1 INCORPORATION OF CONTINENTAL AND URBAN RUN-OFF INTO A COASTAL CIRCULATION MODEL : APPLICATION TO THE CATALAN COAST

A 3D hydrodynamical model has been set up to incorp orate the continental and urban run-off into the Ca talan Coastal waters. Particular attention was paid to introducin g correctly the freshwater plumes and attention was also paid to determinate the influence of the land discharge pro file with regard to the distributed continental run -off. The model domain includes a small part of the Catalan Coast w here the combination of local land topography with torrential rainfall caused considerable local runoff on a shor t pe iod of time with a large impact on the receivi ng coastal waters. The Regional Ocean Modeling System (ROMS) simulatio ns were used to examine the dispersal to a freshwat er delivery from two relevant event; a low river disch arge typical of mean conditions during April 2011 a nd a high discharge representative of the storm event during March 2011 are considered. We have observed the plu me responses to an abrupt change in river discharge. D uring the mean conditions, low salinity water is co n entrated around the rivers mouth while during the flood even t, the plume spread offshore in the direction of ri ver water outflow and turned downstream close to the coast. The diffe renc s between a simulation including the river out fl w as a land forcing and a simulation including river and urban ru off as a land forcing suggested that the urban r unoff plays an important role in the spreading and shape of the ri ver plume.


INTRODUCTION
Coastal seas are ecologically and economically among the most productive areas of the world ocean and have been the subject of many international agreements and conventions looking to preserve and exploit them in a sustainable manner.Additionally, more than 120 million people live in and "impact" on Europe's coastal regions and their quality of life is affected by the environmental status of its regional seas.Most studies on freshwater influence into the coastal sea have focused on estuarine environments (e.g., Wang et al. 2011;Liu et al. 2009;MacCready et al. 2009;Warner et al. 2005;Hetland R.D., 2004) but continental/urban runoff has been rarely considered in coastal circulation although an adequate incorporation of the land boundary fluxes plays a crucial role for assessment and prediction of circulation in coastal waters.
This paper discusses the implementation of the land boundary fluxes for coastal circulation modeling including river and urban runoff in the Catalan coast.In the classical model of buoyant outflow into coastal oceans, the freshwater discharged into the coastal ocean from rivers and runoff is often observed to be incorporated into a narrow coastal current that is typically a few internal Rossby radii wide and that rapidly transports freshwater downshelf (Garvine 1999).However, more recent theoretical, modeling, and laboratory studies (Avicola and Huq 2003a;Fong and Geyer 2002;Nof and Pichevin 2001) revealed a tendency for the formation of a recirculating bulge structure in the vicinity of the outflow in the absence of wind and alongshore current.Different factors and forcing modify the pathways of the freshwater plume (Zhang et al. 2009): outflow angle (Avicola and Huq 2003a,b;Garvine 1999), wind forcing (Fong and Geyer 2001;Berdeal et al. 2002;Lentz and Chapman 2004;Whitney et al. 2005;Choi and Wilkin 2007;MacCready et al. 2009;Xia et al. 2010;Schiller et al. 2011), ambient current (Fong and Geyer 2002;Berdeal et al. 2002;Hickey et al. 2005), tides, and local topography (Schiller et al. 2011).These factors and forcing modify the pathways of the river plume and can make it similar to the classical theory.
Moreover, the details of freshwater dispersal processes can affect ocean stratification (Garvine and Whitney 2006).In this study we use a numerical model to incorporate the land boundary fluxes into a coastal circulation model in the Catalan coast zone during spring 2011 (from March 1, 2011to April 31, 2011) in the framework of the Field AC project (www.field_ac.eu).The Catalan Coast was chosen for this study because local (torrential) rainfall is considered to have a high impact on the quality of coastal waters and it is thus important to be able to predict the land discharge adequately such that it can be incorporated into the coastal scale oceanographic models (Keupers et al. 2011).The study area, as many parts of the Mediterranean environment, is under strong anthropogenic pressure and requires in-depth knowledge of the marine environment for appropriate management (Grifoll et al. 2012).
The study region is adjacent to a relative narrow shelf km in the current study area and 60 km in the south margin near the Ebro Delta.The shelf break is located at the 200 m isobath and the averaged slope is of the order of 10 study exhibits micro-tidal fluctuations, freshwater input, air and variable bathymetry all of which polarized in the along-shelf direction due to the coastal boundary constrain fluctuations are basically controlled by wind stress gradients at synoptic time scales The goal of this paper is to relate land boundary fluxes to different plume discharges.In particular, this paper will compare the relative importance of different freshwater discharges of the plume and the role of land boundary fluxes in determining the structure of the plume.
The outline of this paper is as follows: section 2 introduces the modelling tools, section 3 sho the verification.In section 4, the are summarized in section 5.

Coastal Circulation Model
The ocean model used is the Regional Ocean Modeling System surface, terrain-following numerical Navier-Stokes equations (RANS) using the hydrostatic and Boussinesq approximations (Shchepetkin andMcWilliams 2005, 2009;Haidvogel et al. 2008).ROMS uses finite horizontal curvilinear Arakawa C grid and Momentum and scalar advection and diffusive processes are solved using transport equations equation of state computes the density field that accounts sediment contributions.ROMS provides a flexible structure that allows multiple the model components such as several boundary conditions, bottom-and surface nutrient-phytoplankton-zooplankton data assimilation.The code is written in Fortran90 and runs in serial mode on a single processor, or uses either shared-or distributed processors (Warner et al. 2010).ROMS is open source and freely

Model domain and grid
The Catalan coast is located in the north N and longitude 0º 45' E to 3º 15' E ( Coastal grid (Figure 1, right panel size of approximately 250 m (fine enough to capture the relevant coastal details and oceanographic processes) and 20 sigma vertical levels.The Coastal grid is located at the latitude 41 and longitude 1.89º E to 2.62º E

COASTAL ENGINEERING 2012
region is adjacent to a relative narrow shelf.The Catalan shelf widths are between 20 area and 60 km in the south margin near the Ebro Delta.The shelf break is 00 m isobath and the averaged slope is of the order of 10 -2 .The Catalan Coast under tidal fluctuations, freshwater input, air-sea exchange, large-scale shelf circulation, of which influence circulation processes.The coastal circulation is strongly shelf direction due to the coastal boundary constraint.The along fluctuations are basically controlled by wind stress at short time scales and by remote pressure synoptic time scales (Grifoll et al. 2012).The goal of this paper is to relate land boundary fluxes to different plume discharges.In particular, this paper will compare the relative importance of different freshwater discharges of the plume and the of land boundary fluxes in determining the structure of the plume.The outline of this paper is as follows: section 2 introduces the modelling tools, section 3 sho the verification.In section 4, the simulated salinity dynamics are presented and discus is the Regional Ocean Modeling System (ROMS), a general class of free following numerical models that solve the three dimensional Reynolds equations (RANS) using the hydrostatic and Boussinesq approximations (Shchepetkin andMcWilliams 2005, 2009;Haidvogel et al. 2008).ROMS uses finite-difference approximations on a horizontal curvilinear Arakawa C grid and on a vertical stretched terrain-following coordinate.
advection and diffusive processes are solved using transport equations equation of state computes the density field that accounts for temperature, salinity, and contributions.ROMS provides a flexible structure that allows multiple choices for many of the model components such as several options for advection schemes, turbulence models, lateral and surface-boundary layer submodels, air-sea fluxes, surface drifters, a zooplankton model, and an adjoint model for computing model inverses and data assimilation.The code is written in Fortran90 and runs in serial mode on a single processor, or or distributed-memory architectures (OpenMP or MPI) to run on multiple (Warner et al. 2010).ROMS is open source and freely available (www.myroms.org).located in the north-western Mediterranean at the latitude 40º 45' N to 42º 25' 3º 15' E (Figure 1. Catalan Coast, panel left).This study has used the panel).The Coastal grid has 152x207 grid cells, adopts a horizontal grid size of approximately 250 m (fine enough to capture the relevant coastal details and oceanographic vertical levels.The Coastal grid is located at the latitude 41 E (Figure 1. Catalan Coast and coastal grid, right panel

. Catalan Coast and coastal grid
Catalan shelf widths are between 20 area and 60 km in the south margin near the Ebro Delta.The shelf break is .The Catalan Coast under scale shelf circulation, .The coastal circulation is strongly .The along-shelf current short time scales and by remote pressure The goal of this paper is to relate land boundary fluxes to different plume discharges.In particular, this paper will compare the relative importance of different freshwater discharges of the plume and the The outline of this paper is as follows: section 2 introduces the modelling tools, section 3 shows presented and discussed.The results (ROMS), a general class of free Reynolds-averaged equations (RANS) using the hydrostatic and Boussinesq approximations (Shchepetkin difference approximations on a following coordinate.advection and diffusive processes are solved using transport equations and an for temperature, salinity, and suspendedchoices for many of turbulence models, lateral sea fluxes, surface drifters, a model, and an adjoint model for computing model inverses and data assimilation.The code is written in Fortran90 and runs in serial mode on a single processor, or memory architectures (OpenMP or MPI) to run on multiple available (www.myroms.org).
terranean at the latitude 40º 45' N to 42º 25' ).This study has used the has 152x207 grid cells, adopts a horizontal grid size of approximately 250 m (fine enough to capture the relevant coastal details and oceanographic vertical levels.The Coastal grid is located at the latitude 41.02º N to 41.55º N panel).

Model grid
For this work we have used nested the boundary conditions with enough resolution to reproduce the coastal features.on structured grid models where there is a spatially large grid that encompasses the entire domain of interest.The Regional domain ( enough to cover the entire domain of interest and has a resolution to adequately dynamics.We use a grid that covers the Gulf of Leon and the Balearic Island model a region with increased grid resolution domain covers the Catalan Coast region providing a refine panel).Finally, within the SHECAT defined.This is called the Coastal Barcelona City area (between a bit River mouth) (Figure 2, right panel)

Figure 2. Nested increasing-resolution domains
The one-way nesting is achieved ( forcing) as follows.The Regional which includes several barotropic steps.The prognostic variables of the coarse regional interpolated in space to fill boundary arrays for the SHECAT boundary arrays for the finer Coastal increased spatial resolution and physics are being resolved on both grids with consistent

Initial conditions
Initial conditions for the ocean model fields of salinity, and temperature as well as lateral boundary conditions for currents, sali are obtained from MyOcean products (

Atmospheric forcing
The rainfall forecast that is land boundary fluxes is provided by the Barcelona Super Computing Center (BSC).It is generated by the air quality model CALIOPE ( accumulated rainfall per hour in each grid Forcing winds have been generated by the same air quality model CALIOPE information of the wind velocity per hour i

Land forcing
The run-off data file that has been used as input file for conceptual river and urban runoff (Keupers et al. 2011).The urban runoff combined sewer overflows (CSOs runoff model provides hourly flow time series (m

Initial and boundary conditions
For this work we have used nested increasing-resolution domains in order to introduce the initial and enough resolution to reproduce the coastal features.The method is based on structured grid models where there is a spatially large grid that encompasses the entire domain of (which is the same as the MyOcean Mediterranean Sea cover the entire domain of interest and has a resolution to adequately resolve basin scale dynamics.We use a grid that covers the entire northwest part of the Mediterranean Coast Leon and the Balearic Islands on a 2.3 km scale (Figure 2, left panel).Within the Regional increased grid resolution has been defined.This model is called SHECAT and the domain covers the Catalan Coast region providing a refined grid on a 1 km resolution ithin the SHECAT model another region with increased grid resolution Coastal model and is located offshore of the northeastern districts of a bit north of the Bèsos River mouth and a bit south of the Llobregat panel).
resolution domains way nesting is achieved (for the initial conditions, the boundary c ) as follows.The Regional model (MyOcean data) takes one complete baroclinic time step, which includes several barotropic steps.The prognostic variables of the coarse regional interpolated in space to fill boundary arrays for the SHECAT model which in turn provides the boundary arrays for the finer Coastal model.This allows the region of interest to be resolved with an and to be driven with forcings that are dynamically consistent ng resolved on both grids with consistent formulations.
Initial conditions for the ocean model fields of currents, depth-averaged currents, water level, as well as lateral boundary conditions for currents, salinity, MyOcean products (http://www.myocean.eu/).The method is based on structured grid models where there is a spatially large grid that encompasses the entire domain of Mediterranean Sea grid) is large resolve basin scale Coast including the Within the Regional is called SHECAT and the d grid on a 1 km resolution (Figure 2, central increased grid resolution has been and is located offshore of the northeastern districts of south of the Llobregat oundary conditions and the takes one complete baroclinic time step, which includes several barotropic steps.The prognostic variables of the coarse regional model are then which in turn provides the of interest to be resolved with an that are dynamically consistent because the averaged currents, water level, nity, and temperature models to calculate the land boundary fluxes is provided by the Barcelona Super Computing Center (BSC).It is generated by ) and provides information on the resolution).been generated by the same air quality model CALIOPE and provide coastal model is provided by the s that have been calibrated as part of the FIELD_AC project (m 3 /s) for the four netcdf format).The river /s) for both the Besòs and the Llobregat river.

COMPARISONS WITH OBSERVATIONS
In this section, selected aspects of the simulated results are compared with observations. of the measurements correspond to a field campaign conducted during March and April 2011 over the Catalan inner-shelf (around Bèsos river mouth, FIELD_AC project.Several data are available for model comparison (the data set consists of water velocity time-series from three ADCP, wind data (www.xiom.cat)and several CTDs), bu fields with the salinity measurements from field campaign are located inside the Coastal grid between latitudes longitudes 2º 19'E to 2º 30'E collected and another 42 CTD profiles were collected on the 11 obtained from a monitoring station upstream of the river mouth ( corresponds to "post-rain conditions" associated with a short and intense m 3 s -1 (mean flow conditions are 4 m hours) and very intense storm events, typical of the Mediterranean climate, in small river mountainous drainage basins (Bourrin et al. 2008).associated with a normal Bèsos river discharge.the signal of Besòs River freshwater the rivers mouths.However when the plume spreaded offshore in the direction of river water outflow and to the coast.More detailed analysis indicated that the water exchange rate between the water aroun river mouth and the outside area was proportional to the river discharge.

COMPARISONS WITH OBSERVATIONS
In this section, selected aspects of the simulated results are compared with observations. of the measurements correspond to a field campaign conducted during March and April 2011 over the d Bèsos river mouth, Figure 3, panel left) in the framework of the Several data are available for model comparison (the data set consists of water three ADCP, wind data (www.pontdelpetroli.org),wave data from buoy and several CTDs), but in this paper only the comparison of the with the salinity measurements from CTD measurements are shown.The measurements of the located inside the Coastal grid between latitudes 41º 35'N (Figure 3, right panel).On the 17 th of March 43 CTD profiles were and another 42 CTD profiles were collected on the 11 th of April.Besòs River obtained from a monitoring station upstream of the river mouth (www.aca.cat).The March 17 rain conditions" associated with a short and intense Besòs river discharge of 50 n flow conditions are 4 m 3 s -1 ).These "flash-flood" conditions occur during short (duration storm events, typical of the Mediterranean climate, in small river mountainous basins (Bourrin et al. 2008).The April 11 th measurements correspond to "normal conditions" associated with a normal Bèsos river discharge.

. Field Campaign domain and CTDs position located into the Coastal model domain
salinity during the two sampling periods (March 17 th and April 10 Error!Reference source not found..The surface salinity exhibits larger spatial rain conditionsError!Reference source not found.,Error!Reference source not found., panel right) due to the presence of the signal of Besòs River freshwater.In the normal discharge, low salinity water is concentrated around However when a large amount of river water flows into the sea during offshore in the direction of river water outflow and then turned nalysis indicated that the water exchange rate between the water aroun river mouth and the outside area was proportional to the river discharge.In this section, selected aspects of the simulated results are compared with observations.The bulk of the measurements correspond to a field campaign conducted during March and April 2011 over the in the framework of the Several data are available for model comparison (the data set consists of water , wave data from buoys simulated salinity measurements of the to 41º 47'N and 43 CTD profiles were Besòs River flow data were ).The March 17 th survey Besòs river discharge of 50 flood" conditions occur during short (duration of storm events, typical of the Mediterranean climate, in small river mountainous urements correspond to "normal conditions" April 10 th 2011) is e surface salinity exhibits larger spatial , left panel) than in due to the presence of In the normal discharge, low salinity water is concentrated around into the sea during a flood event, downstream close nalysis indicated that the water exchange rate between the water around the Figure 5 shows two examples of the salinity salinity profiles from ROMS results in the same location (profile 33 and profile 41, indicated by stars in Figure 5, left panel).The water colum constrained to the first meters of the water column influence of the plume is small and localized at the fairly constant horizontal salinity structure.shows the same pattern than the profile from the campaign, but the values of salinity are overestimated up to 0.5 psu (Figure 5, top-right and bottom also present a stratified profile with maximum ROMS profile follows the same pattern but the salinity values are overestimated u A downward transport of freshwater through vertical mixing was evident during the observation period.The difference between the offshore s that observed during field campaigns suggests that the inclusion of vertical mixing and the temporal variation in river discharge should be studied

ANALISYS AND DISCUSSION
The system was run in 2 configurations, conditions.For the river flow both the Bèsos River and the Llobregat River are considered.For the urban runoff, the four CSOs namely Ginebra, Bogatell, Bac de Roda and Prim account.

COASTAL ENGINEERING 2012
Figure 5 shows two examples of the salinity profiles from campaign measurements versus the salinity profiles from ROMS results in the same location (profile 33 and profile 41, indicated by stars The water column is stratified in both surveys and th constrained to the first meters of the water column.During the normal conditions (April 11 influence of the plume is small and localized at the first meters of depth (5-15 meters) fairly constant horizontal salinity structure.ROMS profile result from the same date (April 11 shows the same pattern than the profile from the campaign, but the values of salinity are overestimated right and bottom-right panel).The post-rain conditions ( also present a stratified profile with maximum values between 0 and 8 meters depth.ROMS profile follows the same pattern but the salinity values are overestimated up to A downward transport of freshwater through vertical mixing was evident during the observation period.The difference between the offshore spreading distance of the plume predicted by ROMS, and that observed during field campaigns suggests that the inclusion of vertical mixing and the temporal should be studied in more detail in future theoretical studies. during field campaigns (left panel); measured CTD profiles versus ROMS results

ANALISYS AND DISCUSSION
configurations, low land boundary fluxes and high land boundary fluxes conditions.For the river flow both the Bèsos River and the Llobregat River are considered.For the urban runoff, the four CSOs namely Ginebra, Bogatell, Bac de Roda and Prim (Figure campaign measurements versus the salinity profiles from ROMS results in the same location (profile 33 and profile 41, indicated by stars e plume signal is (April 11 th 2011), the 15 meters), resulting in ROMS profile result from the same date (April 11 th 2011) shows the same pattern than the profile from the campaign, but the values of salinity are overestimated conditions (March 17 th 2011) epth.The shape of the p to 2 psu.A downward transport of freshwater through vertical mixing was evident during the observation preading distance of the plume predicted by ROMS, and that observed during field campaigns suggests that the inclusion of vertical mixing and the temporal in future theoretical studies.

versus ROMS results
high land boundary fluxes conditions.For the river flow both the Bèsos River and the Llobregat River are considered.For the Figure 6) are taken into

SUMMARY AND CONCLUSION
Numerical simulations of Catalan Coastal waters in spring 2011 have been conducted using a ROMS model driven by realistic atmospheric and land forcing.The model results were compared with CTD observations.The shape of the salinity profile was well reproduced but salinity values were overestimated.In the two field surveys carried out at the Catalan coast during the spring of 2011 before and during a flood event and during a normal discharge, the response of a plume to an abrupt change in river discharge was observed.The distribution of low salinity water was different between both discharges.In the normal discharge, low salinity water is concentrated around of the rivers mouths.As a large amount of river flood into the sea during the flood event, the plume spread offshore in the direction of river water outflow and turned downstream close to the coast.
Analysis of the results indicated that the water exchange rate between the water around the river mouth and the outside area was proportional to the river discharge.A downward transport of freshwater through vertical mixing was evident during the observation period.The difference between the offshore spreading distance of the plume from ROMS model results and that observed in the field campaigns suggests that the inclusion of vertical mixing and the temporal variation in the river discharge needs to be investigated further in future theoretical studies.
The difference between the simulation including the river outflow as a land forcing and the simulation including river and urban runoff as a land forcing suggests that urban runoff plays an important role in the spreading and shape of the river plume.
This paper provides a first interpretation the role of land boundary fluxes in determining the structure of the plume in the Catalan Sea and is limited to the period of observation (spring 2011).Further work should focus more on the effect of the river discharge annual cycle on the vertical flow structure.

Figure 1 .
Figure 1.Catalan Coast and coastal grid is used as input to both the river and urban runoff model land boundary fluxes is provided by the Barcelona Super Computing Center (BSC).It is generated by the air quality model CALIOPE (http://www.bsc.es/caliope) and provides information on the ed rainfall per hour in each grid cell of the domain (European, 1km of resolution).been generated by the same air quality model CALIOPE information of the wind velocity per hour in each grid cell of the domain.filethat has been used as input file for the coastal model is provided by and urban runoff models that have been calibrated as part of the FIELD_AC project urban runoff model provides hourly flow time series (m CSOs) that are situated in the Coastal model (netcdf format model provides hourly flow time series (m 3 /s) for both the Besòs and the Llobregat river resolution domains in order to introduce the initial and

Figure 3 .
Figure 3. Field Campaign domain and CTDs position located into the Coastal

.
Modeled surface salinity during March 17 th (left panel) and April 10 th 2011 (right panel)

Figure 6 .
Figure 6.CSOs positions.The results from a simulation land forcing is shown in Figure land forcing are shown in Figure concentrated around the rivers mouths and urban as land forcing.Although and turned downstream close to the coast.

Figure 7 .
Figure 7. Surface salinity ROMS results (psu).Simulation result during flood event (March 11 panel) and including rivers and urban runoff as a forcing (right panel).

Figure 8 .
Figure8. , shows the difference outflow only as a land forcing (Figure7, right panel).This difference suggests that includ role in the spreading and shape of the river plume.

Figure 8 .
Figure 8. Salinity differences (psu) between the result including the river outflow and the result including river and urban runoff as a land forcing.