- Email: [email protected]
Construction of a novel water quality index and quality indicator for reservoir water quality evaluation: A case study in the Amazon region
Journal of Hydrology 522 (2015) 674–683
Contents lists available at ScienceDirect
Journal of Hydrology journal homepage: www.elsevier.com/locate/jhydrol
Construction of a novel water quality index and quality indicator for reservoir water quality evaluation: A case study in the Amazon region T.C. Lobato a,⇑, R.A. Hauser-Davis b,⇑, T.F. Oliveira a, A.M. Silveira a, H.A.N. Silva a, M.R.M. Tavares a, A.C.F. Saraiva c a Universidade Federal do Pará (UFPA), Instituto de Ciências Exatas e Naturais, Faculdade de Estatística and Faculdade de Ciência da Computação, Rua Augusto Correa, 01, CEP: 66075-110 Belém, PA, Brazil b Pontifícia Universidade Católica do Rio de Janeiro (PUC-Rio), Departamento de Química, Laboratório de Bioanalítica, Rua Marquês de São Vicente, 225, Gávea, CEP: 22453-900 Rio de Janeiro, RJ, Brazil c Centro de Tecnologia da ELETRONORTE, Rodovia Arthur Bernardes, S/N – Miramar, Telegrafo sem Fio, CEP: 66.115-000 Belem, PA, Brazil
a r t i c l e
i n f o
Article history: Received 21 July 2014 Received in revised form 27 October 2014 Accepted 7 January 2015 Available online 22 January 2015 This manuscript was handled by Andras Bardossy, Editor-in-Chief, with the assistance of Sheng Yue, Associate Editor Keywords: Hydrological cycle Statistical analyses Water quality indicator Quality index Amazon area
s u m m a r y A novel Quality Indicator (QI) and Water Quality Index (WQI) were constructed in the present study for the evaluation of the water quality of a Hydroelectric Plant reservoir in the Amazon area, Brazil, taking into account the specific characteristics of the Amazon area. Factor analyses were applied in order to select the relevant parameters to be included in the construction of both indices. Quality curves for each selected parameter were then created and the constructed QI and WQI were then applied to investigate the water quality at the reservoir. The hydrological cycle was shown by the indices to directly affect reservoir water quality, and the WQI was further useful in identifying anthropogenic impacts in the area, since water sampling stations suffering different anthropogenic impacts were categorized differently, with poorer water quality, than stations near the dam and the environmental preservation area, which suffer significantly less anthropogenic impacts, and were categorized as presenting better water quality. The constructed indices are thus helpful in investigating environmental conditions in areas that show well-defined hydrological cycles, in addition to being valuable tools in the detection of anthropogenic impacts. The statistical techniques applied in the construction of these indices may also be used to construct other indices in different geographical areas, taking into account the specificities for each area. Ó 2015 Elsevier B.V. All rights reserved.
1. Introduction Water quality is largely determined by natural processes, such as weathering and soil erosion, but also by anthropogenic inputs (Kazi et al., 2009; Singh et al., 2004). To assess human impacts on water quality, variations in space and time and in the biological, physical and chemical processes of natural systems should be considered. One important factor in this context is the hydrological cycle, which directly affects the drainage network of water bodies and may causes surface runoff along riverbanks. This, in turn, may lead to contamination by sediments and several types of pollutants, besides directly affecting the local vegetation and biota (Aubert et al., 2013; Lai et al., 2013; Merten and Minella, 2002). By evaluating these hydrological variations and their effects, it is possible to create management actions that can intervene in the recovery and/or preservation of ecosystems (Abaurrea et al., 2011; Braga et al., 2006; Lucas ⇑ Corresponding authors. E-mail addresses: [email protected] (T.C. Lobato), rachel.hauser. [email protected] (R.A. Hauser-Davis). http://dx.doi.org/10.1016/j.jhydrol.2015.01.021 0022-1694/Ó 2015 Elsevier B.V. All rights reserved.
et al., 2010; Shah et al., 2007; Tundisi and Tundisi, 2008). Water quality assessments also allow for decision-making policies to be made regarding water potability and use. Several water quality assessment studies have been conducted by applying statistical techniques, such as the principal component analysis, which can aid in identifying natural or anthropogenic factors that can cause alterations in water quality (Andrade et al., 2007; Boyacioglu, 2006; Brito et al., 2006; El-Iskandarani et al., 2004; Guedes et al., 2012; Laureano and Navar, 2002; Petersen et al., 2001; Selle et al., 2013; Vialle et al., 2011; Vonberg et al., 2014). Some studies have applied this statistical technique to specifically propose a Water Quality Index (WQI), by using the weighted scores of each analyzed water quality parameter (Haase et al., 2003; Toledo and Nicolella, 2002). WQIs usually take into account general water parameters, such as dissolved oxygen, pH, temperature, turbidity, and NH3 concentrations, among others. UNESCO defines the Amazon region and adjacent areas like the Pantanal as world interest areas due to their unique flora and fauna and great biodiversity. These regions have, however, been increasingly under anthropogenic pressure (Sucksdorff, 1984). One of the
T.C. Lobato et al. / Journal of Hydrology 522 (2015) 674–683
main impacts in these areas has been the construction of Hydroelectric Power Plants reservoirs, which have become the predominant lake type in many regions throughout the world (Lewis, 2000). These tropical areas go though well-defined defined hydrological cycles each year, and this feature has, increasingly, been studied to obtain environmental information which may help in the monitoring of environmental impacts caused by anthropogenic pressures and may lead to decision-making in this regard. However, to the best of our knowledge, no WQI data in the literature has been constructed taking into account the water transparency or hydrological cycle of the Amazon area. In this context, the aims of the present study were to create a new Water Quality Index (WQI) by applying multivariate statistical techniques, taking into account the water transparency and specific hydrological cycle of the Amazon, besides other selected parameters. Another objective was the creation of a Quality Indicator (QI), by constructing water quality index curves, and the subsequent application of the constructed QI to the investigation of the water quality of a Hydroelectric Power Plant reservoir in the Amazon region, in order to verify the impact of this construction on the reservoir’s water quality throughout the well-defined hydrological cycles of the area. 2. Material and methods 2.1. Study area and sampling stations The Tucuruí Hydroelectric Plant is located in the state of Pará, Brazil, at latitude 03°430 –05°150 , longitude 49°120 –50°000 W
675
(Fig. 1). The plant was constructed in September 1984 on the river Tocantins, at approximately 7 km from the town of Tucuruí and 300 km from the city of Belém, the state capital. It is the first large-scale (25 units) hydroelectric project in the Brazilian Amazon rainforest, with an installed capacity of 8370 MW. The main purpose of the dam is hydroelectric power production to the Brazilian states of Maranhão and Pará and navigation between the upper and lower Tocantins river. The reservoir has a total flooded area of approximately 2850 m2, with approximately 50.8 million m3 of water and water residence time of 46 days (WCD, 2000). The area is characterized by rainy (December to May) and dry (June to November) seasons, with annual precipitations between 2250 and 2500 mm. The rainiest month is March and the driest is September (Fisch et al., 1990). The local hydrological cycle in the reservoir is reflected in the local water levels, and, consequently, the reservoir goes through a cycle constituted by a full stage, during March, April and May, an emptying stage, during June, July and August, an empty stage during September, October and November, and a filling stage during December, January and February. Eleven water sampling stations (C1, C2, M1, M3, MR, MBB, MBL, MP, MIP, ML and MJV) are present at the Tucuruí Hydroelectric Plant reservoir also displayed in Fig. 1. All are located upstream and represent the specific characteristics of each area within the reservoir. The large influx of people to this area has also led to deforestation and negative impacts from increased cattle-raising activities. Population increases have also strained the existing infrastructure, or lack thereof (Rebouças et al., 2002; WCD, 2000), leading to concerns regarding anthropogenic influences in the area.
Fig. 1. Location of the Tucuruí Hydroelectric Power Plant reservoir, indicating the water sampling stations.
676
T.C. Lobato et al. / Journal of Hydrology 522 (2015) 674–683
2.2. Physico-chemical water characteristics A total of 180 water samples were collected at the eleven sampling stations, from 2009 to 2012, as displayed in Table 1. The samples were collected from approximately 5 cm below the water surface, using 2.5 L capacity Van Dorn bottles. The samples were stored in 50 mL polypropylene tubes and acidified with nitric acid 1 mol L1, according to USA EPA Standard Method 1060B. The following physico-chemical parameters were determined: chloride (Cl), temperature (T), transparency (S), electrical conductivity (EC), pH, Total Suspended Solids (TSS), ammonia (NH4), nitrate (NO3), chlorophyll a (Cl a), turbidity (Turb.), dissolved oxygen (DO) (Winkler method), phosphates (PO4) and total phosphorous (TP), according to the Standard Methods norms (StandardMethods, 2013). The pH, EC and T parameters were determined in situ. Metal determinations (Ca, Fe, K, Mg, Na) were also conducted, using optical emission spectrometry by inductive coupling plasma (ICP–OES – Varian), a viable, multielemental, simple and quick technique for monitoring macro and micronutrients. The standard reference material used for recovery tests was SLRS4, River Water Reference Material for Trace Metals (National research Council, Canada). 2.3. Construction of the Quality Indicator (QI) The QI was created based on a factor analysis that selected factors that better explain the water quality of the reservoir, by the application of a principal component analysis without the use of a Varimax rotation (Table 2). The Bartlett method was applied to the scores of this factor analysis, to verify if commonalities were above or below 0.50 (Hair et al., 1998; Toledo and Nicolella, 2002). If parameters displayed values below 0.50, the respective variable was excluded from the subsequent analysis. The variables pH, DO, Ca, K, Mg, Na and Cl all showed commonalities below 0.50, not correlating amongst themselves or the other investigated variables. In addition, the values of these parameters were also compared to CONAMA class 2 maximum permitted values for Brazilian waters, and none were higher than these permitted values, in accordance with this legislation. Thus, because these parameters showed very low commonalities, not being responsible for explaining water quality in the study area, and not being worrisome regarding water quality according to the CONAMA legislation, they were excluded from further analyses. The correlation matrix for the analyzed parameters is displayed in Table 3. After this exclusion step, the results of the factor analysis indicated four factors with a total variance of 74.54%. The first factor extracted from the factor analysis explained 41.62% of the data variability, indicating that it is the main factor to be taken into
Table 1 Water sampling frequency per study year at the eleven water sampling stations located at the Tucuruí Hydroelectric Plant reservoir, Brazil. Sampling station
2009
2010
2011
2012
Total
C1 C2 M1 M3 MBB MBL MIP MJV ML MP MR Total
4 6 4 4 3 6 3 1 2 5 5 43
3 1 3 4 4 3 3 5 6 2 6 40
3 6 3 3 3 4 4 6 5 5 5 47
4 3 12 4 4 3 4 4 4 4 4 50
14 16 22 15 14 16 14 16 17 16 20 180
Table 2 Factor analysis results to obtain the relevant parameters to take part in the construction of the Quality Indicator (QI). Component Initial Eigenvalues
1 2 3 4 5 6 7 8 9 10 11
Extraction sums of squared loadings
Total % of variance
Cumulative Total % of % variance
Cumulative %
4.578 41.622 1.436 13.055 1.178 10.709 1.007 9.157 0.756 6.869 0.562 5.105 0.436 3.96 0.399 3.626 0.285 2.59 0.195 1.771 0.169 1.536
41.622 54.677 65.387 74.544 81.413 86.518 90.478 94.104 96.694 98.464 100.00
41.622 54.677 65.387 74.544
4.578 41.622 1.436 13.055 1.178 10.709 1.007 9.157
account, composed by the following parameters: upstream level, EC, S, PO4, TP, TSS, NH4, NO3, T, Turb. and Fe. The Bartlett method was then applied to the scores of this first factor, to verify commonalities above 0.50 (Table 4) (Hair et al., 1998; Toledo and Nicolella, 2002). All scores were above 0.50, and thus, all parameters were taken into account, and subsequently used to construct the QI, displayed in Eq. (1).
QI ¼ 0:19 TP þ 0:19 TSS þ 0:18 Turb: þ 0:18 Fe 0:18 S þ 0:15 PO4 0:09 T þ 0:08 EC þ 0:08 NO3 þ 0:08 NH4 0:05 Upstream water Level
ð1Þ
2.4. Construction of the Water Quality Index (WQI) The WQI created in the present study was based on the index developed by the National Sanitation Foundation (NSF) (Brown et al., 1970). The NSF WQI considers nine weighted parameters to characterize water quality: Dissolved Oxygen (DO), fecal coliforms, pH, biochemical oxygen demand (BOD), total nitrogen (N), total phosphorous (TP), Temperature (T), Turbidity (Turb.) and Total Suspended Solids (TSS). As the WQI created in the present study was adapted to a tropical reservoir, the parameters included in its construction were different, in order to reflect the specific geographical characteristics of the area. With this in mind, a factor analysis without rotation of all the parameters determined in the present study was conducted, and the following parameters were chosen to take part in the WQI construction: EC, S, PO4, TP, TSS, NH4, NO3, Turb., Fe, upstream water levels and T. The factor analysis results are displayed in Table 5. The Bartlett method was then applied to the scores of the four factors, to verify commonalities above 0.50 (Table 6) (Hair et al., 1998; Toledo and Nicolella, 2002). All scores were above 0.50, and thus, all parameters were taken into account, and subsequently used to construct the WQI, displayed in Eq. (2).
WQI ¼
n Y qwi i
ð2Þ
i¼1
where WQI – water quality index, in the range of 0–100; qi – quality of the i-th parameter; n – the size of the parameters selected to compose the index; wi – weight corresponding to the i-th parameter. The results of the factor analysis were used to calculate the quality curves (qi) of the relevant parameters obtained by the application of the QI, for subsequent WQI application. The other qi curves were obtained from correlated linear and non-linear studies. When available, the quality curves from the NSF were applied,
Table 3 Correlation matrix for the analyzed parameters. S
T
DO
pH
EC
Cl
Fe
Ca
Mg
Na
K
NH4
NO3
PO4
TP
TSS
0.186 0.012 180
0.018 0.807 180
0.054 0.469 180
0.261 0 180
0.336 0 180
0.015 0.86 136
0.007 0.93 180
0.023 0.754 180
0.158 0.034 180
0.052 0.49 180
0.157 0.036 180
0.33 0 180
0.172 0.021 180
0.079 0.296 179
0.122 0.102 180
0.261 0 180
0.177 0.017 180
0.111 0.139 180
0.208 0.005 180
0.287 0 180
0.08 0.352 136
0.619 0 180
0.059 0.433 180
0.102 0.173 180
0.09 0.231 180
0.321 0 180
0.229 0.002 180
0.186 0.013 180
0.531 0 179
0.606 0 180
0.642 0 180
0.025 0.736 180
0.005 0.949 180
0.179 0.016 180
0.2 0.019 136
0.219 0.003 180
0.145 0.052 180
0.026 0.727 180
0.044 0.559 180
0.262 0 180
0.02 0.788 180
0.296 0 180
0.286 0 179
0.328 0 180
0.253 0.001 180
1
0.189 0.011 180
0.12 0.109 180
0.089 0.302 136
0.044 0.56 180
0.287 0 180
0.091 0.226 180
0.054 0.468 180
0.179 0.016 180
0.068 0.363 180
0.015 0.837 180
0.018 0.808 179
0.009 0.908 180
0.132 0.076 180
0.141 0.059 180
0.085 0.323 136
0.091 0.225 180
0.114 0.128 180
0.163 0.029 180
0.126 0.091 180
0.088 0.241 180
0.106 0.159 180
0.127 0.089 180
0.062 0.411 179
0.01 0.89 180
0.124 0.099 180
0.145 0.091 136
0.274 0 180
0.355 0 180
0.014 0.856 180
0.072 0.337 180
0.224 0.003 180
0.055 0.467 180
0.049 0.514 180
0.273 0 179
0.256 0.001 180
0.159 0.033 180
1
0.053 0.541 136
0.042 0.628 136
0.003 0.975 136
0.351 0 136
0.404 0 136
0.045 0.6 136
0.033 0.707 136
0.05 0.568 135
0.167 0.052 136
0.083 0.339 136
0.149 0.045 180
0.025 0.743 180
0.225 0.002 180
0.298 0 180
0.26 0 180
0.226 0.002 180
0.568 0 179
0.754 0 180
0.621 0 180
0.103 0.17 180
0.04 0.59 180
0.331 0 180
0.051 0.494 180
0.052 0.49 180
0.213 0.004 179
0.101 0.179 180
0.015 0.84 180
0.084 0.264 180
0.109 0.145 180
0.387 0 180
0.054 0.469 180
0.017 0.817 179
0.059 0.434 180
0.09 0.229 180
0.457 0 180
0.08 0.284 180
0.053 0.482 180
0.036 0.629 179
0.193 0.009 180
0.03 0.693 180
0.16 0.032 180
0.026 0.733 180
0.202 0.007 179
0.358 0 180
0.258 0 180
1
0.291 0 180
0.053 0.481 179
0.26 0 180
0.253 0.001 180
0.111 0.141 179
0.204 0.006 180
0.272 0 180
0.618 0 179
0.467 0 179
1
0.725 0
Upstream water level
Pearson Correlation Sig. (2-tailed) N
1
S
Pearson Correlation Sig. (2-tailed) N
0.186 0.012 180
1
T
Pearson Correlation Sig. (2-tailed) N
0.018 0.807 180
0.177 0.017 180
1
DO
Pearson Correlation Sig. (2-tailed) N
0.054 0.469 180
0.111 0.139 180
0.025 0.736 180
180
pH
Pearson Correlation Sig. (2-tailed) N
0.261 0 180
0.208 0.005 180
0.005 0.949 180
0.189 0.011 180
1
EC
Pearson Correlation Sig. (2-tailed) N
0.336 0 180
0.287 0 180
0.179 0.016 180
0.12 0.109 180
0.141 0.059 180
1
Cl
Pearson Correlation Sig. (2-tailed) N
0.015 0.86 136
0.08 0.352 136
0.2 0.019 136
0.089 0.302 136
0.085 0.323 136
0.145 0.091 136
136
Fe
Pearson Correlation Sig. (2-tailed) N
0.007 0.93 180
0.619 0 180
0.219 0.003 180
0.044 0.56 180
0.091 0.225 180
0.274 0 180
0.053 0.541 136
1
Ca
Pearson Correlation Sig. (2-tailed) N
0.023 0.754 180
0.059 0.433 180
0.145 0.052 180
0.287 0 180
0.114 0.128 180
0.355 0 180
0.042 0.628 136
0.149 0.045 180
1
Mg
Pearson Correlation Sig. (2-tailed) N
0.158 0.034 180
0.102 0.173 180
0.026 0.727 180
0.091 0.226 180
0.163 0.029 180
0.014 0.856 180
0.003 0.975 136
0.025 0.743 180
0.103 0.17 180
1
Na
Pearson Correlation Sig. (2-tailed) N
0.052 0.49 180
0.09 0.231 180
0.044 0.559 180
0.054 0.468 180
0.126 0.091 180
0.072 0.337 180
0.351 0 136
0.225 0.002 180
0.04 0.59 180
0.084 0.264 180
1
K
Pearson Correlation Sig. (2-tailed) N
0.157 0.036 180
0.321 0 180
0.262 0 180
0.179 0.016 180
0.088 0.241 180
0.224 0.003 180
0.404 0 136
0.298 0 180
0.331 0 180
0.109 0.145 180
0.457 0 180
1
NH4
Pearson Correlation Sig. (2-tailed) N
0.33 0 180
0.229 0.002 180
0.02 0.788 180
0.068 0.363 180
0.106 0.159 180
0.055 0.467 180
0.045 0.6 136
0.26 0 180
0.051 0.494 180
0.387 0 180
0.08 0.284 180
0.16 0.032 180
180
NO3
Pearson Correlation Sig. (2-tailed) N
0.172 0.021 180
0.186 0.013 180
0.296 0 180
0.015 0.837 180
0.127 0.089 180
0.049 0.514 180
0.033 0.707 136
0.226 0.002 180
0.052 0.49 180
0.054 0.469 180
0.053 0.482 180
0.026 0.733 180
0.291 0 180
1
PO4
Pearson Correlation Sig. (2-tailed) N
0.079 0.296 179
0.531 0 179
0.286 0 179
0.018 0.808 179
0.062 0.411 179
0.273 0 179
0.05 0.568 135
0.568 0 179
0.213 0.004 179
0.017 0.817 179
0.036 0.629 179
0.202 0.007 179
0.053 0.481 179
0.111 0.141 179
1
TP
Pearson Correlation Sig. (2-tailed)
0.122 0.102
0.606 0
0.328 0
0.009 0.908
0.01 0.89
0.256 0.001
0.167 0.052
0.754 0
0.101 0.179
0.059 0.434
0.193 0.009
0.358 0
0.26 0
0.204 0.006
0.618 0
180
180
180
180
180
180
180
180
180
180
180
179
677
(continued on next page)
T.C. Lobato et al. / Journal of Hydrology 522 (2015) 674–683
Upstream water level
0.778 0 180
180
1
180
T.C. Lobato et al. / Journal of Hydrology 522 (2015) 674–683
TSS
678
Table 4 Results of the Bartlett method to select significant commonalities for the analyzed parameters. Values in bold indicate the factor scores used.
0.642 0 180 0.251 0.001 180 0.17 0.022 180 0.095 0.205 180 Pearson Correlation Sig. (2-tailed) N Turb.
0.187 0.012 180
0.355 0 180
ð4Þ
0.666 0 180
0.121 0.105 180
0.081 0.348 136
0.541 0 180
0.035 0.639 180
0.033 0.663 180
0.025 0.74 180
0.3 0 180
0.141 0.06 180
0.536 0 179
180
0.725 0 180 0.467 0 179 0.272 0 180 0.253 0.001 180 0.258 0 180
179 180 180 180
0.03 0.693 180 0.09 0.229 180
180 180
0.015 0.84 180 0.621 0 180
180 136
0.159 0.033 180
0.115 0.176 0.74 0.449 0.175 0.26 0.025 0.098 0.164 0.017 0.252
1 4; ði ¼ NO3 Þ pi
0.132 0.076 180
180
4
0.276 0.114 0.258 0.593 0.007 0.162 0.082 0.013 0.053 0.057 0.549
ð3Þ
180
0.124 0.099 180
3
0.551 0.014 0.103 0.106 0.106 0.453 0.275 0.08 0.037 0.054 0.262
1 ði ¼ PO4 ; Total P; Turbidity; TFeÞ pi
180 180
0.642 0 180 0.261 0 180
180 N
2
0.053 0.175 0.091 0.08 0.177 0.077 0.154 0.19 0.186 0.181 0.078
qi ¼ 5
qi ¼ 3621:5pi ; ði ¼ NH4 Þ
Pearson Correlation Sig. (2-tailed) N
Component 1
using the limits established by Brazilian legislation CONAMA for class 2 freshwater bodies for each selected parameter. All calculations were made using the STATISTICAÒ software package. To develop the equations used in the construction of the WQI, pi, wi and qi values were calculated: The pi represents the value or concentration relative to the i-th parameter. The pi values of the selected parameters (NO3, NH4, PO4, TP and Turb.) were transformed on a scale from 0.05 to 1.00, where 0.05 was the value attributed to the minimum values and 1.00 to the maximum values obtained for each of these parameters in the collected water samples. For Fe, pi values were processed in the range between 0.05 and 1.60, applying dissolved iron values from the CONAMA legislation in Eq. (3). The qi is the quality of the i-th parameter, a number between 0 and 100, obtained from the respective mean curve of variations in water quality, depending on a concentration or measurement of the respective parameter. The qi values for each parameter were calculated according to Eqs. 3–5 for NH4, followed by the application of an inverse function. The final qi values were very similar to the desired curve representative of water quality (q), and were then multiplied by 5 to obtain values closer to the reality of the study area.
0.253 0.001 180
180
Upstream water level S T EC Fe NH4 PO4 TP TSS Turb NO3
qi ¼ 5
TSS
Table 3 (continued)
0.083 0.339 136
180
NO3 EC Upstream water level
S
T
DO
pH
Cl
Fe
Ca
Mg
Na
K
NH4
PO4
TP
Parameter
ð5Þ
No qi curves for water transparency are available in the literature for the Amazon area. Thus, the information given by qi, which is negative, meaning that better water quality is observed with higher transparency values, and Carlson’s Trophic state index (Carlson, 1977), which takes into account this parameter, were taken into account to create this qi curve. When water transparency increases, the qi increases exponentially according to the interval in which the transparency values are inserted, which is why we calculated qi = eS and multiplied the result by a factor that controls qi growth (Table 7). So, for S 6 0.5 m, qi = 5, indicating lower water quality, and for S > 3.4 m, qi = 100, indicating higher water quality. The upstream level qi was obtained by interpreting the QI. Higher values indicate better water quality than lower values. Water level values were normalized in the range of 5–90 to better interpret the results. The EC qi considered the maximum value stipulated by Brazilian CETESB norms (CETESB, 2009) (100 lS cm1), and the pi. Values
679
T.C. Lobato et al. / Journal of Hydrology 522 (2015) 674–683 Table 5 Factor analysis results to obtain the relevant parameters to take part in the construction of the Water Quality Index (WQI). Component
Initial Eigenvalues
1 2 3 4 5 6 7 8 9 10 11
Extraction sums of squared loadings
Total
% of variance
Cumulative %
Total
% of variance
Cumulative %
Total
% of variance
Cumulative %
4.578 1.436 1.178 1.007 0.756 0.562 0.436 0.399 0.285 0.195 0.169
41.622 13.055 10.709 9.157 6.869 5.105 3.96 3.626 2.59 1.771 1.536
41.622 54.677 65.387 74.544 81.413 86.518 90.478 94.104 96.694 98.464 100
4.578 1.436 1.178 1.007
41.622 13.055 10.709 9.157
41.622 54.677 65.387 74.544
4.1 1.548 1.295 1.257
37.27 14.07 11.774 11.43
37.27 51.34 63.114 74.544
Table 6 Commonalities, autovalues (F), percentage of factor variance and percentage of cumulative variance for the parameters selected to take part of the WQI construction. Values in bold indicate the factor scores used. Variables
Commonalities
Factors 1
Upstream water level S EC Fe NH4 NO3 PO4 TP TSS Turb. T Autovalues (F)
Rotation sums of squared loadings
2
3
4
0.804
0.002
0.696
0.017
0.565
0.695 0.849 0.712 0.64 0.752 0.664 0.784 0.756 0.699 0.845
0.802 0.212 0.838 0.196 0.132 0.757 0.865 0.812 0.79 0.218 4.578
0.168 0.021 0.072 0.773 0.492 0.24 0.1 0.282 0.123 0.162 1.436
0.01 0.054 0.062 0.001 0.663 0.138 0.135 0.124 0.235 0.86 1.178
0.154 0.895 0.021 0.061 0.227 0.115 0.087 0.03 0.066 0.177 1.007
37.27 37.27
14.07 51.34
11.774 63.114
11.43 74.544
% Variance % Cumulative variance
The resulting values of these quality curves indicate the water quality at the study site, with higher qi values indicating higher water quality than lower qi values. All quality curves are shown in Fig. 2. With these quality curve data, the weights, wi, to compose the WQI were calculated. The wi is the weight corresponding to the i-th parameter, a number between 0 and 1, assigned according to its importance to the overall water quality conformation. The sum of the weights wi equals 1 (Toledo and Nicolella, 2002). The wi values were calculated as displayed in Eq. (8):
PL n X F i Ail wi ¼ PL l¼1 ; wi ¼ 1 Pn l¼1 i¼1 F l Ai i¼1
ð8Þ
where: wi – weight of the WQI parameter; F – autovalues of the main components; Ai – explicability of the variable of the principal component; i – assumes values from 1 to n according to the variables selected in the model; l – assumes values from 1 to L according to the number of components and n – the n-th parameter selected to compose the index. 3. Results and discussion
Table 7 Calculations performed to obtain the qi curve for S. S
qi
3.1. Physico-chemical parameters and metal concentrations used for the construction of the QI and WQI
S 6 0.5 0.5 < S 6 2.4 2.4 < S 6 2.9 2.9 < S 6 3.1 3.2 3.3 3.4 S > 3.4
5 5 (eS) 4.6 (eS) 4.2 (eS) 3.85 (eS) 3.5 (eS) 3.17 (eS) 100
The parameters analyzed at the Tucuruí Hydroelectric Power Plant reservoir are displayed in Table 8. The observed variations are due to the different sampling stations environments, and will evidently be reflected in the QI and WQI, discussed in the next section. 3.2. Application of the Quality Indicator (QI)
were transformed in the range between 20 and 80, and then by applying Eq. (6).
qi ¼ 85 pi ; ði ¼ CEÞ
ð6Þ
For temperature we followed the SEMAD recommendation (SEMAD, 2005), using the equation qi = 93, since Brazil does not show much water temperature variations. The TSS qi was estimated by the method proposed by SEMAD (2005), where, when TSS values are higher than 500 units, TSS qi = 30, and when TSS values are lower than 500 units, the qi is calculated as follows:
qTSS ¼ 133:17 eð0:0027TSSÞ 53:17 eð0:0141TSSÞ þ ½ð6:2 e0:00462TSSÞ senð0:0146 TSSÞ
ð7Þ
Low QI values represent better water quality than higher QI values. Consequently, high values for PO4, CE, TP, TSS, NH4, NO3, Turb and Fe negatively affect water quality, causing higher QI values, since they are positively weighted. Likewise, higher water upstream levels, transparency and temperature contribute to better water quality, causing lower QI values, since they are negatively weighted. The upstream water levels used in the construction of the QI are directly correlated to the local hydrological cycle. The application of the QI on the study area during the four distinct phases of the local hydrological cycle throughout the study years showed no significant differences between the study years, with minimum and maximum of 0.96 and 3.52, respectively, meaning that water quality is not significantly modified during each cycle over the years. These results are displayed in Table 9.
680
T.C. Lobato et al. / Journal of Hydrology 522 (2015) 674–683
Fig. 2. Constructed quality curves for the relevant investigated parameters: Total P, NO3, Total Fe, NH4, TSS, turbidity, upstream water level, S, PO4 and EC.
Table 8 Results of the parameters analyzed at the Tucuruí Hydroelectric Power Plant reservoir at the 11 sampling stations from 2009 to 2012. Variables
N
Means
SD
CV (%)
Upstream level (m) S (m) T (°C) DO (mg L1) pH EC (lS cm1) Cl (mg L1) Fe (mg L1) Ca (mg L1) Turb. (NTU) Mg (mg L1) Na (mg L1) K (mg L1) NH4 (lg L1) NO3 (mg L1) PO4 (mg L1) TP (mg L1) TSS (mg L1) Cl a (mg L1)
180 180 180 180 180 180 136 180 180 180 180 180 180 180 180 179 180 180 180
68.26 ± 0.78 2.25 ± 0.15 29.99 ± 0.13 6.15 ± 0.10 7.14 ± 0.04 45.96 ± 1.42 2.82 ± 0.10 0.68 ± 0.08 4.15 ± 0.09 6.55 ± 1.09 1.51 ± 0.06 2.93 ± 0.19 1.50 ± 0.08 52.73 ± 9.34 33.38 ± 5.28 13.50 ± 1.11 24.48 ± 2.31 3.81 ± 0.729 6.00 ± 0.50
5.37 1.02 0.90 0.72 0.29 9.73 0.62 0.57 0.62 7.47 0.43 1.30 0.54 63.94 36.15 7.61 15.81 4.99 3.45
7.87 45.48 2.99 11.69 4.00 21.18 22.13 84.26 15.07 114.16 28.65 44.29 36.29 121.25 108.30 56.36 64.57 130.84 57.58
the means of the QI for each hydrological cycle. The ‘‘filling’’ cycle showed statistically significantly higher QI values, corresponding to December, January and February, while the ‘‘emptying’’ cycle presented statistically significantly lower QI values, corresponding to June, July and August, while the ‘‘empty’’ and ‘‘full’’ cycles were not statistically different. Since significant differences were observed for the filling and empty periods, it is possible to conclude that the dynamic nature of these periods is significantly different from the static nature of dry and empty situations of the reservoir. Fig. 3 shows the box-plots of the QI considering the study years and the hydrological cycle (HC). The results concerning the application of the QI on the water sampling stations to indicate water quality are displayed in Fig. 4. 3.3. Application of the Water Quality Indicator (WQI)
SD = Standard deviation; CV = Coefficient of variation.
As the construction of the QI did not take into account the hydrological cycle as one of its variables, the QI was calculated continuously throughout the four distinct hydrological cycles and divided according to these four cycles (empty, filling, full and emptying). Subsequently, Student’s t-test was applied to compare
The WQI is categorized as displayed in Table 10, following CETESB recommendations. The WQI was able to distinguish water quality at differently impacted sampling sites. The water quality at sampling stations MIP, MJV, ML, MP and M3 was rated as ‘‘Acceptable’’. This was expected, since each of these stations is near a fairly or heavily populated area. Sampling station MP, on the other hand, occupies a larger area, near the reservoir local fishing pole, which suffers from anthropogenic impacts, and was categorized between the ‘‘acceptable’’ and ‘‘poor’’ water quality. Sampling station MIP, located near the city of Itupiranga, which has been intensely urbanized since the late 80s, was also categorized in this water quality range. As was
Table 9 Application of the QI on the study area during the four distinct phases of the local hydrological cycle throughout the study years. Hydrological cycle
N
Minimum
Maximum
Means
SE
SD
Dry Filling Full Emptying
38 47 50 45
0.67 0.88 0.79 0.96
0.76 3.52 2.12 0.87
0.2138 0.5092 0.0594 0.4319
0.04691 0.15313 0.08340 0.05161
0.28918 104.984 0.58970 0.34621
SE = Standard error; SD = Standard deviation.
T.C. Lobato et al. / Journal of Hydrology 522 (2015) 674–683
681
Fig. 3. Box-plots of the QI (2) considering the study years (a) and (b) the hydrological cycle (HC).
Fig. 4. Results concerning the application of the QI on the water sampling stations of the Tucuruí Hydroelectric Power Plant reservoir to investigate water quality.
Table 10 Categories proposed by Cetesb (2005) for the WQI. Value
Classification
80–100
Excellent
52–79
Good
37–51
Acceptable
20–36
Bad
0–19
Poor
Color code
MJV, located in a region with two harbors, ’’Porto Novo’’ (Jacundá municipality) and ‘‘Port of Cologne’’ (municipality of Goianésia of Pará), while ML is located near the indigenous land of the Parakanas Indians, and was categorized mostly as ‘‘acceptable’’. M3, located in the central part of the reservoir, which depends on the water inflow of the reservoir, was also categorized between ‘‘acceptable’’ and ‘‘poor’’ water quality. Despite being located near a fairly populated area, the water quality as C2 and MBB sampling stations were classified as ‘‘Good’’, as was C1. MBL a natural wildlife preserve with no anthropogenic impacts, indicates the applicability of the WQI, since water quality was categorized as ‘‘Good’’ in this sampling station. M1, one of the most important monitoring stations, representing the water captured by electric generating units and the water released downstream of the dam, was also very close to qualifying as ‘‘Good’’. These results regarding the application of the WQI at the Tucuruí Hydroelectric Power Plant reservoir for each sampling site are displayed in Fig. 5, following the same color-code as displayed in Table 10. An overall analysis applying the WQI to the entire reservoir was also conducted. Figs. 6 and 7 were created using the free software gvSIG 1.9 with the implementation of the inverse distance weighted (IDW) interpolation algorithm, where the closest points have greater weight, i.e., when the distance increases their weight decreases. Results show that the water quality for the reservoir during the study years was categorized in the ‘‘Acceptable’’ range throughout the study years, indicating stability in the evaluated parameters of the reservoir (Table 11 and Fig. 7). Reflecting the QI results, the WQI results applied to the entire reservoir also demonstrate that the hydrological cycles directly
Fig. 5. Water Quality Index (WQI) for the Tucuruí Hydroelectric Power Plant reservoir sampling stations from 2009 to 2012.
682
T.C. Lobato et al. / Journal of Hydrology 522 (2015) 674–683
Fig. 6. WQI categorization for the Tucuruí Reservoir from 2009 to 2012.
Fig. 7. WQI categorization for the Tucuruí Reservoir from 2009 to 2012 during the regional hydrologic cycle (dry, filling, full and emptying).
Table 11 Means of the results obtained in the WQI sampling stations in the years 2009, 2010, 2011 and 2012. Stations 2009 2010 2011 2012 C1 46.86 49.00 52.87 50.94 C2 49.35 62.77 56.81 57.46 M1 45.95 51.25 41.23 55.20 M3 38.24 41.46 41.29 43.82 MBB 54.02 54.02 44.93 56.44 MBL 49.02 54.32 55.07 59.99 MIP 34.92 32.15 38.35 37.09 MJV 40.36 37.36 42.41 40.62 ML 40.80 44.15 41.05 42.67 MP 35.28 31.59 44.97 48.64 MR 45.78 53.12 56.67 55.23 Means 43.69 46.47 46.88 49.83 Deviation 6.19 9.99 7.03 7.75 CV 14.17 21.49 15.00 15.55 SD = Standard deviation; CV = Coefficient of variation
Means 49.92 56.60 48.41 41.20 52.35 54.60 35.63 40.19 42.17 40.12 52.70 46.72 7.08 15.15
SD 2.58 5.52 6.10 2.29 5.08 4.49 2.72 2.09 1.56 8.01 4.84 4.12 2.02 49.15
CV (%) 5.17 9.75 12.61 5.55 9.70 8.22 7.63 5.21 3.70 19.96 9.18 8.79 4.53 51.54
SD = Standard deviation; CV = Coefficient of variation.
affect water quality (Table 8 and Fig. 7), since during the flood period water quality was categorized as ‘‘Acceptable’’, close to ‘‘Poor’’, while during emptying the water quality was categorized as ‘‘Good’’, and no significant differences were observed between the empty and full periods (see Table 12).
Table 12 Means of the results obtained in the sampling stations in the years 2009, 2010, 2011 and 2012. Stations Empty Filling Full Emptying C1 50.63 49.59 46.43 55.12 C2 46.44 47.48 57.58 59.85 M1 59.09 38.75 45.33 61.61 M3 47.38 31.75 33.50 53.70 MBB 57.42 43.98 49.16 58.18 MBL 52.29 53.20 52.44 55.95 MIP 45.03 24.98 27.56 46.75 MJV 42.13 30.09 36.77 53.10 ML 44.48 29.21 36.76 61.38 MP 37.49 28.83 48.36 54.39 MR 49.03 49.58 50.25 59.34 Means 48.31 38.86 44.01 56.31 SD 6.37 10.25 9.13 4.39 CV 13.19 26.37 20.74 7.80 SD = Standard deviation; CV = Coefficient of variation
Means 50.44 52.84 51.20 41.58 52.19 53.47 36.08 40.52 42.96 42.27 52.05 46.87 6.22 13.27
SD 3.59 6.86 10.96 10.68 6.83 1.70 11.40 9.72 13.77 11.36 4.89 8.34 3.81 45.69
CV (%) 7.12 12.99 21.40 25.69 13.08 3.18 31.59 24.00 32.06 26.89 9.39 18.85 10.11 53.61
SD = Standard deviation; CV = Coefficient of variation.
4. Conclusions The novel quality indicator and water quality index created in the present study for the evaluation of the water quality of Tucuruí Hydroelectric Plant reservoir were constructed taking into account
T.C. Lobato et al. / Journal of Hydrology 522 (2015) 674–683
specific Amazon characteristics. The application of both indices to investigate water quality at the reservoir showed no significant differences for both between the study years. The QI was useful to determine seasonal differences in water quality due to the hydrological cycle of the area. The hydrological cycle was, in fact, shown by the indices to directly affect reservoir water quality. The WQI was also useful in observing anthropogenic impacts in the area, since water sampling stations in fairly populated areas ranged from ‘‘Poor’’ to ‘‘acceptable’’, while water quality in the environmental preservation area and near the dam was categorized as ‘‘Acceptable’’. These indices can, thus, be helpful in environmental monitoring situations that suffer well-defined hydrological cycles like those in tropical areas, in addition to being valuable tools in the detection of anthropogenic impacts. The statistical techniques applied in the construction of the QI and WQI can also be used to construct other specific indices in different geographical areas, taking into account the specificities for each area. Acknowledgement The main author would like to thank CAPES for the MSc scholarship. References Abaurrea, J., Asin, J., Cebrian, A.C., Garcia-Vera, M.A., 2011. Trend analysis of water quality series based on regression models with correlated errors. J. Hydrol. 400 (3–4), 341–352. Andrade, E.M., Araújo, L.F.P., Rosa, M.F., Gomes, R.B., Lobato, F.A.O., 2007. Fatores determinantes da qualidade das águas superficiais na Bacia do Alto Acaraú, Ceará, Brasil. Ciência Rural 37, 1791–1797. Aubert, A.H., Gascuel-Odoux, C., Merot, P., 2013. Annual hysteresis of water quality: a method to analyse the effect of intra- and inter-annual climatic conditions. J. Hydrol. 478, 29–39. Boyacioglu, H., 2006. Surface water quality assessment using factor analysis. Water SA 32 (3), 389–393. Braga, B., Porto, M., Tucci, C.E.M., 2006. Monitoramento de Quantidade e Qualidade das Águas, Águas Doces no Brasil: Capital Ecológico, Uso e Conservação. Escrituras, pp. 145–160. Brito, L.T.L., Silva, A.S., Srinivasa, V.S., Galvão, C.O., Gheyi, H.R., 2006. Uso de análise multivariada na classificação das fontes hídricas superficiais da bacia hidrográfica do salitre. Engenharia. Agrícola 26, 58–66. Brown, R.M., McLelland, N.J., Deininger, R.A., Tozer, R.G., 1970. A water quality index: do we dare? Water Sewage Works, 339–343. Carlson, R.E., 1977. A trophic state index for lakes. Limnol. Oceanogr. 22 (2), 361– 369. CETESB, 2009. Report – Significado ambiental e sanitário das variáveis de qualidade das águas e dos sedimentos e metodologias analíticas e de amostragem. 40. El-Iskandarani, M., Nasr, S., Okbah, M., Jensen, A., 2004. Principal components analysis for quality assessment of the Mediterranean coastal water of Egypt. Model., Measur. Control C 65, 69–83.
683
Fisch, G.F., Januário, M., Senna, R.C., 1990. Impacto ecológico em Tucuruí (PA): Climatologia. Acta Amazonica 20, 49–60. Guedes, H.A.S. et al., 2012. Aplicação da análise estatística multivariada no estudo da qualidade da água do Rio Pomba, MG. Revista Brasileira de Engenharia Agrícola e Ambiental 16, 558–563. Haase, J., Stringuini, M.H., Silva, M.L.B.C., Rodrigues, M.L.K., Koch, S.M.V., 2003. Qualidade das águas superficiais do litoral norte e médio do Rio Grande do Sul. 22 Congresso brasileiro de engenharia sanitária e ambiental, Joinville – SC. Anais. Rio de Janeiro, ABES, VI 049, p. 17. Hair, J.F., Anderson, R.E., Tathan, R.L., Black, W.C., 1998. Multivariate Data Analysis. Prentice Hall, New Jersey. Kazi, T.G. et al., 2009. Assessment of water quality of polluted lake using multivariate statistical techniques: a case study. Ecotoxicol. Environ. Saf. 72 (2), 301–309. Lai, Y.C., Tu, Y.T., Yang, C.P., Surampalli, R.Y., Kao, C.M., 2013. Development of a water quality modeling system for river pollution index and suspended solid loading evaluation. J. Hydrol. 478, 89–101. Laureano, J.S.F., Navar, J., 2002. An assessment of stream water quality of the Rio San Juan, Nuevo Leon, Mexico, 1995–1996. J. Environ. Qual. 31 (4), 1256–1265. Lewis, W.M., 2000. Basis for the protection and management of tropical lakes. Lakes Reserv. Res. Manage. 5, 35–48. Lucas, A.A.T., Folegatti, M.V., Duarte, S.N., 2010. Qualidade da água em uma microbacia hidrográfica do Rio Piracicaba, SP. Revista Brasileira de Engenharia Agrícola e Ambiental 14, 937–943. Merten, G.H., Minella, J.P., 2002. Qualidade da água em bacias hidrográficas rurais: um desafio atual para a sobrevivência futura. Agroecol. e Desenvol. Rur. Sustent. 3, 33–38. Petersen, W., Bertino, L., Callies, U., Zorita, E., 2001. Process identification by principal component analysis of river water-quality data. Ecol. Model. 138 (1– 3), 193–213. Rebouças, A.C., Braga, B., Tundisi, J.G., 2002. Águas doces no Brasil: capital ecológico, uso e conservação. Escrituras Editora, São Paulo. Selle, B., Schwientek, M., Lischeid, G., 2013. Understanding processes governing water quality in catchments using principal component scores. J. Hydrol. 486, 31–38. SEMAD, 2005. Secretaria de Estado do Meio Ambiente e Desenvolvimento Sustentável de Minas Gerais/Unidade de Coordenação Estadual. Sistema de Cálculo da Qualidade da Água (SCQA) – Estabelecimento das Equações do índice de Qualidade das Águas (IQA). Shah, V.G. et al., 2007. Comparisons of water quality parameters from diverse catchments during dry periods and following rain events. Water Res. 41 (16), 3655–3666. Singh, K.P., Malik, A., Mohan, D., Sinha, S., 2004. Multivariate statistical techniques for the evaluation of spatial and temporal variations in water quality of Gomti River (India) – a case study. Water Res. 38 (18), 3980–3992. Standard Methods, 2013. 1060 Collection and Preservation of Samples. (accessed 10.09.13). Sucksdorff, A., 1984. Pantanal – um paraíso perdido., Rio de Janeiro., 160pp. Toledo, L.G., Nicolella, G., 2002. Índice de qualidade de água em microbacia sob uso agrícola e urbano. Sci. Agric. 59 (1), 181–186. Tundisi, J.G., Tundisi, T.M., 2008. Limnologia. Oficina de Textos, 631pp. Vialle, C. et al., 2011. Monitoring of water quality from roof runoff: Interpretation using multivariate analysis. Water Res. 45 (12), 3765–3775. Vonberg, D., Vanderborght, J., Cremer, N., Pütz, T., Herbst, M., Vereecken, H., 2014. 20 years of long-term atrazine monitoring in a shallow aquifer in western Germany. Water Res. 50, 294–306. WCD, 2000. World Commission on Dams – Tucuruí Hydroelectric Plant (Brazil), p. 297.
Contents lists available at ScienceDirect
Journal of Hydrology journal homepage: www.elsevier.com/locate/jhydrol
Construction of a novel water quality index and quality indicator for reservoir water quality evaluation: A case study in the Amazon region T.C. Lobato a,⇑, R.A. Hauser-Davis b,⇑, T.F. Oliveira a, A.M. Silveira a, H.A.N. Silva a, M.R.M. Tavares a, A.C.F. Saraiva c a Universidade Federal do Pará (UFPA), Instituto de Ciências Exatas e Naturais, Faculdade de Estatística and Faculdade de Ciência da Computação, Rua Augusto Correa, 01, CEP: 66075-110 Belém, PA, Brazil b Pontifícia Universidade Católica do Rio de Janeiro (PUC-Rio), Departamento de Química, Laboratório de Bioanalítica, Rua Marquês de São Vicente, 225, Gávea, CEP: 22453-900 Rio de Janeiro, RJ, Brazil c Centro de Tecnologia da ELETRONORTE, Rodovia Arthur Bernardes, S/N – Miramar, Telegrafo sem Fio, CEP: 66.115-000 Belem, PA, Brazil
a r t i c l e
i n f o
Article history: Received 21 July 2014 Received in revised form 27 October 2014 Accepted 7 January 2015 Available online 22 January 2015 This manuscript was handled by Andras Bardossy, Editor-in-Chief, with the assistance of Sheng Yue, Associate Editor Keywords: Hydrological cycle Statistical analyses Water quality indicator Quality index Amazon area
s u m m a r y A novel Quality Indicator (QI) and Water Quality Index (WQI) were constructed in the present study for the evaluation of the water quality of a Hydroelectric Plant reservoir in the Amazon area, Brazil, taking into account the specific characteristics of the Amazon area. Factor analyses were applied in order to select the relevant parameters to be included in the construction of both indices. Quality curves for each selected parameter were then created and the constructed QI and WQI were then applied to investigate the water quality at the reservoir. The hydrological cycle was shown by the indices to directly affect reservoir water quality, and the WQI was further useful in identifying anthropogenic impacts in the area, since water sampling stations suffering different anthropogenic impacts were categorized differently, with poorer water quality, than stations near the dam and the environmental preservation area, which suffer significantly less anthropogenic impacts, and were categorized as presenting better water quality. The constructed indices are thus helpful in investigating environmental conditions in areas that show well-defined hydrological cycles, in addition to being valuable tools in the detection of anthropogenic impacts. The statistical techniques applied in the construction of these indices may also be used to construct other indices in different geographical areas, taking into account the specificities for each area. Ó 2015 Elsevier B.V. All rights reserved.
1. Introduction Water quality is largely determined by natural processes, such as weathering and soil erosion, but also by anthropogenic inputs (Kazi et al., 2009; Singh et al., 2004). To assess human impacts on water quality, variations in space and time and in the biological, physical and chemical processes of natural systems should be considered. One important factor in this context is the hydrological cycle, which directly affects the drainage network of water bodies and may causes surface runoff along riverbanks. This, in turn, may lead to contamination by sediments and several types of pollutants, besides directly affecting the local vegetation and biota (Aubert et al., 2013; Lai et al., 2013; Merten and Minella, 2002). By evaluating these hydrological variations and their effects, it is possible to create management actions that can intervene in the recovery and/or preservation of ecosystems (Abaurrea et al., 2011; Braga et al., 2006; Lucas ⇑ Corresponding authors. E-mail addresses: [email protected] (T.C. Lobato), rachel.hauser. [email protected] (R.A. Hauser-Davis). http://dx.doi.org/10.1016/j.jhydrol.2015.01.021 0022-1694/Ó 2015 Elsevier B.V. All rights reserved.
et al., 2010; Shah et al., 2007; Tundisi and Tundisi, 2008). Water quality assessments also allow for decision-making policies to be made regarding water potability and use. Several water quality assessment studies have been conducted by applying statistical techniques, such as the principal component analysis, which can aid in identifying natural or anthropogenic factors that can cause alterations in water quality (Andrade et al., 2007; Boyacioglu, 2006; Brito et al., 2006; El-Iskandarani et al., 2004; Guedes et al., 2012; Laureano and Navar, 2002; Petersen et al., 2001; Selle et al., 2013; Vialle et al., 2011; Vonberg et al., 2014). Some studies have applied this statistical technique to specifically propose a Water Quality Index (WQI), by using the weighted scores of each analyzed water quality parameter (Haase et al., 2003; Toledo and Nicolella, 2002). WQIs usually take into account general water parameters, such as dissolved oxygen, pH, temperature, turbidity, and NH3 concentrations, among others. UNESCO defines the Amazon region and adjacent areas like the Pantanal as world interest areas due to their unique flora and fauna and great biodiversity. These regions have, however, been increasingly under anthropogenic pressure (Sucksdorff, 1984). One of the
T.C. Lobato et al. / Journal of Hydrology 522 (2015) 674–683
main impacts in these areas has been the construction of Hydroelectric Power Plants reservoirs, which have become the predominant lake type in many regions throughout the world (Lewis, 2000). These tropical areas go though well-defined defined hydrological cycles each year, and this feature has, increasingly, been studied to obtain environmental information which may help in the monitoring of environmental impacts caused by anthropogenic pressures and may lead to decision-making in this regard. However, to the best of our knowledge, no WQI data in the literature has been constructed taking into account the water transparency or hydrological cycle of the Amazon area. In this context, the aims of the present study were to create a new Water Quality Index (WQI) by applying multivariate statistical techniques, taking into account the water transparency and specific hydrological cycle of the Amazon, besides other selected parameters. Another objective was the creation of a Quality Indicator (QI), by constructing water quality index curves, and the subsequent application of the constructed QI to the investigation of the water quality of a Hydroelectric Power Plant reservoir in the Amazon region, in order to verify the impact of this construction on the reservoir’s water quality throughout the well-defined hydrological cycles of the area. 2. Material and methods 2.1. Study area and sampling stations The Tucuruí Hydroelectric Plant is located in the state of Pará, Brazil, at latitude 03°430 –05°150 , longitude 49°120 –50°000 W
675
(Fig. 1). The plant was constructed in September 1984 on the river Tocantins, at approximately 7 km from the town of Tucuruí and 300 km from the city of Belém, the state capital. It is the first large-scale (25 units) hydroelectric project in the Brazilian Amazon rainforest, with an installed capacity of 8370 MW. The main purpose of the dam is hydroelectric power production to the Brazilian states of Maranhão and Pará and navigation between the upper and lower Tocantins river. The reservoir has a total flooded area of approximately 2850 m2, with approximately 50.8 million m3 of water and water residence time of 46 days (WCD, 2000). The area is characterized by rainy (December to May) and dry (June to November) seasons, with annual precipitations between 2250 and 2500 mm. The rainiest month is March and the driest is September (Fisch et al., 1990). The local hydrological cycle in the reservoir is reflected in the local water levels, and, consequently, the reservoir goes through a cycle constituted by a full stage, during March, April and May, an emptying stage, during June, July and August, an empty stage during September, October and November, and a filling stage during December, January and February. Eleven water sampling stations (C1, C2, M1, M3, MR, MBB, MBL, MP, MIP, ML and MJV) are present at the Tucuruí Hydroelectric Plant reservoir also displayed in Fig. 1. All are located upstream and represent the specific characteristics of each area within the reservoir. The large influx of people to this area has also led to deforestation and negative impacts from increased cattle-raising activities. Population increases have also strained the existing infrastructure, or lack thereof (Rebouças et al., 2002; WCD, 2000), leading to concerns regarding anthropogenic influences in the area.
Fig. 1. Location of the Tucuruí Hydroelectric Power Plant reservoir, indicating the water sampling stations.
676
T.C. Lobato et al. / Journal of Hydrology 522 (2015) 674–683
2.2. Physico-chemical water characteristics A total of 180 water samples were collected at the eleven sampling stations, from 2009 to 2012, as displayed in Table 1. The samples were collected from approximately 5 cm below the water surface, using 2.5 L capacity Van Dorn bottles. The samples were stored in 50 mL polypropylene tubes and acidified with nitric acid 1 mol L1, according to USA EPA Standard Method 1060B. The following physico-chemical parameters were determined: chloride (Cl), temperature (T), transparency (S), electrical conductivity (EC), pH, Total Suspended Solids (TSS), ammonia (NH4), nitrate (NO3), chlorophyll a (Cl a), turbidity (Turb.), dissolved oxygen (DO) (Winkler method), phosphates (PO4) and total phosphorous (TP), according to the Standard Methods norms (StandardMethods, 2013). The pH, EC and T parameters were determined in situ. Metal determinations (Ca, Fe, K, Mg, Na) were also conducted, using optical emission spectrometry by inductive coupling plasma (ICP–OES – Varian), a viable, multielemental, simple and quick technique for monitoring macro and micronutrients. The standard reference material used for recovery tests was SLRS4, River Water Reference Material for Trace Metals (National research Council, Canada). 2.3. Construction of the Quality Indicator (QI) The QI was created based on a factor analysis that selected factors that better explain the water quality of the reservoir, by the application of a principal component analysis without the use of a Varimax rotation (Table 2). The Bartlett method was applied to the scores of this factor analysis, to verify if commonalities were above or below 0.50 (Hair et al., 1998; Toledo and Nicolella, 2002). If parameters displayed values below 0.50, the respective variable was excluded from the subsequent analysis. The variables pH, DO, Ca, K, Mg, Na and Cl all showed commonalities below 0.50, not correlating amongst themselves or the other investigated variables. In addition, the values of these parameters were also compared to CONAMA class 2 maximum permitted values for Brazilian waters, and none were higher than these permitted values, in accordance with this legislation. Thus, because these parameters showed very low commonalities, not being responsible for explaining water quality in the study area, and not being worrisome regarding water quality according to the CONAMA legislation, they were excluded from further analyses. The correlation matrix for the analyzed parameters is displayed in Table 3. After this exclusion step, the results of the factor analysis indicated four factors with a total variance of 74.54%. The first factor extracted from the factor analysis explained 41.62% of the data variability, indicating that it is the main factor to be taken into
Table 1 Water sampling frequency per study year at the eleven water sampling stations located at the Tucuruí Hydroelectric Plant reservoir, Brazil. Sampling station
2009
2010
2011
2012
Total
C1 C2 M1 M3 MBB MBL MIP MJV ML MP MR Total
4 6 4 4 3 6 3 1 2 5 5 43
3 1 3 4 4 3 3 5 6 2 6 40
3 6 3 3 3 4 4 6 5 5 5 47
4 3 12 4 4 3 4 4 4 4 4 50
14 16 22 15 14 16 14 16 17 16 20 180
Table 2 Factor analysis results to obtain the relevant parameters to take part in the construction of the Quality Indicator (QI). Component Initial Eigenvalues
1 2 3 4 5 6 7 8 9 10 11
Extraction sums of squared loadings
Total % of variance
Cumulative Total % of % variance
Cumulative %
4.578 41.622 1.436 13.055 1.178 10.709 1.007 9.157 0.756 6.869 0.562 5.105 0.436 3.96 0.399 3.626 0.285 2.59 0.195 1.771 0.169 1.536
41.622 54.677 65.387 74.544 81.413 86.518 90.478 94.104 96.694 98.464 100.00
41.622 54.677 65.387 74.544
4.578 41.622 1.436 13.055 1.178 10.709 1.007 9.157
account, composed by the following parameters: upstream level, EC, S, PO4, TP, TSS, NH4, NO3, T, Turb. and Fe. The Bartlett method was then applied to the scores of this first factor, to verify commonalities above 0.50 (Table 4) (Hair et al., 1998; Toledo and Nicolella, 2002). All scores were above 0.50, and thus, all parameters were taken into account, and subsequently used to construct the QI, displayed in Eq. (1).
QI ¼ 0:19 TP þ 0:19 TSS þ 0:18 Turb: þ 0:18 Fe 0:18 S þ 0:15 PO4 0:09 T þ 0:08 EC þ 0:08 NO3 þ 0:08 NH4 0:05 Upstream water Level
ð1Þ
2.4. Construction of the Water Quality Index (WQI) The WQI created in the present study was based on the index developed by the National Sanitation Foundation (NSF) (Brown et al., 1970). The NSF WQI considers nine weighted parameters to characterize water quality: Dissolved Oxygen (DO), fecal coliforms, pH, biochemical oxygen demand (BOD), total nitrogen (N), total phosphorous (TP), Temperature (T), Turbidity (Turb.) and Total Suspended Solids (TSS). As the WQI created in the present study was adapted to a tropical reservoir, the parameters included in its construction were different, in order to reflect the specific geographical characteristics of the area. With this in mind, a factor analysis without rotation of all the parameters determined in the present study was conducted, and the following parameters were chosen to take part in the WQI construction: EC, S, PO4, TP, TSS, NH4, NO3, Turb., Fe, upstream water levels and T. The factor analysis results are displayed in Table 5. The Bartlett method was then applied to the scores of the four factors, to verify commonalities above 0.50 (Table 6) (Hair et al., 1998; Toledo and Nicolella, 2002). All scores were above 0.50, and thus, all parameters were taken into account, and subsequently used to construct the WQI, displayed in Eq. (2).
WQI ¼
n Y qwi i
ð2Þ
i¼1
where WQI – water quality index, in the range of 0–100; qi – quality of the i-th parameter; n – the size of the parameters selected to compose the index; wi – weight corresponding to the i-th parameter. The results of the factor analysis were used to calculate the quality curves (qi) of the relevant parameters obtained by the application of the QI, for subsequent WQI application. The other qi curves were obtained from correlated linear and non-linear studies. When available, the quality curves from the NSF were applied,
Table 3 Correlation matrix for the analyzed parameters. S
T
DO
pH
EC
Cl
Fe
Ca
Mg
Na
K
NH4
NO3
PO4
TP
TSS
0.186 0.012 180
0.018 0.807 180
0.054 0.469 180
0.261 0 180
0.336 0 180
0.015 0.86 136
0.007 0.93 180
0.023 0.754 180
0.158 0.034 180
0.052 0.49 180
0.157 0.036 180
0.33 0 180
0.172 0.021 180
0.079 0.296 179
0.122 0.102 180
0.261 0 180
0.177 0.017 180
0.111 0.139 180
0.208 0.005 180
0.287 0 180
0.08 0.352 136
0.619 0 180
0.059 0.433 180
0.102 0.173 180
0.09 0.231 180
0.321 0 180
0.229 0.002 180
0.186 0.013 180
0.531 0 179
0.606 0 180
0.642 0 180
0.025 0.736 180
0.005 0.949 180
0.179 0.016 180
0.2 0.019 136
0.219 0.003 180
0.145 0.052 180
0.026 0.727 180
0.044 0.559 180
0.262 0 180
0.02 0.788 180
0.296 0 180
0.286 0 179
0.328 0 180
0.253 0.001 180
1
0.189 0.011 180
0.12 0.109 180
0.089 0.302 136
0.044 0.56 180
0.287 0 180
0.091 0.226 180
0.054 0.468 180
0.179 0.016 180
0.068 0.363 180
0.015 0.837 180
0.018 0.808 179
0.009 0.908 180
0.132 0.076 180
0.141 0.059 180
0.085 0.323 136
0.091 0.225 180
0.114 0.128 180
0.163 0.029 180
0.126 0.091 180
0.088 0.241 180
0.106 0.159 180
0.127 0.089 180
0.062 0.411 179
0.01 0.89 180
0.124 0.099 180
0.145 0.091 136
0.274 0 180
0.355 0 180
0.014 0.856 180
0.072 0.337 180
0.224 0.003 180
0.055 0.467 180
0.049 0.514 180
0.273 0 179
0.256 0.001 180
0.159 0.033 180
1
0.053 0.541 136
0.042 0.628 136
0.003 0.975 136
0.351 0 136
0.404 0 136
0.045 0.6 136
0.033 0.707 136
0.05 0.568 135
0.167 0.052 136
0.083 0.339 136
0.149 0.045 180
0.025 0.743 180
0.225 0.002 180
0.298 0 180
0.26 0 180
0.226 0.002 180
0.568 0 179
0.754 0 180
0.621 0 180
0.103 0.17 180
0.04 0.59 180
0.331 0 180
0.051 0.494 180
0.052 0.49 180
0.213 0.004 179
0.101 0.179 180
0.015 0.84 180
0.084 0.264 180
0.109 0.145 180
0.387 0 180
0.054 0.469 180
0.017 0.817 179
0.059 0.434 180
0.09 0.229 180
0.457 0 180
0.08 0.284 180
0.053 0.482 180
0.036 0.629 179
0.193 0.009 180
0.03 0.693 180
0.16 0.032 180
0.026 0.733 180
0.202 0.007 179
0.358 0 180
0.258 0 180
1
0.291 0 180
0.053 0.481 179
0.26 0 180
0.253 0.001 180
0.111 0.141 179
0.204 0.006 180
0.272 0 180
0.618 0 179
0.467 0 179
1
0.725 0
Upstream water level
Pearson Correlation Sig. (2-tailed) N
1
S
Pearson Correlation Sig. (2-tailed) N
0.186 0.012 180
1
T
Pearson Correlation Sig. (2-tailed) N
0.018 0.807 180
0.177 0.017 180
1
DO
Pearson Correlation Sig. (2-tailed) N
0.054 0.469 180
0.111 0.139 180
0.025 0.736 180
180
pH
Pearson Correlation Sig. (2-tailed) N
0.261 0 180
0.208 0.005 180
0.005 0.949 180
0.189 0.011 180
1
EC
Pearson Correlation Sig. (2-tailed) N
0.336 0 180
0.287 0 180
0.179 0.016 180
0.12 0.109 180
0.141 0.059 180
1
Cl
Pearson Correlation Sig. (2-tailed) N
0.015 0.86 136
0.08 0.352 136
0.2 0.019 136
0.089 0.302 136
0.085 0.323 136
0.145 0.091 136
136
Fe
Pearson Correlation Sig. (2-tailed) N
0.007 0.93 180
0.619 0 180
0.219 0.003 180
0.044 0.56 180
0.091 0.225 180
0.274 0 180
0.053 0.541 136
1
Ca
Pearson Correlation Sig. (2-tailed) N
0.023 0.754 180
0.059 0.433 180
0.145 0.052 180
0.287 0 180
0.114 0.128 180
0.355 0 180
0.042 0.628 136
0.149 0.045 180
1
Mg
Pearson Correlation Sig. (2-tailed) N
0.158 0.034 180
0.102 0.173 180
0.026 0.727 180
0.091 0.226 180
0.163 0.029 180
0.014 0.856 180
0.003 0.975 136
0.025 0.743 180
0.103 0.17 180
1
Na
Pearson Correlation Sig. (2-tailed) N
0.052 0.49 180
0.09 0.231 180
0.044 0.559 180
0.054 0.468 180
0.126 0.091 180
0.072 0.337 180
0.351 0 136
0.225 0.002 180
0.04 0.59 180
0.084 0.264 180
1
K
Pearson Correlation Sig. (2-tailed) N
0.157 0.036 180
0.321 0 180
0.262 0 180
0.179 0.016 180
0.088 0.241 180
0.224 0.003 180
0.404 0 136
0.298 0 180
0.331 0 180
0.109 0.145 180
0.457 0 180
1
NH4
Pearson Correlation Sig. (2-tailed) N
0.33 0 180
0.229 0.002 180
0.02 0.788 180
0.068 0.363 180
0.106 0.159 180
0.055 0.467 180
0.045 0.6 136
0.26 0 180
0.051 0.494 180
0.387 0 180
0.08 0.284 180
0.16 0.032 180
180
NO3
Pearson Correlation Sig. (2-tailed) N
0.172 0.021 180
0.186 0.013 180
0.296 0 180
0.015 0.837 180
0.127 0.089 180
0.049 0.514 180
0.033 0.707 136
0.226 0.002 180
0.052 0.49 180
0.054 0.469 180
0.053 0.482 180
0.026 0.733 180
0.291 0 180
1
PO4
Pearson Correlation Sig. (2-tailed) N
0.079 0.296 179
0.531 0 179
0.286 0 179
0.018 0.808 179
0.062 0.411 179
0.273 0 179
0.05 0.568 135
0.568 0 179
0.213 0.004 179
0.017 0.817 179
0.036 0.629 179
0.202 0.007 179
0.053 0.481 179
0.111 0.141 179
1
TP
Pearson Correlation Sig. (2-tailed)
0.122 0.102
0.606 0
0.328 0
0.009 0.908
0.01 0.89
0.256 0.001
0.167 0.052
0.754 0
0.101 0.179
0.059 0.434
0.193 0.009
0.358 0
0.26 0
0.204 0.006
0.618 0
180
180
180
180
180
180
180
180
180
180
180
179
677
(continued on next page)
T.C. Lobato et al. / Journal of Hydrology 522 (2015) 674–683
Upstream water level
0.778 0 180
180
1
180
T.C. Lobato et al. / Journal of Hydrology 522 (2015) 674–683
TSS
678
Table 4 Results of the Bartlett method to select significant commonalities for the analyzed parameters. Values in bold indicate the factor scores used.
0.642 0 180 0.251 0.001 180 0.17 0.022 180 0.095 0.205 180 Pearson Correlation Sig. (2-tailed) N Turb.
0.187 0.012 180
0.355 0 180
ð4Þ
0.666 0 180
0.121 0.105 180
0.081 0.348 136
0.541 0 180
0.035 0.639 180
0.033 0.663 180
0.025 0.74 180
0.3 0 180
0.141 0.06 180
0.536 0 179
180
0.725 0 180 0.467 0 179 0.272 0 180 0.253 0.001 180 0.258 0 180
179 180 180 180
0.03 0.693 180 0.09 0.229 180
180 180
0.015 0.84 180 0.621 0 180
180 136
0.159 0.033 180
0.115 0.176 0.74 0.449 0.175 0.26 0.025 0.098 0.164 0.017 0.252
1 4; ði ¼ NO3 Þ pi
0.132 0.076 180
180
4
0.276 0.114 0.258 0.593 0.007 0.162 0.082 0.013 0.053 0.057 0.549
ð3Þ
180
0.124 0.099 180
3
0.551 0.014 0.103 0.106 0.106 0.453 0.275 0.08 0.037 0.054 0.262
1 ði ¼ PO4 ; Total P; Turbidity; TFeÞ pi
180 180
0.642 0 180 0.261 0 180
180 N
2
0.053 0.175 0.091 0.08 0.177 0.077 0.154 0.19 0.186 0.181 0.078
qi ¼ 5
qi ¼ 3621:5pi ; ði ¼ NH4 Þ
Pearson Correlation Sig. (2-tailed) N
Component 1
using the limits established by Brazilian legislation CONAMA for class 2 freshwater bodies for each selected parameter. All calculations were made using the STATISTICAÒ software package. To develop the equations used in the construction of the WQI, pi, wi and qi values were calculated: The pi represents the value or concentration relative to the i-th parameter. The pi values of the selected parameters (NO3, NH4, PO4, TP and Turb.) were transformed on a scale from 0.05 to 1.00, where 0.05 was the value attributed to the minimum values and 1.00 to the maximum values obtained for each of these parameters in the collected water samples. For Fe, pi values were processed in the range between 0.05 and 1.60, applying dissolved iron values from the CONAMA legislation in Eq. (3). The qi is the quality of the i-th parameter, a number between 0 and 100, obtained from the respective mean curve of variations in water quality, depending on a concentration or measurement of the respective parameter. The qi values for each parameter were calculated according to Eqs. 3–5 for NH4, followed by the application of an inverse function. The final qi values were very similar to the desired curve representative of water quality (q), and were then multiplied by 5 to obtain values closer to the reality of the study area.
0.253 0.001 180
180
Upstream water level S T EC Fe NH4 PO4 TP TSS Turb NO3
qi ¼ 5
TSS
Table 3 (continued)
0.083 0.339 136
180
NO3 EC Upstream water level
S
T
DO
pH
Cl
Fe
Ca
Mg
Na
K
NH4
PO4
TP
Parameter
ð5Þ
No qi curves for water transparency are available in the literature for the Amazon area. Thus, the information given by qi, which is negative, meaning that better water quality is observed with higher transparency values, and Carlson’s Trophic state index (Carlson, 1977), which takes into account this parameter, were taken into account to create this qi curve. When water transparency increases, the qi increases exponentially according to the interval in which the transparency values are inserted, which is why we calculated qi = eS and multiplied the result by a factor that controls qi growth (Table 7). So, for S 6 0.5 m, qi = 5, indicating lower water quality, and for S > 3.4 m, qi = 100, indicating higher water quality. The upstream level qi was obtained by interpreting the QI. Higher values indicate better water quality than lower values. Water level values were normalized in the range of 5–90 to better interpret the results. The EC qi considered the maximum value stipulated by Brazilian CETESB norms (CETESB, 2009) (100 lS cm1), and the pi. Values
679
T.C. Lobato et al. / Journal of Hydrology 522 (2015) 674–683 Table 5 Factor analysis results to obtain the relevant parameters to take part in the construction of the Water Quality Index (WQI). Component
Initial Eigenvalues
1 2 3 4 5 6 7 8 9 10 11
Extraction sums of squared loadings
Total
% of variance
Cumulative %
Total
% of variance
Cumulative %
Total
% of variance
Cumulative %
4.578 1.436 1.178 1.007 0.756 0.562 0.436 0.399 0.285 0.195 0.169
41.622 13.055 10.709 9.157 6.869 5.105 3.96 3.626 2.59 1.771 1.536
41.622 54.677 65.387 74.544 81.413 86.518 90.478 94.104 96.694 98.464 100
4.578 1.436 1.178 1.007
41.622 13.055 10.709 9.157
41.622 54.677 65.387 74.544
4.1 1.548 1.295 1.257
37.27 14.07 11.774 11.43
37.27 51.34 63.114 74.544
Table 6 Commonalities, autovalues (F), percentage of factor variance and percentage of cumulative variance for the parameters selected to take part of the WQI construction. Values in bold indicate the factor scores used. Variables
Commonalities
Factors 1
Upstream water level S EC Fe NH4 NO3 PO4 TP TSS Turb. T Autovalues (F)
Rotation sums of squared loadings
2
3
4
0.804
0.002
0.696
0.017
0.565
0.695 0.849 0.712 0.64 0.752 0.664 0.784 0.756 0.699 0.845
0.802 0.212 0.838 0.196 0.132 0.757 0.865 0.812 0.79 0.218 4.578
0.168 0.021 0.072 0.773 0.492 0.24 0.1 0.282 0.123 0.162 1.436
0.01 0.054 0.062 0.001 0.663 0.138 0.135 0.124 0.235 0.86 1.178
0.154 0.895 0.021 0.061 0.227 0.115 0.087 0.03 0.066 0.177 1.007
37.27 37.27
14.07 51.34
11.774 63.114
11.43 74.544
% Variance % Cumulative variance
The resulting values of these quality curves indicate the water quality at the study site, with higher qi values indicating higher water quality than lower qi values. All quality curves are shown in Fig. 2. With these quality curve data, the weights, wi, to compose the WQI were calculated. The wi is the weight corresponding to the i-th parameter, a number between 0 and 1, assigned according to its importance to the overall water quality conformation. The sum of the weights wi equals 1 (Toledo and Nicolella, 2002). The wi values were calculated as displayed in Eq. (8):
PL n X F i Ail wi ¼ PL l¼1 ; wi ¼ 1 Pn l¼1 i¼1 F l Ai i¼1
ð8Þ
where: wi – weight of the WQI parameter; F – autovalues of the main components; Ai – explicability of the variable of the principal component; i – assumes values from 1 to n according to the variables selected in the model; l – assumes values from 1 to L according to the number of components and n – the n-th parameter selected to compose the index. 3. Results and discussion
Table 7 Calculations performed to obtain the qi curve for S. S
qi
3.1. Physico-chemical parameters and metal concentrations used for the construction of the QI and WQI
S 6 0.5 0.5 < S 6 2.4 2.4 < S 6 2.9 2.9 < S 6 3.1 3.2 3.3 3.4 S > 3.4
5 5 (eS) 4.6 (eS) 4.2 (eS) 3.85 (eS) 3.5 (eS) 3.17 (eS) 100
The parameters analyzed at the Tucuruí Hydroelectric Power Plant reservoir are displayed in Table 8. The observed variations are due to the different sampling stations environments, and will evidently be reflected in the QI and WQI, discussed in the next section. 3.2. Application of the Quality Indicator (QI)
were transformed in the range between 20 and 80, and then by applying Eq. (6).
qi ¼ 85 pi ; ði ¼ CEÞ
ð6Þ
For temperature we followed the SEMAD recommendation (SEMAD, 2005), using the equation qi = 93, since Brazil does not show much water temperature variations. The TSS qi was estimated by the method proposed by SEMAD (2005), where, when TSS values are higher than 500 units, TSS qi = 30, and when TSS values are lower than 500 units, the qi is calculated as follows:
qTSS ¼ 133:17 eð0:0027TSSÞ 53:17 eð0:0141TSSÞ þ ½ð6:2 e0:00462TSSÞ senð0:0146 TSSÞ
ð7Þ
Low QI values represent better water quality than higher QI values. Consequently, high values for PO4, CE, TP, TSS, NH4, NO3, Turb and Fe negatively affect water quality, causing higher QI values, since they are positively weighted. Likewise, higher water upstream levels, transparency and temperature contribute to better water quality, causing lower QI values, since they are negatively weighted. The upstream water levels used in the construction of the QI are directly correlated to the local hydrological cycle. The application of the QI on the study area during the four distinct phases of the local hydrological cycle throughout the study years showed no significant differences between the study years, with minimum and maximum of 0.96 and 3.52, respectively, meaning that water quality is not significantly modified during each cycle over the years. These results are displayed in Table 9.
680
T.C. Lobato et al. / Journal of Hydrology 522 (2015) 674–683
Fig. 2. Constructed quality curves for the relevant investigated parameters: Total P, NO3, Total Fe, NH4, TSS, turbidity, upstream water level, S, PO4 and EC.
Table 8 Results of the parameters analyzed at the Tucuruí Hydroelectric Power Plant reservoir at the 11 sampling stations from 2009 to 2012. Variables
N
Means
SD
CV (%)
Upstream level (m) S (m) T (°C) DO (mg L1) pH EC (lS cm1) Cl (mg L1) Fe (mg L1) Ca (mg L1) Turb. (NTU) Mg (mg L1) Na (mg L1) K (mg L1) NH4 (lg L1) NO3 (mg L1) PO4 (mg L1) TP (mg L1) TSS (mg L1) Cl a (mg L1)
180 180 180 180 180 180 136 180 180 180 180 180 180 180 180 179 180 180 180
68.26 ± 0.78 2.25 ± 0.15 29.99 ± 0.13 6.15 ± 0.10 7.14 ± 0.04 45.96 ± 1.42 2.82 ± 0.10 0.68 ± 0.08 4.15 ± 0.09 6.55 ± 1.09 1.51 ± 0.06 2.93 ± 0.19 1.50 ± 0.08 52.73 ± 9.34 33.38 ± 5.28 13.50 ± 1.11 24.48 ± 2.31 3.81 ± 0.729 6.00 ± 0.50
5.37 1.02 0.90 0.72 0.29 9.73 0.62 0.57 0.62 7.47 0.43 1.30 0.54 63.94 36.15 7.61 15.81 4.99 3.45
7.87 45.48 2.99 11.69 4.00 21.18 22.13 84.26 15.07 114.16 28.65 44.29 36.29 121.25 108.30 56.36 64.57 130.84 57.58
the means of the QI for each hydrological cycle. The ‘‘filling’’ cycle showed statistically significantly higher QI values, corresponding to December, January and February, while the ‘‘emptying’’ cycle presented statistically significantly lower QI values, corresponding to June, July and August, while the ‘‘empty’’ and ‘‘full’’ cycles were not statistically different. Since significant differences were observed for the filling and empty periods, it is possible to conclude that the dynamic nature of these periods is significantly different from the static nature of dry and empty situations of the reservoir. Fig. 3 shows the box-plots of the QI considering the study years and the hydrological cycle (HC). The results concerning the application of the QI on the water sampling stations to indicate water quality are displayed in Fig. 4. 3.3. Application of the Water Quality Indicator (WQI)
SD = Standard deviation; CV = Coefficient of variation.
As the construction of the QI did not take into account the hydrological cycle as one of its variables, the QI was calculated continuously throughout the four distinct hydrological cycles and divided according to these four cycles (empty, filling, full and emptying). Subsequently, Student’s t-test was applied to compare
The WQI is categorized as displayed in Table 10, following CETESB recommendations. The WQI was able to distinguish water quality at differently impacted sampling sites. The water quality at sampling stations MIP, MJV, ML, MP and M3 was rated as ‘‘Acceptable’’. This was expected, since each of these stations is near a fairly or heavily populated area. Sampling station MP, on the other hand, occupies a larger area, near the reservoir local fishing pole, which suffers from anthropogenic impacts, and was categorized between the ‘‘acceptable’’ and ‘‘poor’’ water quality. Sampling station MIP, located near the city of Itupiranga, which has been intensely urbanized since the late 80s, was also categorized in this water quality range. As was
Table 9 Application of the QI on the study area during the four distinct phases of the local hydrological cycle throughout the study years. Hydrological cycle
N
Minimum
Maximum
Means
SE
SD
Dry Filling Full Emptying
38 47 50 45
0.67 0.88 0.79 0.96
0.76 3.52 2.12 0.87
0.2138 0.5092 0.0594 0.4319
0.04691 0.15313 0.08340 0.05161
0.28918 104.984 0.58970 0.34621
SE = Standard error; SD = Standard deviation.
T.C. Lobato et al. / Journal of Hydrology 522 (2015) 674–683
681
Fig. 3. Box-plots of the QI (2) considering the study years (a) and (b) the hydrological cycle (HC).
Fig. 4. Results concerning the application of the QI on the water sampling stations of the Tucuruí Hydroelectric Power Plant reservoir to investigate water quality.
Table 10 Categories proposed by Cetesb (2005) for the WQI. Value
Classification
80–100
Excellent
52–79
Good
37–51
Acceptable
20–36
Bad
0–19
Poor
Color code
MJV, located in a region with two harbors, ’’Porto Novo’’ (Jacundá municipality) and ‘‘Port of Cologne’’ (municipality of Goianésia of Pará), while ML is located near the indigenous land of the Parakanas Indians, and was categorized mostly as ‘‘acceptable’’. M3, located in the central part of the reservoir, which depends on the water inflow of the reservoir, was also categorized between ‘‘acceptable’’ and ‘‘poor’’ water quality. Despite being located near a fairly populated area, the water quality as C2 and MBB sampling stations were classified as ‘‘Good’’, as was C1. MBL a natural wildlife preserve with no anthropogenic impacts, indicates the applicability of the WQI, since water quality was categorized as ‘‘Good’’ in this sampling station. M1, one of the most important monitoring stations, representing the water captured by electric generating units and the water released downstream of the dam, was also very close to qualifying as ‘‘Good’’. These results regarding the application of the WQI at the Tucuruí Hydroelectric Power Plant reservoir for each sampling site are displayed in Fig. 5, following the same color-code as displayed in Table 10. An overall analysis applying the WQI to the entire reservoir was also conducted. Figs. 6 and 7 were created using the free software gvSIG 1.9 with the implementation of the inverse distance weighted (IDW) interpolation algorithm, where the closest points have greater weight, i.e., when the distance increases their weight decreases. Results show that the water quality for the reservoir during the study years was categorized in the ‘‘Acceptable’’ range throughout the study years, indicating stability in the evaluated parameters of the reservoir (Table 11 and Fig. 7). Reflecting the QI results, the WQI results applied to the entire reservoir also demonstrate that the hydrological cycles directly
Fig. 5. Water Quality Index (WQI) for the Tucuruí Hydroelectric Power Plant reservoir sampling stations from 2009 to 2012.
682
T.C. Lobato et al. / Journal of Hydrology 522 (2015) 674–683
Fig. 6. WQI categorization for the Tucuruí Reservoir from 2009 to 2012.
Fig. 7. WQI categorization for the Tucuruí Reservoir from 2009 to 2012 during the regional hydrologic cycle (dry, filling, full and emptying).
Table 11 Means of the results obtained in the WQI sampling stations in the years 2009, 2010, 2011 and 2012. Stations 2009 2010 2011 2012 C1 46.86 49.00 52.87 50.94 C2 49.35 62.77 56.81 57.46 M1 45.95 51.25 41.23 55.20 M3 38.24 41.46 41.29 43.82 MBB 54.02 54.02 44.93 56.44 MBL 49.02 54.32 55.07 59.99 MIP 34.92 32.15 38.35 37.09 MJV 40.36 37.36 42.41 40.62 ML 40.80 44.15 41.05 42.67 MP 35.28 31.59 44.97 48.64 MR 45.78 53.12 56.67 55.23 Means 43.69 46.47 46.88 49.83 Deviation 6.19 9.99 7.03 7.75 CV 14.17 21.49 15.00 15.55 SD = Standard deviation; CV = Coefficient of variation
Means 49.92 56.60 48.41 41.20 52.35 54.60 35.63 40.19 42.17 40.12 52.70 46.72 7.08 15.15
SD 2.58 5.52 6.10 2.29 5.08 4.49 2.72 2.09 1.56 8.01 4.84 4.12 2.02 49.15
CV (%) 5.17 9.75 12.61 5.55 9.70 8.22 7.63 5.21 3.70 19.96 9.18 8.79 4.53 51.54
SD = Standard deviation; CV = Coefficient of variation.
affect water quality (Table 8 and Fig. 7), since during the flood period water quality was categorized as ‘‘Acceptable’’, close to ‘‘Poor’’, while during emptying the water quality was categorized as ‘‘Good’’, and no significant differences were observed between the empty and full periods (see Table 12).
Table 12 Means of the results obtained in the sampling stations in the years 2009, 2010, 2011 and 2012. Stations Empty Filling Full Emptying C1 50.63 49.59 46.43 55.12 C2 46.44 47.48 57.58 59.85 M1 59.09 38.75 45.33 61.61 M3 47.38 31.75 33.50 53.70 MBB 57.42 43.98 49.16 58.18 MBL 52.29 53.20 52.44 55.95 MIP 45.03 24.98 27.56 46.75 MJV 42.13 30.09 36.77 53.10 ML 44.48 29.21 36.76 61.38 MP 37.49 28.83 48.36 54.39 MR 49.03 49.58 50.25 59.34 Means 48.31 38.86 44.01 56.31 SD 6.37 10.25 9.13 4.39 CV 13.19 26.37 20.74 7.80 SD = Standard deviation; CV = Coefficient of variation
Means 50.44 52.84 51.20 41.58 52.19 53.47 36.08 40.52 42.96 42.27 52.05 46.87 6.22 13.27
SD 3.59 6.86 10.96 10.68 6.83 1.70 11.40 9.72 13.77 11.36 4.89 8.34 3.81 45.69
CV (%) 7.12 12.99 21.40 25.69 13.08 3.18 31.59 24.00 32.06 26.89 9.39 18.85 10.11 53.61
SD = Standard deviation; CV = Coefficient of variation.
4. Conclusions The novel quality indicator and water quality index created in the present study for the evaluation of the water quality of Tucuruí Hydroelectric Plant reservoir were constructed taking into account
T.C. Lobato et al. / Journal of Hydrology 522 (2015) 674–683
specific Amazon characteristics. The application of both indices to investigate water quality at the reservoir showed no significant differences for both between the study years. The QI was useful to determine seasonal differences in water quality due to the hydrological cycle of the area. The hydrological cycle was, in fact, shown by the indices to directly affect reservoir water quality. The WQI was also useful in observing anthropogenic impacts in the area, since water sampling stations in fairly populated areas ranged from ‘‘Poor’’ to ‘‘acceptable’’, while water quality in the environmental preservation area and near the dam was categorized as ‘‘Acceptable’’. These indices can, thus, be helpful in environmental monitoring situations that suffer well-defined hydrological cycles like those in tropical areas, in addition to being valuable tools in the detection of anthropogenic impacts. The statistical techniques applied in the construction of the QI and WQI can also be used to construct other specific indices in different geographical areas, taking into account the specificities for each area. Acknowledgement The main author would like to thank CAPES for the MSc scholarship. References Abaurrea, J., Asin, J., Cebrian, A.C., Garcia-Vera, M.A., 2011. Trend analysis of water quality series based on regression models with correlated errors. J. Hydrol. 400 (3–4), 341–352. Andrade, E.M., Araújo, L.F.P., Rosa, M.F., Gomes, R.B., Lobato, F.A.O., 2007. Fatores determinantes da qualidade das águas superficiais na Bacia do Alto Acaraú, Ceará, Brasil. Ciência Rural 37, 1791–1797. Aubert, A.H., Gascuel-Odoux, C., Merot, P., 2013. Annual hysteresis of water quality: a method to analyse the effect of intra- and inter-annual climatic conditions. J. Hydrol. 478, 29–39. Boyacioglu, H., 2006. Surface water quality assessment using factor analysis. Water SA 32 (3), 389–393. Braga, B., Porto, M., Tucci, C.E.M., 2006. Monitoramento de Quantidade e Qualidade das Águas, Águas Doces no Brasil: Capital Ecológico, Uso e Conservação. Escrituras, pp. 145–160. Brito, L.T.L., Silva, A.S., Srinivasa, V.S., Galvão, C.O., Gheyi, H.R., 2006. Uso de análise multivariada na classificação das fontes hídricas superficiais da bacia hidrográfica do salitre. Engenharia. Agrícola 26, 58–66. Brown, R.M., McLelland, N.J., Deininger, R.A., Tozer, R.G., 1970. A water quality index: do we dare? Water Sewage Works, 339–343. Carlson, R.E., 1977. A trophic state index for lakes. Limnol. Oceanogr. 22 (2), 361– 369. CETESB, 2009. Report – Significado ambiental e sanitário das variáveis de qualidade das águas e dos sedimentos e metodologias analíticas e de amostragem. 40. El-Iskandarani, M., Nasr, S., Okbah, M., Jensen, A., 2004. Principal components analysis for quality assessment of the Mediterranean coastal water of Egypt. Model., Measur. Control C 65, 69–83.
683
Fisch, G.F., Januário, M., Senna, R.C., 1990. Impacto ecológico em Tucuruí (PA): Climatologia. Acta Amazonica 20, 49–60. Guedes, H.A.S. et al., 2012. Aplicação da análise estatística multivariada no estudo da qualidade da água do Rio Pomba, MG. Revista Brasileira de Engenharia Agrícola e Ambiental 16, 558–563. Haase, J., Stringuini, M.H., Silva, M.L.B.C., Rodrigues, M.L.K., Koch, S.M.V., 2003. Qualidade das águas superficiais do litoral norte e médio do Rio Grande do Sul. 22 Congresso brasileiro de engenharia sanitária e ambiental, Joinville – SC. Anais. Rio de Janeiro, ABES, VI 049, p. 17. Hair, J.F., Anderson, R.E., Tathan, R.L., Black, W.C., 1998. Multivariate Data Analysis. Prentice Hall, New Jersey. Kazi, T.G. et al., 2009. Assessment of water quality of polluted lake using multivariate statistical techniques: a case study. Ecotoxicol. Environ. Saf. 72 (2), 301–309. Lai, Y.C., Tu, Y.T., Yang, C.P., Surampalli, R.Y., Kao, C.M., 2013. Development of a water quality modeling system for river pollution index and suspended solid loading evaluation. J. Hydrol. 478, 89–101. Laureano, J.S.F., Navar, J., 2002. An assessment of stream water quality of the Rio San Juan, Nuevo Leon, Mexico, 1995–1996. J. Environ. Qual. 31 (4), 1256–1265. Lewis, W.M., 2000. Basis for the protection and management of tropical lakes. Lakes Reserv. Res. Manage. 5, 35–48. Lucas, A.A.T., Folegatti, M.V., Duarte, S.N., 2010. Qualidade da água em uma microbacia hidrográfica do Rio Piracicaba, SP. Revista Brasileira de Engenharia Agrícola e Ambiental 14, 937–943. Merten, G.H., Minella, J.P., 2002. Qualidade da água em bacias hidrográficas rurais: um desafio atual para a sobrevivência futura. Agroecol. e Desenvol. Rur. Sustent. 3, 33–38. Petersen, W., Bertino, L., Callies, U., Zorita, E., 2001. Process identification by principal component analysis of river water-quality data. Ecol. Model. 138 (1– 3), 193–213. Rebouças, A.C., Braga, B., Tundisi, J.G., 2002. Águas doces no Brasil: capital ecológico, uso e conservação. Escrituras Editora, São Paulo. Selle, B., Schwientek, M., Lischeid, G., 2013. Understanding processes governing water quality in catchments using principal component scores. J. Hydrol. 486, 31–38. SEMAD, 2005. Secretaria de Estado do Meio Ambiente e Desenvolvimento Sustentável de Minas Gerais/Unidade de Coordenação Estadual. Sistema de Cálculo da Qualidade da Água (SCQA) – Estabelecimento das Equações do índice de Qualidade das Águas (IQA). Shah, V.G. et al., 2007. Comparisons of water quality parameters from diverse catchments during dry periods and following rain events. Water Res. 41 (16), 3655–3666. Singh, K.P., Malik, A., Mohan, D., Sinha, S., 2004. Multivariate statistical techniques for the evaluation of spatial and temporal variations in water quality of Gomti River (India) – a case study. Water Res. 38 (18), 3980–3992. Standard Methods, 2013. 1060 Collection and Preservation of Samples.