Impact of climate change and urban development scenarios on waste water overflows from the combined sewage in Nantes, France

Transcription

1 Impact of climate change and urban development scenarios on waste water overflows from the combined sewage in Nantes, France V. Mahaut* 1,3, H. Andrieu 2, 3 and C. Joannis 2 1 École d architecture, Faculté de l aménagement, Université de Montréal, 2940, chemin de la Côte-Sainte-Catherine, Montréal (Québec) H3T 1B9, Canada 2 Department of Geotechnical Engineering, Environment and Risks, IFSTTAR, Route de Bouaye CS4, Bouguenais Cedex, Nantes, France 3 IRSTV, 1 rue de la Noë, Nantes *Corresponding author, valerie.mahaut@umontreal.ca ABSTRACT In Nantes (France) important overflowings from the combined sewer network towards the Erdre, a tributary of the Loire, occur regularly and contribute to the pollution of the water resource. The present paper intends to assess the impact of climate change and of the different urban development scenarios on overflowing volumes of the sanitary sewer network in the natural environment. It proposes and discusses a method using climate model outputs at the temporal and spatial resolutions available for users (daily step and a regional resolution) without an additional downscaling stage. The future evolution of overflowing is simulated by a hydrological model of the combined sewer system of Nantes at a daily time step. This model uses daily recorded data for six years. The preliminary results of this case study indicate that, in Nantes, climate change might result in a decrease of total overflows from the combined sewer network towards the Erdre. On the other hand, the increasing population associated to the urban development might accentuate overflowings. This study moderates different development scenarios in terms of sewer network management. KEYWORDS Climate change, conceptual model, overflow, sustainable urban development, waste water. INTRODUCTION Nantes-Métropole (France) is settled on the river Loire at the confluence of different tributaries: Erdre, Sèvre, Chézine. Important overflowings from the combined sewer network toward the Erdre are regularly recorded in winter and during long or intense rain events. These waste waters overflows contribute to the pollution of the water resource. Climate change and urban evolution could contribute to the increase of this vulnerability of water supply. This study deals with the assessment of the impact of climate change and of different urban development scenarios on the overflowing volumes of the sanitary sewer network in the natural environment. Studies that assessed the impact of climate change for the hydrology of urban areas (Semadeni-Davies et al., 2008; Olson et al., 2008; Franczyk and Chang, 2009) all used thin downscaling data from atmospheric models. The head line of the present research was to use climate model outputs at the temporal and spatial resolutions available for users Mahaut et al. 1

2 (Pagé C. et al., 2008) that is a daily step and a regional resolution (50km) downscaling. These were done without an additional downscaling stage that many of them cannot perform by themselves and which might appear to them as a questionable procedure. It establishes a combined sewer model from a six-year record of transfer flows and overflowing volumes and extrapolates this model with climate change data issued from a climate model (Figure 2). The results and the discussion about the assessment of climate change impact on a hydrology purpose in Nantes is the pretext to discuss on the relevance of the results from a low level of downscaling and desegregation model analysing the question of climate change. Section 2 presents the case study of Nantes. The methodology and the assumptions made to develop a model to simulate the behavior of waste water overflows from the combined sewage according to the weather are addressed in Section 3. Section 4 deals with different scenarios of climate change and urban development drawn up and combined. These are used, in Section 5, in this model in order to extrapolate the long-term evolution of this behavior. The conclusions will focus on the relevance of the methodology and the quality of the results provided by the study. CASE STUDY The case-study in Nantes (Figure 1) covers the territory between the Erdre and the Loire which is approximately 144 km² and represents different land uses: urbanized area, agriculture, prairies, marshlands and woodlands. The average ratio of impervious surfaces reaches less than 17%. A tot = km² α tot = 16.8 % Erdre Erdre Separated sewer network A s = km² α s = 16.0 % Broutelle Combined sewer network A c = 4.2 km² α c = 41.1 % Canal Duchesse Anne Q T,DA Q O,DA Q O,Br Loire Outlet of the catchment Duchesse Anne Loire Figure 1. Territory of Nantes-Métropole drained by the sewer network flowing to the outlet in Duchesse Anne. A is the surface and α is the impervious ratio of the area covered by each sewer system. Q T,DA is transfer flow rate just before the fixed weir discharges in Duchesse Anne. Q O,DA and Q O,Br are overflow rates at the fixed weir discharges in Duchesse Anne and Broutelle points. The densely urbanized part closed to the old centre (4.2 km² with an impervious ratio of 41.1%) is drained by a combined sewer system and the rest of the urbanized part of this territory is drained by a separated sewer system. Several monitored overflowing devices equip the sewer system. Transfer flow rates at the catchment outlet and total overflow rates are estimated from measurement devices. Daily averages are extracted for this study for the sixyear period from 2004 to 2009 and used to calibrate a combined sewer model. 2 Impact of climate change on waste water overflows, Nantes

3 METHODS The objectives of the model and the adopted approach are motivated by characteristics of the climate data resulting from the atmospheric model. This data is available at a daily step much longer than the time step required to reproduce the dynamics of overflows and that of the response time of urban catchments. The chosen method aims to reproduce the statistical distribution of the overflows over a long period of several years. To reach this objective, a two-part model is drawn up to simulate the daily average behavior of the observed overflows (Figure 2). 1 th Observed recorded data step Observed rainfall Observed Observed Observed PET total transfer flows overflows Model 1 Model 2 Rainfall PET f 1 [Rainfall P, PET] Total transfer flows f 2 [Total transfer flows] Overflow s 2 nd Climate change Urban step Rainfall, ETP scenarios Evolution scenarios Figure 2. Two-part model process in a two-step method. Firstly, parameters of models 1 and 2 were assessed thanks to observed recorded date. Secondly, evolution scenarios are assessed and implemented into the model. Model 1 to simulate total transfer flows Model 1 addresses the modeling of the total transfer flow at the catchment outlet according to rainfall (P, [mm]) and Potential Evapo-Transpiration (PET, [mm]). Model 1 is inspired from Reynaud et al. (2008) and Reynaud (2009). The total transfer flow at the catchment outlets regroups several components catchment including transfer flow at the outlet and overflows upstream, Q T,tot = Q T,DA + Q O,Br, is daily simulate following three hydrological components: a waste water part, Q ww ; a rainwater drainage part, Q rain ; a ground water drainage part, Q ground : Q T,tot = Q T,DA + Q O,Br = Q ww + Q rain + Q ground [m³] The rainwater drainage component is due to the runoff on impervious surfaces of the territory drained by the combined sewer network. It s supposed that connection defects from the rainwater system to waste water sewer system are insignificant in the area drained by the separated sewer network: Q rain = α c. A c. (P Pi c ), where P is the observed daily rain [mm] and Pi represents initial losses over the surface area drained by the combined sewer network. The ground water drainage component is due to the infiltration of soil water into the non waterproof pipes of the underground network (the sewer network on the territory drained by the combined sewer network and the waste water network on the territory drained by the separated sewer network): Q ground = (S k / S max ). δ ww. A tot, Mahaut et al. 3

4 where S k represents the soil moisture condition [mm] relative to its maximum value S max [mm] and indirectly the saturation level of groundwater. δ ww [mm] represents the drainage coefficient which regroups the influence of both the density of pipes on the territory drained by the sewer network and the efficiency of the drainage of soil water by the sewer pipes. S k is estimated by the following mass balance at a daily time step from S k-1, the daily previous weather (rain P k-1 [mm] and evapotranspiration PET k-1 [mm]) and the loss of groundwater drainage by the sewer network (waste water sewer and rain water sewer). S k = S k-1 + [(P k-1 -Pi tot ). (1-α tot )] [PET k-1. (1-α tot ) (S k-1 / S max )] - [δ tot. (S k-1 / S max )], where Pi tot is the initial loss rain due to the evaporation at the surface of the impervious material of the total territory and δ tot [mm] represents both the drainage coefficient of the underground network and the density of pipes on the total territory. The coefficients δ ww and δ tot are different because the average densities of pipes and the drainage coefficient are different on each territory. Flows recorded data. This model was calibrated thanks to six years of monitoring of the sewer network from 2004 to Only 5.7% of the data series was rejected by the city operator for dysfunction or maintenance causes. Meteorological recorded data. Recorded data of rainfall (P, [mm]) and evapotranspiration (PET, [mm]) at the Bouguenais meteorological airport station are available for the period analyzed. This point is approximately at 20 km on the south-west of the geographic centre of the Duchesse Anne catchment. Parameters calibration. The waste water component, Q ww, mainly depends on the demography and is an assumed constant. Its value is taken to be equal to the minimum of the observed Q T,tot corresponding on the summery period where the effect of drainage of the soil water is lower. Model 1 regroups seven parameters (Pi c, Pi tot, S max, S min, S initial, δ ww, δ tot ). Data from the all periods from 2004 to 2009 is used for calibrating the model. Results follow in Table 1 and in Figure 3. Discussion on the calibration results. The drainage coefficients δ ww and δ tot are very different. The separated sewer network drains a territory covered by a less dense urbanization with a high rate of periurban land use. The soil moisture condition (S k ) varies seasonally and is consistent to observations recorded in Nantes. The determination coefficient of simulated outflow is 76%, and the Nash coefficient is about 50% which represent correct simulation results. Table 1. Summary of observed or measured data, calibrated parameters and result coefficient. Observed and measured data Calibrated parameters Result coefficients A tot 14,412.4 ha Pi c mm Correlation coefficient α tot Pi tot mm 87.1 % A c ha S max 95.3 mm Determination coefficient α c S min 4.2 mm 75.9 % A s 13,992.0 ha S initial 95.3 mm Nash coefficient α s δ ww mm 48.8 % Q ww 14,309.0 m³ δ tot mm 4 Impact of climate change on waste water overflows, Nantes

5 01/01/ /03/ /05/ /07/ /09/ /11/ /01/ /03/ /05/ /07/ /09/ /11/ /01/ /03/ /05/ /07/ /09/ /11/ /01/ /03/ /05/ /07/ /09/ /11/ /01/ /03/ /05/ /07/ /09/ /11/ /01/ /03/ /05/ /07/ /09/ /11/2009 Total transit flows[m3/j] 12 th International Conference on Urban Drainage, Porto Alegre/Brazil, September Observed flows Modelized flows Figure 3. Comparison between observed and modelled totals transit flows during a six years monitoring period of the sewer network and climatic recording (2004 to 2009). Model 2 to simulate total overflows Model 2 concerns the analysis of behavior on the total overflows in the catchment according to total transfer flows. Daily overflows (Q O,tot = Q O,DA + Q O,Br ) and daily transfer flows (Q T,tot = Q T,DA + Q O,Br ) are correlated (Figure 4) but this relation bears a high degree of unpredictable variability. This is due partly to the less accurate daily step values imposed by the characteristics of the climate model. The daily step is not representative of the dynamics processes on the catchment. The simulation of daily overflows from total daily flow rates introduces a statistical component which aims to reproduce the random variability illustrated in Figure 4. The statistical distribution is cut into four parts to benefit, in each part, from a specific approximation by second degrees equations and from appropriate variability of the random. The parameters of the random component are selected to make the histogram of modeled overflow as close as possible to the histogram on observed overflows characterized by their number and their magnitude. Figure 4. On the left: relation between daily overflows, Q O,tot and daily transfer flows, Q T,tot. On the right: histogram distribution of the overflow days depending on the overflow volume. LONG-TERM EVOLUTION SCENARIOS The model representing the overflowing of the sewer system is forced with climate change scenarios combined with urban development scenarios in order to assess the influence of these two factors on the evolution of the overflowing. Climate change scenarios Climate change data is provided by the Arpege-climate model of Meteo-France (Pagé C. et al., 2008; Pagé C., 2008; Terray and Boé, 2008). The data consists of time series of climate parameters (wind, temperature, rain, evapotranspiration and atmospheric pressure) with a spatial resolution of 50 km at a daily step for a period from 1950 to The chosen grid point includes the city of Nantes. Each time series depends on the initial conditions of CO 2 concentration in the atmosphere and on IPCC scenarios of evolution of this concentration. Mahaut et al. 5

6 Figure 5 illustrates the temporal evolution of yearly rainfall and PET (Potential Evapo- Transpiration) from 1950 to Yearly rain [mm] averaged on decades Yearly PET [mm] averaged on decades Figure 5. Comparison of climate scenarios: evolution of yearly rain and PET parameters averaged on decades for IPCC scenarios simulated by Arpege (discontinuous lines) and for observed dates (black continuous line). The period presents the same data for climate scenarios (black discontinuous line) and different evolutions depending on the IPCC scenarios (B1 in green discontinuous line; A1B in blue discontinuous line; A2 in red continuous line). Discussion on the climate change comparison. Even if yearly rain presents a large variation for a decade to another, average trend of rain should decrease 10 to 20% from 1990 to The scenarios A1B and A2 do not present significant difference in the evolution of the yearly rain. The evolution of the Potential Evapo-Transpiration should present 12 to 30% increase on the same period. Urban development scenarios Different urban scenarios were also assessed and implemented in the model. It is assumed in this study that the following parameters - Q ww, α c, α s, δ ww and δ tot - depend directly on the evolution of the population (the waste water volume) or on the evolution of the urban settlement (the impervious ratios and the drainage coefficients into the soil). The population of Nantes increased of 10% from 1990 to 2000 (nearly 1% per year). The urbanized surfaces doubled from 1950 to 2000 (1.4% per year). These evolutions are not homogeneous in terms of time and space on the territory of Nantes-Métropole. They are not easily predictable for the next 90 years on the territory covered by the study case. Nantes-Métropole is waiting for a continuous increasing of population and urbanized surfaces for the next decade but it is assumed that the territory covered by this study is less under pressure than other parts of the city and that the increasing will decelerate by half from the last half of the century. The proposed scenarios are summarized in the Table 2. Table 2. Summary of the scenarios. The first and the second numbers in each box correspond to the annual increase ratio from 2010 to 2050 and from 2051 to Parameters Scenarios Q ww 14,309 m³ - +1%/+.5% +.5%/+.25% +.5%/+.25% +.5%/+.25% +.5%/+.25% +.5%/+.25% α c %/+.1% α s %/+.25% +.25%/+.1% +.5%/+.25% - δ ww mm %/+.25% +.25%/+.1% - - δ tot mm %/+.25% +.25%/+.1% Impact of climate change on waste water overflows, Nantes

7 The scenario 1 takes into account the climate change outputs without any other evolution of the parameters. The scenarios 2 and 3 add the impact of the population evolution via the Q ww parameter. The scenarios 4, 5, 6 and 7 add the impact of the evolution of the urban settlement with a sprawl settlement (scenario 4 and 5: an increase of both the impervious and the drainage areas) and with a densification of the existing urban settlement (scenarios 6 and 7: an increase of the impervious areas without an increase of the drainage areas). RESULTS AND DISCUSSION The impact of the different urban scenarios on the overflowing is compared on Figure 6 (left) for the climate scenario A1B. The comparison between the urban scenarios and the climate change scenarios is made at Figure 6 (right). Figure 6. On the left: Total overflows Q O, tot for climate scenario A1B and for the seven urban scenarios defined on Table 2. On the right: Total overflows for climate scenario B1, A1B and A2 and for the urban scenarios 4 (sprawl development) and 6 (densification development). Comparison of scenarios 1, 2 and 3: if the constant term Q ww in the equation of model 1 represents the daily waste water volume and is assumed proportional to the demography, the increase of the population in Nantes is the prevailing parameter on the impact of the overflowing behavior. The climate change would decrease total overflows from the combined sewer network toward the Erdre (scenario 1) but the increase of this base flow (scenario 2 and 3) accentuates overflowing and can invert the tendency (scenario 2). Comparison of scenarios 3, 4 and 5: the doubling of the increase ratio of both the impervious and the drainage areas causes a 10% increase of the overflowing. Comparison of scenarios 4 and 6: a densification development (scenario 6) of the existing urban settlement allows to save nearly 30% of overflowing in comparison to a sprawl development (scenario 4). Urban development avoiding laying of pipes for rain and waste water are advised. Comparison of scenarios 6 and 7. The increase of the urban development assessed in the scenarios 6 and 7 are respectively of 39% (on the territory of the separated sewer network) and 14% (on the territory of the combined sewer network). It represents a surface of 5500 ha and 59 ha, that is to say a 10-² ratio. Despite this ratio, the impact of the increase of impervious areas in the territory drained by the separated sewer network is prevailing (16% increase) on the overflows into the Erdre. In term of impact on overflowing volume into the environment, urban development is advised within the territory with separated sewer network to avoid waste water overflowing during rain periods. Mahaut et al. 7

8 Comparison of climate scenarios and urban scenarios. The distances between the more optimistic and the more pessimistic scenarios for each urban scenario are of the same order than, for example, the distance between the sprawl scenario (4) and the densification scenario (6). The urban development is not less important than the climate change on the environmental impacts. CONCLUSIONS This study shows that, in Nantes, climate change would decrease total overflows from the combined sewer network toward the Erdre. On the other hand, the increase of the demography, the impervious areas and the drainage sewer system still accentuate overflowing into the environment. The results suggest recommendations in terms of urban development to mitigate the impact of urban development of the metropolis. The adopted approach motivated by characteristics of the climate data resulting from the atmospheric model without an additional downscaling stage than the daily step gives tendencies of evolution. These tendencies are average but can be compared to one other to distinguish the prevailing parameters. However, the extrapolation of the model parameters to urban parameters needs to be refined and checked. The assumption made on the base flow of the transfer flow, for example, is open to doubt because it could present a base drainage flow component as well as the waste water component. The importance of this part of drainage in the base transfer flow could have a significant impact on the results and discussion about the impact of the evolution scenario of the drainage parameters. ACKNOWLEDGEMENT The authors thank Nantes-Métropole and Météo-France for providing the hydrological data of the sewer network and the meteorological data of Nantes. They thank the Région Pays de la Loire that has financed this research. REFERENCES Raynaud O. (2009). Un modèle hydrologique conceptuel pour l'évaluation des rejets de réseaux séparatifs d'eaux usées par temps de pluie, PhD. thesis, Ecole Centrale de Nantes, France. Raynaud O., Joannis C., Schoefs F., Billard F. (2008). A model-based assessment of infiltration and inflow in the scope of controlling separate sanitary overflows at pumping stations. Proc. 10 th Int. Conf. on Urban Drainage, Edinburgh, Scotland, UK, Pagé C., Terray L., Boé J. (2008). Projections climatiques à échelle fine sur la France pour le 21 ème siècle: les scénarii SCRATCH08, Climate Modelling and Global Change, TR/CMGC/08/64, CERFACS Pagé C. (2008). Format des données SAFRAN et scenarios climatiques désagrégés au CERFACS, Climate Modelling and Global Change, TR/CMGC/08/27, CERFACS Terray L., Boé J. (2008). Weather regimes and downscalling, La Houille Blanche , Semadeni-Davies A., Hernebring C., Svensson G., Gustafsson L.-G. (2008) The impact of climate change and urbanisation on drainage in Helsingborg, Sweden: combined sewer system, Journal of Hydrology, 350, Olson J., Berggren M., Viklander M. (2008). Applying climate model precipitation scenarios for urban hydrological assessment : A case study in Kalmar Cityn Sweden, Atmospheric Research, 92, Franczyk J. and Chang H. (2009). The effects of climate change and urbanization on the runoff of the Rock Creek basin in the Portland metropolitan area, Oregon, USA, Hydrological Processes, 23, Impact of climate change on waste water overflows, Nantes