glovis) tool. Figure 3 displays a sample Landsat 8 image tile for 15 April 2015 that covers Olushandja Dam.

Figure 3

Landsat 8 image tile of path 180, row 072, showing the location of Olushandja Dam.

Figure 3

Landsat 8 image tile of path 180, row 072, showing the location of Olushandja Dam.

The Digital Elevation Model (DEM) map covering the study area was obtained from USGS Earth Explorer (http:/www.earthexplorer.usgs.gov) and four ground control points (GCP) were established. The dam outline and canal route were obtained by digitizing a Google Earth satellite image and converting to a shape file (.shp) format in Quantum Geographical Information System (GIS or QGIS) software. All files were then imported into ILWIS and were assigned to the same coordinate system as that of the study area.

The Landsat 8 imagery (cloud-free or with cloud cover of <10%) over the area was selected for analysis. Radiometric calibration to convert digital number (DN) values to physical units, at sensor spectral radiance (watts/(m2/srad/μm)) and then to top-of-atmosphere (TOA) reflectance was done in GIS software, Environment for Visualizing Images (ENVI 5.0.3) using the same equations described by Chander et al. (2009) and USGS (2013). In the ENVI program, a band math expression was applied to rescale the floating reflectance value to surface reflectance of between 0 and 1.

After the radiometric calibration and atmospheric correction, pre-processed bands with reflectance were imported in the Integrated Land and Water Information System (ILWIS) GIS environment. The image raster maps were geometrically corrected by using the nearest-neighbour resampling technique and assigning the correct georeference as that of the study area. This study used pixel reflectance values for the four visible and near-infrared bands (2, 3, 4 and 5) of the Landsat 8 images for each selected date based on the GPS locations of every sampling point. This was done in ILWIS through the ‘map cross’ operation that matched up the sampling point raster map with a raster map of each processed band (after image re-sampling). Reflectance values of each band on every image at specific sampling points were determined.

Development of algorithms

The algorithms were then developed through a simple linear regression analysis to determine the best-fit model. The surface reflectances of each band and water quality parameter values at a specific sampling point were regressed. Reflectance values of all bands were selected to be independent of each water quality variable. Independent variables that form a relationship with reflectance by showing a high coefficient of determination (R2) were selected based on the literature (Alparslan et al. 2007; He et al. 2008; El-Saadi et al. 2014). The R2 compares estimated and actual y-values, and ranges in value from 0 to 1. A value of 1 denotes a strong correlation in the sample, meaning there is no difference between the estimated value and the actual value. At the other extreme, if the coefficient of determination is 0, the regression equation is not helpful in predicting a y-value (Alparslan et al. 2007).

Application of the algorithms and validation

The developed algorithms were applied to resampled multispectral reflectance image bands. Interpolation was performed on the six sampling locations to come up with a spatial variation map of each parameter. The inverse distance weighted (IDW) interpolation (Parida et al. 2017) technique was used for interpolating water quality data at respective point locations to grid points for comparison with satellite retrieved estimates.

Furthermore, the predicted values for each parameter were extracted from every parameter satellite raster map based on GPS location. The predicted values were validated against field measured data to test whether the satellite-based predictions were able to estimate water quality parameters. This was done by simple regression analysis using field data (for 25 January and 10 February 2015) that were randomly selected from the main dataset and not used in the development of the algorithms to reduce errors. To assess the detailed spatial distribution of water quality in Olushandja Dam, quantitative attribute maps were produced by applying algorithms to the original Landsat datasets.

RESULTS

Field measured data

The mean field measured water quality parameter data acquired from four sampling campaigns (26 Feb–15 April 2015) were calculated at each of the six points. The field data that correspond to the same dates as the satellite data (Table 1) are eventually used for model development.

Table 1

Average values of measured data for four sampling campaigns at six sampling points in Olushandja Dam

Sampling pointAlgae counts (cells/ml)Turbidity (NTU)NH3 (mg/L)TN (mg/L)TPU (mg/L)TSS (mg/L)
380 18.70 0.10 0.90 0.19 11.75 
590 9.88 0.02 0.93 0.04 4.50 
2,374 5.97 0.02 0.95 0.04 6.13 
684 19.74 0.33 1.42 0.06 22.88 
864 10.77 0.02 1.03 0.05 6.33 
3,028 54.97 0.06 3.78 0.36 27.20 
Sampling pointAlgae counts (cells/ml)Turbidity (NTU)NH3 (mg/L)TN (mg/L)TPU (mg/L)TSS (mg/L)
380 18.70 0.10 0.90 0.19 11.75 
590 9.88 0.02 0.93 0.04 4.50 
2,374 5.97 0.02 0.95 0.04 6.13 
684 19.74 0.33 1.42 0.06 22.88 
864 10.77 0.02 1.03 0.05 6.33 
3,028 54.97 0.06 3.78 0.36 27.20 

Note: NH3 = ammonia, TN = total nitrogen, TP = total phosphorus and TSS = total suspended solids.

Remote sensed data

The spectral band reflectance values at each sampling point on the selected four images corresponding to field data (26 February and 15 April 2015) are given in Table 2. From the table, low reflectance values (<0.1) on the water body can be noticed and this could be due to the fact that the water is relatively clean and this makes it have low reflective spectral radiation (He et al. 2008).

Table 2

Average spectral band reflectance for four sampling campaigns at six sampling points in Olushandja Dam

Sampling pointBand2Band3Band4Band5
0.011 0.042 0.035 0.095 
0.016 0.047 0.026 0.020 
0.059 0.057 0.057 0.057 
0.004 0.012 0.012 0.008 
0.010 0.050 0.027 0.015 
0.000 0.004 0.024 0.004 
Sampling pointBand2Band3Band4Band5
0.011 0.042 0.035 0.095 
0.016 0.047 0.026 0.020 
0.059 0.057 0.057 0.057 
0.004 0.012 0.012 0.008 
0.010 0.050 0.027 0.015 
0.000 0.004 0.024 0.004 

Developed algorithms

Field measured data (Table 1) were regressed with reflectance values (given in Table 2) using a simple linear regression method and algorithms were formulated and are given in Table 3. Looking at the R2 (coefficient of determination) and F-values (which approach 1 for good performance), the results of regression analysis can be interpreted as follows; all the derived regression models have good regressive correlation for all parameters (i.e. total algae, turbidity, TN, TP, TSS and NH3). Total algae, turbidity, TN, TP and TSS have R2 values of 0.999, 0.986, 0.987, 0.980, and 0.988 respectively. The R2 values show that there is a high accuracy in predicting all the water quality parameters for the whole dam.

Table 3

Derived retrieval algorithms for each parameter through regression analyses

Water quality variable(R2)Derived algorithms
Algae 0.999 =− 54.7081 − 26,766.6(λ2) − 42,687.5(λ3) + 137,000.1(λ4) − 23,819.7(λ5) 
Turbidity 0.986 =15.31856 − 956.806(λ2) − 747.376 (λ3) + 1,742.455 (λ4) + 165.173(λ5) 
TN 0.987 =1.047532 − 54.928 (λ2) − 46.2947(λ3) + 120.8943 (λ4) − 19.223(λ5) 
TP 0.98 =− 0.00309 − 8.78139(λ2) − 4.99958(λ3) + 15.31713(λ4) − 0.3916(λ5) 
TSS 0.988 =27.08987 + 10.80036(λ2) − 507.708(λ3) + 95.37331(λ4) + 27.8424(λ5) 
Water quality variable(R2)Derived algorithms
Algae 0.999 =− 54.7081 − 26,766.6(λ2) − 42,687.5(λ3) + 137,000.1(λ4) − 23,819.7(λ5) 
Turbidity 0.986 =15.31856 − 956.806(λ2) − 747.376 (λ3) + 1,742.455 (λ4) + 165.173(λ5) 
TN 0.987 =1.047532 − 54.928 (λ2) − 46.2947(λ3) + 120.8943 (λ4) − 19.223(λ5) 
TP 0.98 =− 0.00309 − 8.78139(λ2) − 4.99958(λ3) + 15.31713(λ4) − 0.3916(λ5) 
TSS 0.988 =27.08987 + 10.80036(λ2) − 507.708(λ3) + 95.37331(λ4) + 27.8424(λ5) 

Note: TN = total nitrogen, TP = total phosphorus and TSS = total suspended solids; λ = pixel reflectance values at bands 2–5.

The R2 values for all water quality parameters were high and close to 1, which shows that there is a significant relationship between satellite data and in-situ observed measurements.

The F-value in one-way analysis of variance (ANOVA) aided in assessing whether the variance between the means of Landsat 8 derived water quality values and field measured water quality relate. Calculated F-values (not displayed) show that all algorithms exceeded the 95% level of confidence. Li (2009) also found algorithms for turbidity in Shakespeare Bay to exceed the 90% confidence level. Furthermore, NH3 showed no significant correlation with satellite data and had low F-value (less than 4.53), thus it was excluded in the prediction.

Prediction of water quality parameter from satellite data

The study demonstrated that all four bands of Landsat 8 data contribute to the derived water quality parameters. The equations in Table 3 were applied to multispectral Landsat datasets to predict water quality parameters over the entire dam and at each sampling point. Five water quality variables in Olushandja Dam (turbidity, total suspended solids, total nitrogen, total phosphorus and total algae counts) were derived from Landsat 8 imagery on 26 February, 14 and 30 March as well as 15 April 2015 (four sampling campaigns). This was done after applying the developed algorithms to the original bands with surface reflectance and summary statistics (Table 4), and the spatial distribution (Figure 4(a)–4(e)) of the predicted five water quality parameters are presented.

Table 4

Summary statistics of predicted water qualities over the entire Olushandja Dam

StatisticsConcentrations
Algae (cells/ml)Turbidity (NTU)TN (mg/L)TP (mg/L)TSS (mg/L)
Minimum 4.7 0.00 0.00 0.0 
Maximum 21,204 356.0 13.20 2.60 42.0 
Average 1,650 83.2 1.39 0.43 8.8 
Standard deviation 3,143 63.2 2.18 0.40 16.3 
NamWater – – – – 
Canadian – 50 – – – 
UN/ECE – – < 300 < 10 – 
StatisticsConcentrations
Algae (cells/ml)Turbidity (NTU)TN (mg/L)TP (mg/L)TSS (mg/L)
Minimum 4.7 0.00 0.00 0.0 
Maximum 21,204 356.0 13.20 2.60 42.0 
Average 1,650 83.2 1.39 0.43 8.8 
Standard deviation 3,143 63.2 2.18 0.40 16.3 
NamWater – – – – 
Canadian – 50 – – – 
UN/ECE – – < 300 < 10 – 
Figure 4

Spatial variation of predicted water quality parameters.

Figure 4

Spatial variation of predicted water quality parameters.

Spatial variation of predicted water quality parameters

The resultant quantitative maps for the five water quality variables (total algae content, turbidity, and concentration of total nitrogen, total phosphorus and total suspended solids) are presented in Figure 4(a)–4(e). The resultant maps clearly show that different water quality parameters yield several spatially distributed patterns over the reservoir over time. Some parameters show very low predicted values, especially in areas with dense vegetation in the inner and along the edge of the dam closer to the land and some show the opposite. This could be due to the reflectance ability of specific bands, especially band 5 (near-infrared) that reflects more highly on vegetation and interaction of water and land than on water (USGS 2013).

The resultant maps demonstrated that total algae contents and concentrations TN and TSS are highly variable over the dam while turbidity and TP have less spatial variance. Turbidity is found to be above the recommended limits of drinking water standards by NamWater as well as for recreational purposes by Canadian guidelines (50 NTU). Other studies (He et al. 2008; Li 2009; Lehmann 2010) found high turbidity levels in different water bodies. Total nitrogen and total phosphorus values are found to be within the UNECE (1994) standards for maintaining aquatic life, which are <300 mg/L.

Validation of algorithms

The relationship between Landsat predicted values and in-situ observed measurements of water quality parameters at each of the six sampling points for 15 April 2015 are presented in Figure 5(a)–5(e). Turbidity, TN, and TP and total algae count showed medium to strong positive linear relationship between Landsat predicted and measured values while TSS showed a weak linear relationship. The regression coefficients between predicted and measured values were: turbidity (R2 = 0.767); TN (R2 = 0.798); TP (R2 = 0.907); TSS (R2 = 0.284) and total algae count (R2 = 0.851). Total suspended solids showed a lower coefficient than all other parameters. Looking at the R2 values, it can be concluded that the prediction models are best fitted to derive the four water quality parameters other than TSS.

Figure 5

(a–e): Relationship between Landsat predicted versus field measured data: (a) total algae content, (b) turbidity level, (c) total nitrogen (TN), (d) total phosphorus (TP) and (e) total suspended solids (TSS).

Figure 5

(a–e): Relationship between Landsat predicted versus field measured data: (a) total algae content, (b) turbidity level, (c) total nitrogen (TN), (d) total phosphorus (TP) and (e) total suspended solids (TSS).

Figure 5(a)–5(e) confirms the strength of the regression models by showing correlation with high coefficient of determination (R2) between predicted values and measured data. It was found that TP has higher R2 followed by total algae counts, TN and turbidity respectively. TSS had a low correlation coefficient.

Methodological framework for monitoring water quality

Based on the result of this study, Landsat data has proven its ability to derive water quality parameters, data that are comparable to the in-situ observed measurements. Therefore, a proposed framework for predicting water qualities for Olushandja Dam is presented in Figure 6.

Figure 6

A proposed framework for predicting water quality parameters based on remote sensing.

Figure 6

A proposed framework for predicting water quality parameters based on remote sensing.

The framework will help responsible institutions such as NamWater and the Department of Water Affairs under the Directorate of Resources Management in the Ministry of Agriculture, Water and Forestry and interested members of the community to implement a continuous monitoring of water quality for improved decision-making. The framework will also be useful to the Ministry of Fisheries as there are fishery resources, the Ministry of Health and Social Services, students, community members, water resources managers, planners, developers and any other interested parties. The application of this framework would require among others: personnel with some knowledge of GIS and remote sensing; and financial resources (for transportation and laboratory costs).

DISCUSSION

Summary

This research found that regression analysis-based retrieval algorithms are ideal for water quality retrieval in Olushandja Dam, north-central Namibia. Obtaining the spatial variation of water quality parameters allows decision makers to manage the critical resource in near-real-time. Remote sensing is therefore recommended for frequent and continuous monitoring of Olushandja Dam as it has the ability to provide the information on the spatio-temporal variability of surface water quality. Detailed discussion and future applications of the present work are described in the following sections.

Regression-analysis-based algorithms for water quality retrieval

In-situ-based water quality retrieval (Bhuyar et al. 2019a, 2019b) brings limitation in comprehensive water quality assessment. This study used Landsat 8, 30 m spatial resolution imagery reflectance values and in-situ observed data to develop regression-analysis-based retrieval algorithms from November 2014 to June 2015. The study found that the majority of water quality parameter regression algorithms had high correlation coefficients (R2) between 0.75 and 0.99. Therefore, the study concludes that the developed regression algorithms are the best fit to predict water quality parameters from satellite data. The study further applied the developed algorithms to derive water quality parameters at selected points in the dam. Correlation analysis was performed between Landsat 8 predicted and field measured data (Masocha et al. 2018). Alparslan et al. (2007) used Landsat ETM pixel reflectance at Ömerli Dam, which is a vital potable water resource of Istanbul City, Turkey and also found suspended solid matter and total phosphate to have high R2 of 0.9999 and 0.9906, respectively. On the other hand, Reza (2008) showed that there is a relationship between the level of suspended solids and MODIS radiance or reflectance in the seawater region around Penang Island of northern Malaysia. Lim et al. (2008) revealed that the total auspended solids (TSS) algorithm developed in the same island (Penang) from optical properties is a promising TSS model for high-accuracy TSS mapping using satellite data. Furthermore, He et al. (2008) found a fairly high correlation between Landsat TM imagery DN values with water quality variables (algae, turbidity, TN and TP) at the Guanting Reservoir in Beijing, China. All the above studies have concluded that water quality can be successfully derived through remote sensing data. Our study shows that there is generally a good correlation for all the parameters except total suspended solids, suggesting that the algorithms can be used to retrieve and predict water quality data.

Spatial variation of retrieved water quality parameters

The spatial distribution of water quality parameters retrieved from satellite data demonstrate similar patterns as the in-situ observed measurements based on a statistical comparison (R2 of >0.75 except for total suspended solids). Hafeez et al. (2019) found similar results with turbidity amongst other parameters, in a comparison of satellite retrieval of reflectance data with in situ reflectance data for a study in Hong Kong. The results of this study show that there is substantial spatial variation in the concentration of water quality parameters both longitudinally and latitudinally due to several factors. These include rainfall runoff processes (Bonansea et al. 2015), nutrient accumulation due to decomposition of plants and animals as well as low or no flow especially at the end part of the dam. In addition, it could be due to human activities such as fishing, abstraction of water and swimming, which would encourage mixing of water constituents. All these factors affect water quality.

In addition to the six sampling points, the Landsat predicted data at specific sampling points were also compared with water quality standards for the different uses. Turbidity was found to be above the NamWater (1998) guidelines of 5 NTU but below the Canadian (2012) guidelines for recreational water quality (50 NTU). On the other hand, TN and TP are below ECE (1994) standards for maintenance of aquatic life, which are <300 mg/L and <10 mg/L respectively. Water quality parameters outside recommended ranges are likely to cause complications in drinking water treatment as well as for human and aquatic life.

Suitability of medium-resolution satellite images for water quality monitoring

This study proves that medium resolution images such as those of Landsat are able to predict water quality parameters based on high regression coefficients (>0.75), as observed in studies by Concha & Schott (2014), Chen & Quan (2012), He et al. (2008) and Hellweger et al. (2004). The study also showed that Landsat 8 has capabilities in modelling the water quality of relatively clean water bodies that have low spectral radiation just like Landsat TM as recommended by Peterson et al. (2020), Pu et al. (2019) and He et al. (2008). Li (2009), however, used different Landsat missions (Landsat 5) in the Chesapeake Bay, USA, for water quality monitoring. However, the in-situ observed measurements remain critical in the process of monitoring water quality using remote sensing techniques (Ritchie et al. 2003). Namibia has cloud-free sky most times of the year, which makes remote sensed water quality monitoring possible.

Future application of the present work

The future practical application of the present work includes the improved capacity to predict water quality status for data-scarce environments such as north-central Namibia where there is a deterioration of water quality. Such efforts potentially contribute towards improved environmental and watershed management in an integrated water resources management approach. Based on the strong relationship between field-based water quality results and satellite-based water quality retrieval, our results also help in rapid water quality appraisals of other reservoirs in Namibia. In addition, the developed algorithms can be used to assess anthropogenic land-use activities in pollution of reservoirs. Multiple linear regression analyses of land-use and land-cover variable combinations can further be developed and used to develop equations for estimating water quality parameters. Example applications of such work can also be found in many global water resources studies (Christian et al. 2000; Goodman et al. 2020; Kim et al. 2020; Pasika & Gandla 2020; Topp et al. 2020). A water quality prediction framework for Olushandja Dam, north-central Namibia, will continue to use freely available earth observation data to contribute towards improved management of pollution. Future work should also focus on developing a mobile-based and a web-based application for rapid retrieval and communication of water quality information.

CONCLUSIONS AND RECOMMENDATIONS

The capacity to predict water quality status for data-scarce environments can potentially contribute towards improved water quality management. The study demonstrated that a linear regression statistical approach can be used to develop algorithms for retrieving water quality data from satellite imagery as the water quality values retrieved from remote sensing techniques had a high correlation with field measured data (R2 > 0.98). The algorithms developed have a good prediction accuracy for turbidity, total nitrogen, total phosphorus and total algae count as the linear relationship between Landsat predicted and measured values had R2 of 0.77–0.91 while prediction for total suspended solids showed a weak linear relationship (R2 = 0.28). The developed algorithms can be used to predict water quality variables for the whole Olushandja Dam but require more sampling points to further improve the accuracy of the regression models. We recommend further development of a GIS web-based application for rapid retrieval and communication of water quality information by water managers, water users and stakeholders in the Cuvelai Basin in north-central regions of Namibia.

ACKNOWLEDGEMENTS

This paper presents part of the research results of an MSc study by Taimi Kapalanga at the University of Zimbabwe (Department of Civil Engineering) under a WaterNet Fellowship under the supervision team of Eng. Zvikomborero Hoko and Mr Webster Gumindoga. The authors would like to thank NamWater and NamLab for laboratory equipment and water quality analysis. Mr Lloyd Chikwiramakomo improved the maps in this manuscript.

DATA AVAILABILITY STATEMENT

All relevant data are included in the paper or its Supplementary Information.

REFERENCES

REFERENCES
Alparslan
E.
,
Aydöner
C.
,
Tufekci
V.
&
Tüfekci
H.
2007
Water quality assessment at Ömerli Dam using remote sensing techniques
.
Environmental Monitoring and Assessment
135
,
391
398
.
Angula
M. N.
&
Kaundjua
M. B.
2016
The changing climate and human vulnerability in north-central Namibia
.
Jàmbá: Journal of Disaster Risk Studies
8
(
2
),
200. http://dx.doi.org/10.4102/jamba.v8i2.20
.
Bain
R.
,
Johnston
R.
&
Slaymaker
T.
2020
Drinking water quality and the SDGs
.
npj Clean Water
3
(
1
),
37
.
https://doi.org/10.1038/s41545-020-00085-z
.
Bauer
M.
,
Dekker
A.
,
DiGiacomo
P.
,
Greb
S.
,
Gitelson
A.
,
Herlevi
A.
&
Kuster
T.
2007
Inland and Nearshore Coastal Water Quality Remote Sensing
.
GEOSS Group on Earth Observation
,
Geneva
,
Switzerland
.
Bhuyar
P.
,
Sundararaju
S.
,
Rahim
M. H. A.
,
Ramaraj
R.
,
Maniam
G. P.
&
Govindan
N.
2019a
Microalgae cultivation using palm oil mill effluent as growth medium for lipid production with the effect of CO2 supply and light intensity
.
Biomass Conversion and Biorefinery
.
https://doi.org/10.1007/s13399-019-00548-5
.
Bhuyar
P.
,
Hong
D. D.
,
Mandia
E.
,
Rahim
M. H. A.
,
Maniam
G. P.
&
Govindan
N.
2019b
Desalination of polymer and chemical industrial wastewater by using green photosynthetic microalgae, Chlorella sp
.
Maejo International Journal of Energy and Environmental Communication
1
(
3
),
9
19
.
Bhuyar
P.
,
Hong
D. D.
,
Mandia
E.
,
Rahim
M. H. A.
,
Maniam
G. P.
&
Govindan
N.
2020
Salinity reduction from poly-chem-industrial wastewater by using microalgae (Chlorella sp.) collected from coastal region of peninsular Malaysia
.
J. Biol. Med. Open Access
1
,
105
.
Brando
V. E.
&
Dekker
A. G.
2003
Satellite hyperspectral remote sensing for estimating estuarine and coastal water quality
.
Geoscience and Remote Sensing
41
(
6
),
1378
1387
.
Chander
G.
,
Markham
B. L.
&
Helder
D. L.
2009
Summary of current radiometric calibration coefficients for Landsat MSS, TM, ETM + , and EO-1 ALI sensors
.
Remote Sensing of Environment
113
,
893
903
.
Chapra
M. U.
2015
Innovation and authenticity in Islamic finance
. In:
Keynote Forum Lecture Delivered at the Inaugural Session of the Eighth Harvard University Forum on Islamic Finance
,
19–20 April 2008, Harvard Law School, Cambridge, MA, USA
.
Chawira
M.
,
Dube
T.
&
Gumindoga
W.
2013
Remote sensing based water quality monitoring in Chivero and Manyame lakes of Zimbabwe
.
Physics and Chemistry of the Earth
66
,
38
44
.
Christian
H. J.
,
Blakeslee
R. J.
,
Goodman
S. J.
&
Mach
D. M.
2000
Algorithm Theoretical Basis Document for the Lightning Imaging Sensor
.
Marshall Space Flight Center, NASA, USA
.
Concha
J. A.
&
Schott
J. R.
2014
A model-based ELM for atmospheric correction over Case 2 water with Landsat 8
.
Proc. SPIE 9111 (Ocean Sensing and Monitoring 6)
,
911112
.
CuveWaters
2014
The Interactive Water Information and Planning Tool (IWIP) of the Cuvelai-Etosha Basin
, 2014 edn.
CuveWaters Project
,
Germany/Namibia
.
Dekker
A. G.
,
Vos
R. J.
&
Peters
S. W. M.
2002
Analytical algorithms for lake water TSM estimation for retrospective analyses of TM and SPOT sensor data
.
International Journal of Remote Sensing
23
,
15
35
.
El-Saadi
A. M.
,
Yousry
M. M.
&
Jahin
H. S.
2014
Statistical estimation of Rosetta branch water quality using multi-spectral data
.
Water Science
28
,
18
30
.
Goodman
J. A.
,
Lay
M.
,
Ramirez
L.
,
Ustin
S. L.
&
Haverkamp
P. J.
2020
Confidence levels, sensitivity, and the role of bathymetry in coral reef remote sensing
.
Remote Sensing
12
,
3
,
496
.
https://doi.org/10.3390/rs12030496
.
Hafeez
S.
,
Wong
M. S.
,
Ho
H. C.
,
Nazeer
M.
,
Nichol
J.
,
Abbas
S.
,
Tang
D.
,
Lee
K. H.
&
Pun
L.
2019
Comparison of machine learning algorithms for retrieval of water quality indicators in case-II waters: a case study of Hong Kong
.
Remote Sensing
11
(
6
),
617
.
Hellweger
F. L.
,
Schlosser
P.
,
Lall
U.
&
Weissel
J. K.
2004
Use of satellite imagery for water quality studies in New York Harbor
.
Estuarine, Coastal and Shelf Science
61
,
437
448
.
Irenosen
O. G.
,
Festus
A. A.
&
Coolborn
A. F.
2012
Water quality assessment of the Owena Multi-Purpose Dam, Ondo State, Southwestern Nigeria
.
Journal of Environmental Protection
3
,
14
25
.
IWRMPJVN
2010
Integrated Water Resources Management Plan for Namibia. IWRM Plan Joint Venture Namibia, Ministry of Agriculture, Water and Forestry
,
Windhoek
,
Namibia
.
Kallio
K.
2000
Remote sensing as a tool for monitoring lake water quality
. In:
Hydrological and Limnological Aspects of Lake Monitoring
(P. Heinonen, G. Ziglio & A. Van der Beken, eds)
,
John Wiley & Sons
,
New York, NY
,
USA
, pp.
237
245
.
Kelbert
T.
2016
Encounters at the Water Point – An Ethnography of the Travelling Model of Community-Based Water Management and Its Application to Rural Water Supply in Namibia
.
PhD thesis, University of Cologne, Cologne, Germany
.
Kim
Y. H.
,
Son
S.
,
Kim
H.-C.
,
Kim
B.
,
Park
Y.-G.
,
Nam
J.
&
Ryu
J.
2020
Application of satellite remote sensing in monitoring dissolved oxygen variabilities: a case study for coastal waters in Korea
.
Environment International
134
,
105301
.
https://doi.org/https://doi.org/10.1016/j.envint.2019.105301
.
Kolberg
H.
2002
Preliminary Inventory of Namibia's Wetlands
.
Directorate of Scientific Services
,
Ministry of Environment and Tourism
,
Windhoek
,
Namibia
.
Available from: http://www.drfn. info/arc/pdf/reprints/drfn2979.pdf (accessed 29 August 2014)
.
Lehmann
F. L.
2010
Water Quality of the Swakoppoort Dam, Namibia. Master's thesis, IHF Hydrologie Freiburg
,
University of Freiburg, Freiburg im Breisgau, Germany
.
Li
D.
2009
A Novel Remote Sensing-Based Water Quality Monitoring System for the Chesapeake Bay Using Landsat 5 Thematic Mapper
.
Thomas Jefferson High School for Science and Technology, Alexandria, VA, USA
.
Lim
H. S.
,
MatJafri
M. Z.
,
Abdullah
K.
,
Alias
A. N.
&
Saleh
N. M.
2008
Water quality mapping using remote sensing technique in Penang Straits, Malaysia
.
Proc. SPIE
7105
,
71050A
.
https://doi.org/10.1117/12.800573
.
McMillan
H.
,
Seibert
J.
,
Petersen-Overleir
A.
,
Lang
M.
,
White
P.
,
Snelder
T.
,
Rutherford
K.
,
Krueger
T.
,
Mason
R.
&
Kiang
J.
2017
How uncertainty analysis of streamflow data can reduce costs and promote robust decisions in water management applications.
Water Resources Research
53
(
7
),
5220
5228
.
Mendelsohn
J.
,
El Obeid
S.
&
Roberts
C.
2000
A Profile of North-Central Namibia. Environmental Profiles Project
,
Gamsberg Macmillan Publishers
,
Windhoek
,
Namibia
.
Mendelsohn
J.
,
Jarvis
A.
&
Robertson
T.
2013
A Profile and Atlas of the Cuvelai-Etosha Basin
, 1st edn.
RAISON and Gondwana Collection
,
Windhoek
,
Namibia
.
Mendelsohn
J.
,
Jarvis
A.
&
Robertson
T.
2016
A Profile and Atlas of the Cuvelai-Etosha Basin
.
John Meinert Printing
,
Windhoek, Namibia
.
NAMWATER
1998
Guidelines for Evaluating Drinking-Water for Human Consumption with Respect to the Chemical, Physical and Bacteriological Qualities
.
Namibia Water Corporation
,
Windhoek, Namibia
.
NamWater
2012
Management of the Upper Swakop Basin: Key Water Security to the Central Area of Namibia
(
Pazvakawambwa
G. T.
,
Tjipangandjara
K. F.
&
Chulu
E.
, eds),
NamWater
,
Windhoek
,
Namibia
.
NamWater
2013
Infrastructures – Olushandja Dam, Namibia Water Corporation Ltd, Windhoek, Namibia
.
Oberholster
P. J.
&
Ashton
P. J.
2008
State of the Nation Report: An Overview of the Current Status of Water Quality and Eutrophication in South African Rivers and Reservoirs
Parliamentary Grant Deliverable
,
Council for Scientific and Industrial Research (CSIR), Pretoria
,
South Africa
.
Olet
I.
2010
Water Quality Monitoring of Roxo Reservoir Using Landsat Images and In-Situ Measurements. MS thesis
,
International Institute for Geo-Information Science and Earth Observation
,
Enschede
,
The Netherlands
.
Parida
S.
,
Patsha
A.
,
Bera
S.
&
Dhara
S.
2017
Spectroscopic investigation of native defect induced electron-phonon coupling in GaN nanowires.
Journal of Physics D: Applied Physics
50
,
275103
.
Pasika
S.
&
Gandla
S. T.
2020
Smart water quality monitoring system with cost-effective using IoT
.
Heliyon
6
(
7
),
e04096
.
https://doi.org/10.1016/j.heliyon.2020.e04096
.
Pu
F.
,
Ding
C.
,
Chao
Z.
,
Yu
Y.
&
Xu
X.
2019
Water-quality classification of inland lakes using Landsat8 images by convolutional neural networks
. Remote Sensing
11
(
14
),
1674
Ritchie
J. C.
,
Zimba
P. V.
&
Everitt
J. H.
2003
Remote sensing techniques to assess water quality
.
Photogrammetric Engineering & Remote Sensing
69
(
6
),
695
704
.
Salama
M. S.
,
Dekker
A.
,
Su
Z.
,
Mannaerts
C. M.
&
Verhoef
W.
2009
Deriving inherent optical properties and associated inversion-uncertainties in the Dutch Lakes
.
Hydrology and Earth System Sciences
13
(
7
),
1113
1121
.
Schaeffer
B. A.
,
Schaeffer
K. G.
,
Keith
D.
,
Lunetta
R. S.
,
Conmy
R.
&
Gould
R. W.
2013
Barriers to adopting satellite remote sensing for water quality management
.
International Journal of Remote Sensing
34
(
21
),
7534
7544
.
Schalles
J. F.
,
Gitelson
A. A.
,
Yacobi
Y. Z.
&
Kroenke
A. E.
1998
Estimation of chlorophyll a from time series measurements of high spectral resolution reflectance in an eutrophic lake
.
Journal of Phycology
34
,
383
390
.
SDP10
2001
Water use and Environmental Conditions along the Etaka Canal in Northern Central Namibia. Summer Desertification Programme
,
Desert Research Foundation of Namibia
,
Windhoek, Namibia
.
Shuuya
M. K.
&
Hoko
Z.
2014
Trends and impacts of pollution in the Calueque–Oshakati Canal in north-Central Namibia on water treatment
. In:
Combating Water Scarcity in Southern Africa
(
Msangi
J. P.
, ed.),
Springer, Dordrecht
,
The Netherlands
, pp.
43
60
.
Somvanshi
S.
,
Kunwar
P.
,
Singh
N. B.
,
Shukla
S. P.
&
Pathak
V.
2012
Integrated remote sensing and GIS approach for water quality analysis of Gomti river, Uttar Pradesh
.
International Journal of Environmental Sciences
3
(
1
),
62
74
.
Tomppo
E.
,
Czaplewski
R. L.
&
Mäkisara
K.
2002
The Role of Remote Sensing in Global Forest Assessment: A Remote Sensing Background Paper for Kotka IV Expert Consultation
.
Food and Agriculture Organization
,
Rome, Italy
.
Topp
S. N.
,
Pavelsky
T. M.
,
Jensen
D.
,
Simard
M.
&
Ross
M. R. V.
2020
Research trends in the use of remote sensing for inland water quality science: moving towards multidisciplinary applications
.
Water
12
(
1
),
169
.
https://doi.org/10.3390/w12010169.
UN
2011
Namibia Floods: Situation Report #4
.
United Nations, Office of the Resident Coordinator
,
Windhoek
,
Namibia
.
UNECE
1994
ECE Standard Statistical Classification of Surface Freshwater Quality for the Maintenance of Aquatic Life
.
United Nations Economic Commission for Europe (UNECE), Geneva
,
Switzerland
.
UN-Water
2011
Policy Brief on Water Quality
.
UN-Water, New York, USA. Available from: http://www.unwater.org/downloads/waterquality_policybrief (accessed 12 May 2015)
.
USGS
2013
Using the USGS Landsat 8 Product
.
US Geological Survey
,
USA
.
Vignolo
A.
,
Pochettino
A.
&
Cicerone
D.
2006
Water quality assessment using remote sensing techniques: Medrano Creek, Argentina
.
Journal of Environmental Management
81
(
4
),
429
433
.
Wait
A. D.
,
Tuit
C. B.
&
Maney
J. P.
2020
Forensic sampling practices for oil spills in the marine environment
.
Environmental Forensics
21
,
310
318
.
Wang
X. J.
&
Ma
T.
2001
Application of remote sensing techniques in monitoring and assessing the water quality of Taihu Lake
.
Bulletin of Environmental Contamination and Toxicology
67
(
6
),
863
870
.
Wang
Y.
,
Xia
H.
,
Fu
J.
&
Sheng
G.
2004
Water quality change in reservoirs of Shenzhen, China: detection using LANDSAT/TM data
.
Science of the Total Environment
328
(
1–3
),
195
206
.
Waxter
M. T.
2014
Analysis of Landsat Satellite Data to Monitor Water Quality Parameters in Tenmile Lake, Oregon, Civil and Environmental Engineering. Master's Project Reports 2
.
Portland State University
,
OR, USA
.
Zuccari Fernandes Braga
C.
,
Setzer
A. W.
&
Drude de Lacerda
L.
1993
Water quality assessment with simultaneous Landsat-5 TM data at Guanabara Bay, Rio de Janeiro, Brazil
.
Remote Sensing of Environment
45
(
1
),
95
106
.