Abstract

The stormwater runoff carries different pollutants that can reduce the quality of receiving waters due to diffuse pollutant loads. This research was aimed at evaluating the concentration of pollutants in stormwater and the application of SWMM (Storm Water Management Model) to an urban catchment in Lake Paranoá watershed to carry out the simulation of flow discharge with the hydraulic model, and subsequently to estimate the loads conveyed to the lake in ordinary events of precipitation. This study was carried out based on rainfall and runoff monitoring during events. It was confirmed that this model's results fit well in simulation of this type of watershed, leading to high value of the Nash–Sutcliffe coefficient after calibration but, as expected, precipitation distribution is a very important factor for calibration. Concerning water quality, it was observed that the event mean concentration values of suspended solids and chemical oxygen demand were high, indicating that the diffuse pollution is an important source of pollution of the receiving waters. The monitoring and modelling of stormwater are essential to identify diffuse pollution discharge, in searching for a sustainable solution.

INTRODUCTION

The urbanization process may generate impacts on water resources which are directly related to the alteration of the natural conditions of the soil surfaces; for example, the increase of impervious area results in reduction of infiltration rate into the soil, and consequently reduction of groundwater recharge and increase of runoff and pollutant loads (House et al. 1993; Xu et al. 2017).

Beyond the hydrological impacts, watersheds with urban occupation may also suffer impacts in hydraulic terms. In addition to the increase of the volume drained superficially, there is also a significant increase in the velocity of surface runoff, in response to the smaller roughness of the paved surfaces and pipes that constitute the drainage networks (Du et al. 2012). The consequences are, among other effects, increased soil erosive potential and increased sediment transport that can carry aggregate pollutants and promote sedimentation in rivers and lakes (Righetto et al. 2009). Monitoring of stormwater flows is important to analyse the urban growth influence and the changes in land use patterns. It is known that the selection of the most appropriate method to measure the flow velocity depends on the characteristics and conditions of the channel, as well as the available resources, since there are several types of equipment and techniques that can be used (Barbosa et al. 2012).

Hydrological and water quality models are also effective tools to predict environmental impacts and to set water quality objectives, because receiving water bodies in developing countries are often highly polluted due to combined sewer overflows, wrong connections of sewers to the drainage system, and uncontrolled industrial discharges (Camacho et al. 2017). These impacts are also observed in Brazil, where the sewer system is considered completely separated; however, it is common for illegal sewage to be discharged into the drainage network. Thus, water quality monitoring is a very important step and it remains a very complex process due to a large number of factors to consider such as monitoring locations, selection of water quality parameters, monitoring frequency and identification of monitoring objectives (Behmel et al. 2016).

Model use has become more widespread due to scientific and technological developments, which today give easy access to a relatively high level of sophistication. This allows the improvement of hydrological modelling, so that in addition to the simulation of the water flow generation and conveyances other phenomena such as the transport of pollutants carried by the water are also incorporated. The hydrological models allow predictions of hydrological change responses, which can provide support for decision-making, water resources planning, and flood protection, for example (Beven 2012).

There are many models available, such as DR3M–QUAL, HSPF, SWMM, and STORM, that can help in dealing with urban water studies. One of the most used models is SWMM – Storm Water Management Model – a dynamic hydrological model that is very accessible because it is available for free download from USEPA (United States Environmental Protection Agency), and, since its creation in 1971, has undergone great advances and updates over time (Rossman 2009).

Drainage water flow monitoring produces information essential for diffuse pollution analysis, like calculation of loads of pollutants carried to the receiving water bodies. In addition, flow and precipitation data in urban basins can be used as a basis for the development and application of sustainable techniques and even the prevention of future impacts (Garcia & Paiva 2006; Jia et al. 2014). Thus, the scarcity of data, especially when dealing with small urban basins, becomes an obstacle to the application of hydrological models considering the need for reliable data for analysis of efficiency, calibration, and verification of the model.

The lack of field data available in Brazil also reflects a reduced number of researches about the urban drainage stormwater since the models must be calibrated to better fit the different areas in which they are used. Among the model parameters in the studies carried out, a great sensitivity of the models to the percentage of impervious area parameter has been verified. The percentage of impervious area has great influence in the peak flow simulation, as shown by Collodel (2009) in the studies developed in the Córrego do Gregório basin, located at Rio Grande do Sul, Brazil. Garcia & Paiva (2006) also verified the sensitivity of the model to the percentage of impervious area parameter, and pointed out the need for a good dataset on land use and occupation.

The study developed by Costa et al. (2017) in Brasilia-DF, concluded that urban stormwater is an important contribution to the Lake Paranoá pollution. Their results showed that diffuse pollution load is very high in the first events of the rainy season and it is higher at the beginning of each event. Modelling indicated that structural measures such as retention ponds and infiltration trenches could promote significant pollutant removal.

Lake Paranoá is an artificial reservoir and Riacho Fundo River is its main tributary and the most polluted. Starting in 2018, Lake Paranoá waters began to be used for domestic water supply and this fact increased the importance of better knowledge on nutrients loads and sediments produced by the Riacho Fundo urban area and discharged to Lake Paranoá. A model that can perform reliable simulations of flows and pollutant loads can be very important to the studies about non-point source pollution.

The main objective of this study is to model processes using the calibrated and verified SWMM to simulate the runoff, to analyse the impacts of flood waves on the outfall of the urban drainage network system of the Riacho Fundo area and to evaluate the drainage water quality and pollutant loads.

MATERIALS AND METHODS

Study area

The Riacho Fundo River catchment has about 225.5 km2 and is the most urbanized of the Lake Paranoá watershed. The studied area was, specifically, the urban area of Riacho Fundo I (Figure 1); the urban catchment drains an area of approximately 2.3 km2 with 50,000 inhabitants. The slope varies from 0 to 12%, and the elevations vary between 1,157.26 and 1,217.99 m.

Figure 1

Studied area and equipment location.

Figure 1

Studied area and equipment location.

In general, the drainage pipe network is composed of concrete pipes with diameters ranging between 400 and 1,500 mm. The total length of conduits in this area is 20,568 m, and 48.53 minutes is the estimated flow travel time. Land use and occupation classification showed that the impervious area of Riacho Fundo I is approximately 52% of the total area. In residential lots, it is very common that the impervious area covers the whole lot. The standard residential lot area is approximately 150 m2, with many houses/buildings with two to five floors. In general, solid wastes are properly disposed of in containers; however, there are some spots where debris and building materials such as sand and gravel can be found. Riacho Fundo I tends to produce high pollutant loads due to the dense urban occupation.

Monitoring

The flow and water quality monitoring station was installed at the discharge channel of the urban drainage network. Stormwater flow and water level in the discharge channel were monitored between November 2017 and February 2018. Also, two rain gauges (CBE: fire department; ETE: wastewater treatment plant) were installed and operated. The channel water levels recorded by a data logger were transformed to flow data using a stage–discharge rating curve. The flow was measured using an acoustic Doppler velocimeter. The results obtained in the monitoring step were used in the calibration and verification process of modelling. The summary of rainfall events used in this study is shown in Table 1.

Table 1

Summary of monitoring of water quality samples and rainfall events

Rainfall event (mm/dd/yyyy) Antecedent dry days Number of samples Sample start time (hh:mm) Rainfall amount (mm)
 
Rainfall duration (min)
 
ETE CBM ETE CBM 
11/25/2017 24 04:49 25.8 26.2 448 441 
12/05/2017 11:06 2.2 16.4 60 25 
12/12/2017 16:07 13.0 5.2 133 99 
01/03/2018 15:50 23.8 6.8 113 85 
01/05/2018 15:30 12.4 7.2 303 161 
01/08/2018 18:09 18.2 4.4 24 17 
01/14/2018 10:25 9.0 5.2 22 14 
01/15/2018 19a 14:12 26.0 3.8 84 39 
01/25/2018 04:10 5.8 8.4 65 72 
01/29/2018 19 16:30 52.80 50.2 152 186 
02/01/2018 13a 01:30 22.0 27.8 116 140 
02/03/2018 23 16:18 36.2 29.6 413 439 
Rainfall event (mm/dd/yyyy) Antecedent dry days Number of samples Sample start time (hh:mm) Rainfall amount (mm)
 
Rainfall duration (min)
 
ETE CBM ETE CBM 
11/25/2017 24 04:49 25.8 26.2 448 441 
12/05/2017 11:06 2.2 16.4 60 25 
12/12/2017 16:07 13.0 5.2 133 99 
01/03/2018 15:50 23.8 6.8 113 85 
01/05/2018 15:30 12.4 7.2 303 161 
01/08/2018 18:09 18.2 4.4 24 17 
01/14/2018 10:25 9.0 5.2 22 14 
01/15/2018 19a 14:12 26.0 3.8 84 39 
01/25/2018 04:10 5.8 8.4 65 72 
01/29/2018 19 16:30 52.80 50.2 152 186 
02/01/2018 13a 01:30 22.0 27.8 116 140 
02/03/2018 23 16:18 36.2 29.6 413 439 

aA very small volume was collected in one of the samples; therefore the analysis was made for one sample less than the total collected.

Water sampling was carried out using an ISCO 6712 autosampler. The time interval between samplings established was every 5 minutes after the water level had reached 40 cm in order to collect samples that could describe the characteristics of the water quality during a storm event. The parameters of water quality measured were total dissolved solids (TDS), total suspended solids (TSS), COD (chemical oxygen demand), nitrate, nitrite, total phosphorus and reactive phosphorus.

Modelling

Mathematical modelling was performed with SWMM, using PCSWMM software, with the objective of evaluating the runoff during storm events in the Riacho Fundo I urban center and to obtain parameters compatible with the land use and occupation of the region.

For the modelling process, the studied area was divided into subcatchments by the Thiessen polygons method, and CN (curve number) coefficients were attributed to each one. The urban drainage network data and orthophotos with 0.24 m spatial resolution were supplied by NOVACAP (Urbanizing Company of the New Capital of Brazil), topography by SEDUH (Secretary of State for Urban Development and Housing); and soil type by EMBRAPA (Brazilian Agricultural Research Corporation).

The simulations were carried out using the rainfall registered by the two rain gauges. The Soil Conservation Service method was adopted to evaluate infiltration, and flood wave propagation was performed using dynamic wave theory (Table 2).

Table 2

Model SWMM options chosen

Simulation options Chosen options 
Process model Rainfall/Runoff 
Flow routing 
Infiltration model Curve number 
Routing method Dynamic wave. Force main equation: Darcy–Weisbach  
Simulation period Each event 
Storage depression capacity Permeable areas: calibrated 
Impervious area: calibrated 
Manning Drainage network devices: 0.018 (calibrated) 
Impervious area: 0.012 (calibrated) 
Pervious area: 0.070 (calibrated) 
Simulation options Chosen options 
Process model Rainfall/Runoff 
Flow routing 
Infiltration model Curve number 
Routing method Dynamic wave. Force main equation: Darcy–Weisbach  
Simulation period Each event 
Storage depression capacity Permeable areas: calibrated 
Impervious area: calibrated 
Manning Drainage network devices: 0.018 (calibrated) 
Impervious area: 0.012 (calibrated) 
Pervious area: 0.070 (calibrated) 

During the calibration process, we tried to adjust the model parameters to the studied area. Thus, some changes were made in the depression storage on pervious area (Dstore perv), area of the subcatchment (area), width of flow in the subcatchment (width), curve number (CN) and roughness of the channels of the drainage network (roughness).

Runoff water quality analysis

To ensure sustainable development of the urban area, while protecting the water quality, an integrated analysis of runoff and diffuse pollution is required to evaluate the effect of urbanization and the possible mitigation solutions (Capodaglio et al. 2003).

Thus, the mean concentration per event (EMC) has been widely used to indicate the total pollutant load discharged into the receiving body, as reported by Novotny (2003). By this method of quantifying diffuse loads, the ratio between the mass of pollutant transported by volume of water over time (EMC) was evaluated using Equation (1). The EMC is considered a concise parameter that represents a very variable dataset, in addition to facilitating the comparison between different events and locations. In the absence of water quality data, EMC values reported in the literature can be used as a guide to estimate the likely range of pollution load (Chiew & McMahon 1999). 
formula
(1)
C(t) is the pollutant concentration at time t, Q (t) is the discharge flow rate at sampling time t; and dt is the time interval between the sampling.

RESULTS AND DISCUSSION

Rainfall events

From the rainfall-runoff monitoring, it was observed that very intense rainfalls did not occur in the observed period. For all events the return periods estimated by the regional intensity-duration-frequency equation were not greater than 1 year. In the monitoring step, there was great difficulty in the measurement of flows in the discharge channel, since the velocities are very high and, also, in the reach upstream of the monitoring station the channel is very steep with a sharp bend and steps, causing high velocities and instabilities. As also observed by Yang et al. (2017), finding an adequate location for the monitoring station is important for accurate measurement of discharge in sewer systems. The events used in the modelling process occurred on 24th December 2017, 3rd January 2018 and 15th January 2018 and their characteristics are shown in Table 3. Both rain gauges' data were used as input in the modelling, averaged by Thiessen method.

Table 3

Precipitation characteristics

  12/24/2017 01/03/2018 01/15/2018 
Maximum flow (L/s) 5,091.23 2,839.30 3,549.44 
Precipitation height (mm) ETE 18.8 23.8 26 
CBM 13.4 6.8 3.8 
Rainfall duration (min) ETE 27 113 84 
CBM 23 85 39 
  12/24/2017 01/03/2018 01/15/2018 
Maximum flow (L/s) 5,091.23 2,839.30 3,549.44 
Precipitation height (mm) ETE 18.8 23.8 26 
CBM 13.4 6.8 3.8 
Rainfall duration (min) ETE 27 113 84 
CBM 23 85 39 

Runoff modelling

As the first step, modelling was carried out using the model recommended parameters and assumptions. The simulations without any calibration of parameters, led to underestimated peak flow, as could be seen in Figure 2. The maximum simulated value for total outflow on the 24th December 2017 event was about 1,200 L/s and the maximum value observed was 5,091 L/s. For the 3rd January 2018 event the simulated peak flow was1,250 L/s and the observed value was 2,869 L/s, with 58.2% of estimated error.

Figure 2

24th December 2018 and 3rd January 2018 hydrograms simulated and observed.

Figure 2

24th December 2018 and 3rd January 2018 hydrograms simulated and observed.

The event used for the calibration process was the 15th January 2018 event, which presented characteristics such as rainfall temporal distribution similar to the prescribed design rainfall for the region. Before the calibration process, the peak flow simulated presented an error of about 48.9% and, after calibration, the error was about 4%. The comparison between the model response before and after calibration can be seen in Figure 3.

Figure 3

(a) Comparison between flow observed and simulated before calibration; and (b) comparison between flow observed and simulated after calibration.

Figure 3

(a) Comparison between flow observed and simulated before calibration; and (b) comparison between flow observed and simulated after calibration.

The simulated flow error was evaluated by Nash–Sutcliffe efficiency (NSE) and the coefficient of determination (R2). For the event of 15th January 2018, the NSE before calibration was 0.51 and R2 was 0.83, and these values after calibration process were NSE: 0.823 and R2: 0.84.

The uncertainty attributed to each parameter was defined according to SWMM manual recommendations; thus, the uncertainties adopted were: subcatchment area (15%), subcatchment width (50%), CN (10%), depression storage on pervious area (50%), conduit roughness (20%). The changes made for each one at the calibration process were: subcatchment area (12%), subcatchment width (0%), CN (9%), depression storage on pervious area (13%), conduit roughness (16.7%).

In the calibration procedure, it was possible to carry out sensitivity analysis and CN was identified as the most sensitive parameter, which is expected due to its direct relation with the soil infiltration. However, this is not a parameter that should be allowed large changes if the characteristics of the soil type and land use and occupation are carefully assigned in the input data (Beven 2012).

The sensitivity analysis also verified that two other parameters were very sensitives to changes, the depression storage on pervious area and the conduit roughness. Costa (2013) and Formiga et al. (2016), who developed studies modelling Brazilian urban drainage catchments with SWMM, obtained different results: both found the percentage of impervious area as the most sensitive parameter.

After the analysis involved in the calibration process, it is expected that the model responds more efficiently to other simulations carried out for the study area. The model verification process was carried out with the events of 24th December 2017 and 3rd January 2018, previously described.

The values for NSE and R2 in the 24th December 2017 rainfall event were −0.437 and 0.082, after calibration the values were 0.022 (NSE) and 0.416 (R2), and the 3rd January 2018 rainfall event presented even better values after calibration, since the values for NSE and R2 were −3.06 and 0.514 before the calibration process, after that the values were 0.641 (NSE) and 0.722 (R2). The model efficiency coefficient mean values indicated that, for both simulated rainfall events, modelling could be improved by calibration to give better representations of the hydrograms, as shown in Figure 4.

Figure 4

24th December 2018 and 3rd January 2018 hydrograms with calibrated model.

Figure 4

24th December 2018 and 3rd January 2018 hydrograms with calibrated model.

Even for the 24th December 2017 event, the NSE value was improved. As can be seen in Table 1, for the 15th January 2018 and 3rd January 2018 events the precipitation heights registered by ETE rain gauge were much higher than the precipitation height registered at CBM rain gauge, but for the 24th December 2017 event the values registered by both rain gauges were of the same magnitude, but temporally separated, making difficult the peak time calibration.

The spatial distribution of precipitation events is a very important characteristic for the SWMM simulations. Del Giudice & Padulano (2016), who modelled on SWMM and made the sensitivity analyses and calibration using a genetic algorithm for a drainage basin in western Naples, used only rainfall events that proved to be uniform within the basin, to avoid problems coming from spatial variation of rain depth. Krebs et al. (2014) analyzed the spatial resolution on SWMM and affirmed that the spatial variability did affect model performances.

Water quality

The behaviour of some pollutants in urban stormwaters was evaluated from water quality data obtained. In terms of concentration of TSS and TDS the values were between 4.5 and 522 mg/L and between 16 and 679 mg/L, respectively.

As discussed, EMC values may allow a comparison of the concentration of pollutant generated in different areas due to the characteristics of each location. The EMC for TDS and TSS calculated for the area in the study is shown in Figure 5. Righetto et al. (2017), working on an urban drainage basin in Natal-RN (northeast Brazil) with a 0.14 km2 area, obtained a maximum EMC value of 772.1 mg/L and a lowest of 57.9 mg/L, while Costa (2013) found EMC values ranging from 4.2 to 26.3 mg/L. The values observed in this study ranged from 22.9 to 201.2 mg/L.

Figure 5

(a) Total nitrogen EMC; (b) total phosphorus EMC; (c) reactive phosphorus EMC; (d) TDS EMC; (e) TSS EMC; (f) COD EMC.

Figure 5

(a) Total nitrogen EMC; (b) total phosphorus EMC; (c) reactive phosphorus EMC; (d) TDS EMC; (e) TSS EMC; (f) COD EMC.

In general, it is observed that the events with higher values of EMC are those with the greater antecedent dry period, like the event of 25th January 2018, and even the events of 12th December 2017 and 14th January 2018. The low EMC value for the 26th November 2017 event can also be explained by the time of sample collection, carried out a few hours after the beginning of the surface runoff due to problems in the automatic sampler, which started the sampling after two flood waves.

In the analysis of nutrients, the concentrations of different forms of nitrogen found in water samples give information about the stage of pollution, with recent pollution associated with ammonia, while nitrate indicates older pollution. In this case, the range of nitrate varied from 0.01 to 0.8 mg/L, nitrite from 0.003 to 0.123 mg/L and ammonia from 0.009 to 2.9 mg/L.

The total nitrogen EMC values observed at the studied area during the period did not reach 1 mg/L, EMC values much lower than those observed in the work developed by Costa (2013) in another urban area of the region, where contribution of domestic sewage was observed and the highest values of EMC were between 1.5 and 2.5 mg/L. However, the values observed by Costa (2013) are close to the total nitrogen EMC values observed at AR Riacho Fundo I where a smaller contribution of domestic sewage was detected.

It was also observed that the total phosphorus concentration varied from 0.01 to 0.63 mg/L and reactive phosphorus varied from 0.02 to 0.59 mg/L. Gomes (2014) monitored an experimental basin in Natal-RN urban area, where total phosphorus EMC ranged from 0.14 to 0.67 mg/L, values higher than those observed in this study, ranging from 0.01 to 0.24 mg/L (Figure 5(b)). Analyzing phosphorus loads carried by surface runoff, the influence of soil use and occupation is evidenced. Pinheiro & Deschamps (2008) verified that, in the watershed of the Fortuna stream in Timbó-SC, where the main uses are forest (69%) and agriculture (10%) and the other areas are divided between pisciculture and breeding cattle and pigs, orthophosphate (reactive phosphorus) loads ranged from 6.51 to 62.79 kg/(ha·yr), values very different from the loads observed in this work, ranging from 0.182 to 273.75 kg/(ha·yr).

In this sense, the loads generated in the urban area of RA Riacho Fundo I are closer to those observed by Costa (2013) at the Iate Clube (Brasília-DF) urban area of Brasilia, where total phosphorus loads varied from 0.04 to 1.140 kg/(ha·d) and reactive phosphorus from 0.016 to 0.456 kg/(ha·d). However, it is important to note that the Iate Clube area drainage system also receives contributions of irregular domestic sewage connections in addition to the loads from urban storm drainage and also that the typologies of land use and occupation are very different. Thus, the studied area tends to produce much higher diffuse pollutant loads of phosphorus than Brasilia, considering that at the point of sampling no domestic sewage was observed in the drainage channel during the dry period.

The COD analysis was used to quantify the concentration of organic matter in the water sample and the values found in this work varied from below detection by the used method to 196 mg/L.

Figure 5(f) shows COD EMC values, in which the event of 25th January 2018 presented higher COD loads (approximately 70 mg/L), which is a high value when compared to the EMC values of drainage waters from a rural area in the Taquarizinho River basin located in Mato Grosso do Sul, obtained by Oliveira (2007), where the highest EMC was 45.83 mg/L. On the other hand, the values are much lower than the largest EMC, approximately 330 mg/L, registered in the urban area of the Iate Clube where, as already mentioned, the drainage system receives a contribution from domestic sewage irregular connections (Costa 2013).

CONCLUSIONS

It is reasonable to state that SWMM can be very useful for the management of Lake Paranoá watershed concerning stormwater flows and diffuse pollution generated by the urban areas that cover half of the watershed area. The model simulated accurately the flow hydrographs, but pollutant loads simulation needs improvement.

It had been noted that in simulating real events SWMM works well when the rainfall spatial distribution was uniform throughout the watershed, This could be expected considering that in the simulations averaged rainfall values were adopted for the whole area. In this study, a good fit was achieved after calibration, leading to NSE and R2 as high as 0.82 for events more regularly distributed on the catchment.

In the simulations, the CN parameter was identified as the most sensitive parameter, which is reasonable due to its direct relation with soil infiltration. However, this is an input data parameter estimated through a geographic information system based on soil type and land use and, thus, it should not be subjected to large changes in the calibration process and only small adjustment, such as 10%, might be accepted.

Another important fact that must be noted is the complexity of stormwater drainage studies, and the lack of field data makes them even more difficult. Therefore, it is important to increase and keep monitoring systems along the urban watershed to collect field information and to support the hydrological and hydraulic models.

In general, the urban water quality analysis shows that nutrient and organic matter (COD) loads in urban stormwaters are, on average, lower than the values observed in areas where irregular sewage connections to the stormwater system were identified. The relatively high diffuse pollution loads in Riacho Fundo was not expected considering that no sewage discharge to the stormwater collection system was observed. This probably indicates that the characteristics of occupation and population in this area are leading to higher diffuse pollution generation than in Brasilia city, which can be considered consistent with the fact that in Riacho Fundo the percentage of impermeable area is higher and average population income is lower.

These results show the need for urban stormwater diffuse pollutant load reduction before discharging to the receiving water body, using best management practices to help in reducing the risk of Lake Paranoá re-eutrophication, after a water quality improvement achieved by 20 years of hard work and very high investment in sewage treatment.

ACKNOWLEDGEMENTS

This work was supported by Adasa, Caesb, NOVACAP, and FAP/DF Companhia Urbanizadora da Nova Capital do Brasil, Companhia de Saneamento Ambiental do Distrito Federal and Fundação de Apoio à Pesquisa do Distrito Federal – Governo do Distrito Federal. The authors are also grateful to ChiWater for PCSWMM license.

REFERENCES

REFERENCES
Barbosa
A. E.
,
Fernandes
J. N.
&
David
L. M.
2012
Key issues for sustainable urban stormwater management
.
Water Research
46
(
20
),
6787
6798
.
Behmel
S.
,
Damour
M.
&
Ludwig
R.
2016
Water quality monitoring strategies – a review and future perspectives
.
Science of the Total Environment
571
,
1312
1329
.
Beven
K. J.
2012
Rainfall-Runoff Modelling: The Primer
,
2nd edn
.
John Wiley & Sons
,
Thomson Digital, Noida, India
.
Camacho
L.
,
Navas
A.
&
Sierra
R.
2017
Dynamic water quality monitoring and modelling of polluted urban rivers. Study case: Tunjuelo River at Bogotá City – Colombia
. In:
Proceedings of 14th ICUD – International Conference on Urban Drainage. 14th IWA/IAHR International Conference on Urban Drainage (ICUD 2017)
, pp.
1584
1588
.
Chiew
F. H. S.
&
McMahon
T. A.
1999
Modelling runoff and diffuse pollution loads in urban areas
.
Water Science and Technology
39
(
12
),
241
248
.
Collodel
M. G.
2009
Aplicação do modelo hidrológico SWMM na avaliação de diferentes níveis de detalhamento da bacia hidrográfica submetida ao processo de transformação chuva-vazão (Application of the SWMM Hydrological Model in the Evaluation of Different Levels of Detail of the Hydrographic Basin Submitted to the Rainfall-Flow Transformation Process)
.
Master's thesis
,
University of São Paulo
,
São Carlos
,
Brazil
.
Costa
M. E. L.
2013
Monitoramento e modelagem de águas de drenagem urbana na bacia do lago Paranoá (Monitoring and Modeling of Urban Drainage Water in the Paranoá Lake Basin)
.
Master's thesis
,
University of Brasilia
,
Brasilia
,
Brazil
.
Costa
M. E. L.
,
Souza
F. P.
,
Roque
D. F. D.
&
Koide
S
, .
2017
Analysis of urban stormwater runoff quality and first flush phenomenon
. In:
Proceedings of 14th ICUD – International Conference on Urban Drainage. 14th IWA/IAHR International Conference on Urban Drainage (ICUD 2017)
, pp.
465
472
.
Garcia
J. I. B.
&
Paiva
E. M. C. D.
2006
Monitoramento hidrológico e modelagem da drenagem urbana da bacia do Arroio Cancela - RS (Hydrological monitoring and urban drainage modeling of the Arroio Cancela basin – RS)
.
Revista Brasileira de Recursos Hídricos, Porto Alegre
11
(
4
),
1
10
.
Gomes
K. M. F.
2014
Modelagem da poluição difusa em uma bacia de drenagem urbana (Diffuse Pollution Modelling in an Urban Drainage Basin)
.
Master's thesis
,
Federal University of Rio Grande do Norte
,
Natal
,
Brazil
.
House
M. A.
,
Ellis
J. B.
,
Herricks
E. E.
,
Hvitved-Jacobsen
T.
,
Seager
J.
,
Lijklema
L.
,
Aalderink
H.
&
Clifforde
I. T.
1993
Urban drainage – impacts on receiving water quality
.
Water Science and Technology
27
(
12
),
117
158
.
Jia
H.
,
Yao
H.
,
Tang
Y.
&
Shawlin
Y. U.
2014
LID-BMPS planning for urban runoff control and case study
.
Shuikexue Jinzhan/Advances in Water Science
25
(
2
),
260
267
.
Krebs
G.
,
Kokkonen
T.
,
Valtanen
M.
,
Setälä
H.
&
Koivusalo
H.
2014
Spatial resolution considerations for urban hydrological modelling
.
Journal of Hydrology
512
,
482
497
.
Novotny
V.
2003
Water Quality: Diffuse Pollution and Watershed Management
,
2nd edn
.
John Wiley & Sons
,
Hoboken, New Jersey, USA
.
Oliveira
H. A. R.
2007
Qualidade da drenagem pluvial rural: Rio Taquarizinho – MS (Quality of Rural Rainwater Drainage: Taquarizinho River – MS)
.
Master's thesis
,
Federal University of Mato Grosso do Sul
,
Campo Grande
,
Brazil
.
Righetto
A. M.
,
Moreira
L. F. F.
&
Sales
T. E. A.
2009
Capítulo 1 - Manejo de águas pluviais urbanas (Chapter 1 – Stormwater Management)
. In:
 Manejo de Águas Pluviais
(
Righetto
A. M.
, ed.).
ABES
,
Rio de Janeiro, Brazil
.
Righetto
A. M.
,
Gomes
K. M.
&
Freitas
R. F. S.
2017
Poluição difusa nas águas pluviais de uma bacia de drenagem urbana (Rainwater diffuse pollution of an urban drainage basin)
.
Revista de Engenharia Sanitária e Ambiental
22
(
6
),
1109
1120
.
Rossman
L. A.
2009
Storm Water Management Model User's Manual Version 5.0
.
U S Environmental Protection Agency
,
Cincinnati, OH, USA
.
Xu
P.
,
Gao
F.
,
He
J.
,
Ren
X.
&
Xi
W.
2017
Modelling and optimization of land use/land cover change in a developing urban catchment
.
Water Science and Technology
75
(
11
),
2527
2537
.
Yang
H.
,
Zhu
D. Z.
&
Liu
Y.
2017
Effect of Doppler flow meter position on discharge measurement in surcharged manholes
.
Water Science and Technology
77
(
3
),
647
654
.