Physical flows modelling

The modelling approach within BRIDGE integrates different types of models from mesoscale air quality models to urban canopy models. The energy and water fluxes are measured and modelled at local scale. The fluxes of carbon and pollutants are modelled and their spatio-temporal distributions are estimated. These fluxes are dynamically simulated in a 3D context by using state-of-the-art numerical models. The outputs of the above models lead to indicators which define the state of the urban environment and are incorporated into the BRIDGE Decision Support System (DSS). The cascade modelling technique from large to local scale is the main methodology applied in BRIDGE and allows estimating the pollutant concentrations and the fluxes associated to varying geographical extents of urban development scenarios. 

Mesoscale meteorological models such as MM5 (Fifth-Generation NCAR/Penn State Mesoscale Model), developed by PSU/NCAR United States, and WRF (Weather Research and Forecasting Model) are considered within BRIDGE for simulating the meteorological variables through the numerical simulation of the atmospheric flows based on the Navier-Stokes equations. MM5 and WRF models simulate the atmospheric flow (meteorology only) in a 3D cube with spatial resolutions on about 1-100km with domains between 20-50 km (urban domain) to thousands of km (Europe). These models require as input the meteorological boundary conditions coming from the outer domain. This information is typically obtained from Global Meteorological Models. MM5 and WRF models give detailed information of all meteorological variables and fluxes involved in the atmospheric flow (comprehensive descriptions of atmospheric motions; pressure, moisture, and temperature fields; momentum, moisture, and heat fluxes; turbulence, cloud formation, precipitation, and atmospheric radiative characteristics). They also require land use and topographic information which should be adequate to the specific spatial resolution that they are run. 

These meteorological models provide input information (meteorological fluxes and variables) to chemical transport models such as CAMx (Comprehensive Air quality Model with extensions), developed by ENVIRON International Cooperation from California, United States of America, and CMAQ (Community Multiscale Air Quality Modelling System), developed by EPA United States, which estimate the pollutant concentrations in the atmosphere. These models provide detailed information, at mesoscale level, of the flux exchange between atmosphere and soil and pollutant concentrations in the air as an average per grid cell and time unit in a 4D (space + time) environment. They can simulate the emission, dispersion, chemical reaction, and removal of pollutants in the troposphere by solving the pollutant continuity equation for each chemical species on a system of nested three-dimensional grids. Therefore, CAMx and CMAQ require additional input data such as emission data (amount of pollutant emitted in a grid cell per second). CAMx and CMAQ can estimate the ozone concentrations (and other secondary pollutants) in the atmosphere. In addition, CAMx and CMAQ models use different aerosol models to estimate primary and secondary PM concentrations in the atmosphere. 

Therefore, the following combined models will be used within BRIDGE: 

The MM5-CAMx air quality modelling system is composed by the chemistry-transport model CAMx, forced by the MM5 meteorological fields. This modelling system uses the meteorological fields driven by the MM5 model and CAMx computes the atmospheric concentrations of various gaseous and aerosols. The MM5-CMAQ (Figure 1) is a model composed by the mesoscale meteorological non-hydrostatic model MM5 and the chemical transport model CMAQ. 

WRF/CHEM (Figure 1) is a new generation of mesoscale models based on the on-line simulation of chemical and meteorological processes which is more suitable for climate studies. It is a model which simulates the meteorology and the chemistry simultaneously and it is, therefore, being closer to the atmospheric reality. WRF/CHEM is based on the same principles with MM5 and WRF but it also includes the chemical solver in every time-step together with the meteorology. The computer demand in substantially higher than using MM5-CMAQ or MM5-CAMx or WRF-CMAQ/CAMx.



Figure 1: Visual representation of outputs of different mesoscale models such MM5-CMAQ or WRF/CHEM.


At the local scale of a few hundred meters, the two models used have different objectives and are CFD (Computational Fluid Dynamics) models. The MICROSYS (Microscale Air Quality Simulation System), developed by UPM, is a microscale fluid dynamics model which includes chemical dispersion and transformation of species. It can estimate the meteorological variables and heat fluxes in the complex urban environment. MICROSYS can receive boundary conditions from the mesoscale models (MM5-CMAQ and/or WRF/CHEM) and runs over microscale domains (domains range between few km to meters). The atmospheric flow is also solved using numerical methods with higher spatial and temporal resolution (according to the Courant law) in a 4D environment. The MICROSYS model requires similar input information with the mesoscale models but, in addition, it requires information related to urban characteristics (buildings geometry and heights, soil type, asphalt, cement type, tree types, roof materials, etc.) in order to have a more detailed description on the heat flux exchange between surface and the atmosphere. In order to obtain a detailed emission dataset with a very high spatial resolution (meters), it is usually combined with a traffic model (for urban environments) based on the Cellular Automata Traffic Model (CAMO), developed in UPM. MICROSYS (Figure 2) can then provide a detailed description of the heat flux exchange, pollutants concentration and can assimilate different types of observational datasets such as surface temperature, air temperature, etc.



Figure 2: Visual representation of outputs of MICROSYS and CAMO.


The VADIS model, developed in 1998 at the Department of Environment and Planning of UAVR, can obtain detailed information related to the biosphere-atmosphere heat flux exchange at surface level and simulate the passive and chemically active pollutants in the micro-scale urban domain. More specifically, it can estimate pollutant dispersion in the atmosphere under variable wind conditions. The model requires information that allows characterising the simulation domain, the meteorological conditions at the entrance of the domain, and the emissions for the considered period of time. It uses air temperature and wind speed and direction as meteorological initial conditions at the entrance of the domain and at a specified reference height. More specifically, initial vertical profiles, the turbulent kinetic energy and the energy dissipation at the entrance of the domain are used to describe the variation with height of the mean wind speed. It also requires information related to road traffic emission which is usually produced by a traffic model in an urban context, such as the Transport Emission Model for Line Sources (TREM) using detailed data on vehicles counting. The output data is constituted by the three wind velocity components, turbulent viscosity, turbulent kinetic energy, energy dissipation, pressure, temperature, and pollutant concentration in each grid cell. 

The WRF/CHEM-UCM-MICROSYS model is an integration of mesoscale models and microscale CFD models. This model – with different versions – can be applied to scales up to 200 m of spatial resolution over urban areas in BRIDGE and also to microscales with MICROSYS with feedbacks to mesoscale domains. It is a combination of the models described before. The amount of output variables of one of these models is quite large and could be larger than 400 variables. 

Another microscale model to be used at BRIDGE is the Local-scale Urban Meteorological Parameterization Scheme (LUMPS)3, which is a microscale surface flux model that utilises standard meteorological observations and land cover characteristics. It can model the variability in fluxes both spatially and temporally. The sub-models utilised within LUMPS are used to calculate: a) the net all-wave radiation using NARP4 (Net All-wave Radiation Parameterization); b) the storage heat flux using the OHM3 (Objective Hysteresis Model); c) the latent heat flux using LUMPS3 and SUES2 (Single-source Urban Evapotranspiration -interception Scheme); d) the turbulent sensible heat flux using LUMPS3 and the residual method (using the surface energy balance). In addition the Urban Water Balance (UWB)1 is currently being added to LUMPS to simulate the urban hydrologic cycle which impacts directly on latent heat flux and therefore turbulent sensible heat flux. 

At urban scale, the URBAIR (Urban Air Quality) model evaluates air quality and dispersion patterns and it is a second generation Gaussian plume model intended to be used for distances up to about 10 km from the source. URBAIR is a steady state atmospheric dispersion model, based on boundary layer scaling parameters, instead of relying on Pasquill stability classification. The model was developed for simulating passive or buoyant gas dispersion and deposition at local and urban scales. It is designed to allow consideration of dispersion in rural or urban areas, including the treatment of building effects. To characterize the meteorological conditions within the simulation domain, the model requires meteorological information driven by mesoscale meteorological models or by surface measurements and upper air soundings databases. URBAIR requires also the characterization of topography, land-use and the emissions of anthropogenic sources, which can be provided by inventories or by the Transport Emission Model for Line Sources (TREM), using vehicles counting data. Emission and meteorology information is defined on an hourly or daily basis. The output data is constituted by the meteorological parameters and pollutant concentration at user-specified receptor points or spatially distributed over a regular grid. 

The Advanced Canopy-Atmosphere-Soil Algorithm (ACASA) model, developed by the University of California Davis (UCD), estimates energy and mass fluxes between surface and the atmosphere. It treats the surface and associated fluxes as an interconnected system, and the atmosphere, the urban surface, and the soil are represented as a multi-layer system. The ACASA domain extends maximally to 100m above the city and plant canopy elements to ensure applicability of the turbulence assumptions. ACASA incorporates higher-order closure principles for turbulent statistics to predict effects that higher-order turbulent kinetic and thermodynamic processes have on the surface microenvironment and associated fluxes of heat, moisture, momentum and carbon. These processes include turbulent production and dissipation to turbulence kinetic energy, turbulent vertical transport of heat, mass, and momentum fluxes. Using a set of governing equations, ACASA creates vertical profiles of temperature, humidity, mean wind, and CO2 concentration. ACASA has several additional features including: (1)radiative transfer within the surface layers, and (2)surface heat storage processes. ACASA is being coupled with the mesoscale model WRF in order to identify how multiple environmental factors, in particularly climate variability, population density, and species distribution, impact future carbon cycle prediction across a wide geographical range. 

SURFEX is an externalised surface scheme that contains various modules allowing describing the exchanges of water, momentum, and energy on 4 tiles of surface: sea, lake, vegetation, and the city. A grid value is then simply an area averaged value of the different tiles present in the grid cell. Over urban surfaces, SURFEX includes the Town Energy Balance (TEB) single layer urban canopy module. Urban canopy is assumed to be an isotropic array of street canyons. The advantage over more comprehensive urban surface schemes, which include parameterizations for the canyon orientation and heterogeneous buildings morphology, is that relatively few individual surface energy balance evaluations need to be resolved, radiation interactions are simplified, and therefore computational time is kept low. TEB simulates heat and water exchanges over three generic surfaces (roof, wall, and road), where heat transfers are computed through several layers of materials. Anthropogenic heat and vapor releases from buildings, vehicles and chimneys can also be added. TEB utilises standard surface thermal parameters and observed or simulated atmospheric parameters and radiation data from above roof level and returns the fluxes and urban canyon climate characteristics (air temperature, humidity, wind) at the neighborhood scale. Despite the simplification hypotheses, offline simulations of TEB have been shown to accurately reproduce surface energy balance, canyon air temperature, energy consumption and surface temperatures observed in dense urban areas for various seasons. 

The SCADIS model solves the wind and turbulence fields over a heterogeneous scene. The scene-specific terrain topography and forest canopy height and leaf area density (or possible a corresponding description for an urban canopy) can be described in detail, to provide a realistic description of the scene. SCADIS then computes the footprint, for an eddy flux tower or other such measurement instrument, based on the wind and turbulence fields. The SCADIS Footprint Calculator operates in a 2D mode (x-horizontal, z-vertical). The horizontal resolution can be adjusted by the user, and the vertical domain is up to 3 km, so that all of the atmospheric boundary layer is contained. The SCADIS Footprint Calculator assumes neutral atmospheric stratification. 

The hydrological model to be applied within BRIDGE is called SIMGRO. SIMGRO can produce detailed information on all the hydrological processes present in an urban environment, especially through the module SIMGRO-urban. SIMGRO-urban describes the rainfall runoff process of urban areas, including paved and unpaved areas, the unsaturated zone, plant-atmosphere relations and the sewerage system. Modules for groundwater flow and surface water flow can be included, depending on relevance and data availability. Typical model outcomes are: sewerage outflow, (reduction of) evapotranspiration, groundwater recharge. With SIMGRO-urban the hydrological impact of measures (like green roofs), land use change (urbanisation) and climate change can be assessed. 

In addition, a model having a completely different context is also used. The NKUA NN model is developed by NKUA for the estimation and prediction of the urban heat island intensity in various locations of a large urban area. The specific model is based on neural networks technology. Neural networks are a computational technique that simulates the operation of the human brain’s neurons. To some extent, the NN approach is a non-algorithmic, black box strategy, which is trainable. The purpose is to train the neural black-box to learn the correct response or output (e.g. classification) for each of the training samples. This strategy is attractive to the system designer, since the required amount of a priori knowledge and detailed knowledge of the internal system operation is minimal. After training the internal (neural) structure of the artificial implementation the NN is self-organized to enable extrapolation when faced with new, yet similar, patterns, on the basis of experience with the training set. The application of NN model to the urban structure requires hourly ambient air temperature and hourly humidity values for various urban positions. Moreover the NN model requires a series of urban stations and a reference station which is placed in a rural area. It can model the night-time, as well as the daytime heat island intensity. 

A Regional Climate Model (regCM3) is also integrated in BRIDGE modelling setup. It is a climate version of the MM5 and/or WRF meteorological mesoscale models, which can be used to simulate regional climate periods provided global initial and boundary conditions. It can produce information on the climate evolution for future scenarios, to provide climate variables (temperature, wind, humidity, PBL height, etc.) and fluxes under climate change. 

Finally, a CA module was integrated in BRIDGE DSS for the simulation of land use dynamics. The CA is used to determine the spatial distribution of an aggregate land-use demand, taking into account the interaction between different land-uses, as well as the physical, environmental and institutional factors characterizing each cell. CA can easily account for the planning decisions whose broader effects in terms of a spatial distribution of land-uses have to be evaluated. In CA adopted in BRIDGE the neighbourhood is defined as the circular region around the cell with a typical radius that ranges from 0.5 Km to 1.5 km depending on the grid resolution. The typical output of the CA are maps showing the predicted evolution of land uses in the area of interest, over a predefined period of time. By varying the inputs into the CA model, it can be used to explore the future urban development of the area under consideration under alternative spatial planning and policy scenarios. 

Regarding the models implemented in the framework of the BRIDGE project, some are used in ON-LINE mode and others in OFF-LINE mode. The models called as “off-line” are those models that require a large amount of data and high computer demands due to their size and long duration of the simulations. So they have to be run “in-house” in clusters and/or supercomputers instead of being implemented into the DSS tool. The results of the “off-line” simulations are stored in the DSS database to the specific format required by the DSS tool. The “on-line” models are integrated in the DSS tool due to their simplicity and limited computer resources, and therefore they run by the BRIDGE DSS tool for each case study.

The OFF-LINE models are the following:

·    WRF-UCM model (run by UPM)

·    MICROSYS (CFD) model (run by UPM)

·    MM5-CAMx model (run by UAVR)

·    VADIS (CFD) model (run by UAVR)

·    NKUA Models (Neural Network) (run by NKUA)

·    ACASA model (run by CMCC)

The ON-LINE models are the following: 

·    LUMPS V.5 model (produced by KCL)

·    SURFEX model (produced by CNRM)

·    SCADIS model (UHEL)

·    URBAIR model (UAVR)

·    SIMGRO model (ALTERRA)