1 Introduction

During the last decade, substantial progress in both meso-meteorological and numerical weather prediction (NWP) modelling and in the description of urban atmospheric processes, have been achieved. For instance, state-of-the-art nested NWP models have been developed which utilise land-use databases down to 1 km resolution or finer, enabling the provision of high quality urban meteorological data. Thus, NWP models are now approaching the necessary horizontal and verticalresolution to provide weather forecasts for the urban scale (e.g. Baklanov et al., 2002, 2008).

Many urban features can influence the atmospheric flow, its turbulence regime, the microclimate, and, accordingly modify the transport, dispersion, and deposition of atmospheric pollutants within urban areas, namely:

  • Local-scale non-homogeneities, such as sharp changes of roughness and heat fluxes;

  • Sheltering effects of buildings on wind;

  • Redistribution of eddies, from large to small, due to buildings;

  • Trapping of radiation in street canyons;

  • Effects on urban soil structure;

  • Differing diffusivities of heat and water vapour in the canopy layer;

  • Anthropogenic heat fluxes;

  • The so-called urban heat island;

  • Urban internal boundary layers and the urban mixing height;

  • Effects of pollutants (including aerosols) on urban meteorology and climate;

  • Urban effects on clouds and precipitation.

Despite the increased resolution and various improvements, current operational NWP models still have several shortcomings with respect to urban areas including:

  • Urban areas are mostly described by the same sub-surface, surface, and boundary layer formulations as rural areas.

  • These formulations do not account for specific urban dynamics and energetics or for their impacts on the simulation of the atmospheric urban boundary layer (UBL) and its intrinsic characteristics (e.g. internal boundary layers, urban heat islands, precipitation patterns).

  • NWP models have not been primarily developed for air pollution and emergency modelling and as such, their outputs need to be designed as suitable input for such urban-scale models.

Apart from Urban Air Quality Information and Forecasting Systems (UAQIFS) per se, improved urban meteorological forecasts will also provide information to city managers regarding additional hazardous urban climate features (e.g. urban runoff and flooding, ice and snow accumulation, high urban winds or gusts, heat or cold stress in growing cities and/or a warming climate). Moreover, the availability of reliable urban scale weather forecasts might be relevant in assisting the emergency management of fires, accidental toxic emissions, potential terrorist actions, etc.

2 Methodology for Urbanization of City-Scale Meteorological Models

2.1 FUMAPEX Strategy to Improve NWP and Meso-Scale Meteorological Models for Urban Areas

The FUMAPEX (FUMAPEX, 2005; Baklanov et al., 2005) strategy to improve NWP and meso-scale meteorological models includes the following aspects for the urbanisation of relevant submodels or processes:

  1. (i)

    Model down-scaling, including increasing vertical and horizontal resolution and nesting techniques (one- and two-way nesting);

  2. (ii)

    Modification of high-resolution urban land-use classifications, parameterizations and algorithms for roughness parameters in urban areas, based on the morphometric method;

  3. (iii)

    Specific parameterization of the urban fluxes in meso-scale models;

  4. (iv)

    Modelling/parameterization of meteorological fields in the urban sublayer;

  5. (v)

    Calculation of the urban mixing height based on prognostic approaches.

The following meso-meteorological and NWP models were used for urban conditions or for different variants of the “urbanisation” scheme (user/developer teams are in brackets, cf. http://www.fumapex.dmi.dk): 1. DMI-HIRLAM (DMI); 2. Local Model LM (DWD, MeteoSwiss, EPA Emilia-Romagna); 3. MM5 (CORIA, met.no, UH); 4. RAMS (CEAM, Arianet); 5. Topographic Vorticity-Mode (TVM) Mesoscale Model (UCL); 6. Finite Volume Model FVM (EPFL); 7. SUBMESO model (ECN).

2.2 Urban Fluxes and Sublayer Parameterisation

Two main approaches to simulate urban canopy effects are considered :

  1. 1.

    Modifying existing non-urban approaches (e.g., the Monin-Obukhov similarity theory,) for urban areas by finding appropriate values for the effective roughness lengths, displacement height, and heat fluxes (adding anthropogenic heat flux, heat storage capacity and albedo change). In this case, the lowest model level is close to the top of the urban canopy (displacement height), and a new analytical model is suggested for the urban roughness sublayer which is the critical region where pollutants are emitted and where people live (Zilitinkevich and Baklanov, 2004).

  2. 2.

    Alternatively, source and sink terms are added in the momentum, energy and turbulent kinetic energy equations to represent the effects of buildings. Different parameterizations (Masson, 2000; Martilli et al., 2002) have been developed to estimate the radiation balance (shading and trapping effect of the buildings), the heat, the momentum and the turbulent fluxes inside the urban canopy, considering a simple geometry of buildings and streets (3 surface types: roof, wall and road).

Several options for the integrated FUMAPEX urban module which can be used with NWP models have been suggested. In the first stage, four modules for model urbanisation (Fig. 8.1) were developed for further testing and for implementation in NWP models or their post-processors. It included the following modules:

  1. 1.

    DMI module: Based on the first approach, this includes a new diagnostic analytical parameterisation of the wind profile in the urban canopy layer (Zilitinkevich and Baklanov, 2004; Baklanov, 2008) and corrections to the surface roughness (with the incorporation of displacement height) for urban areas, and heat fluxes (adding anthropogenic heat fluxes, e.g., via heat/energy production/use in the city, heat storage capacity and albedo change) within existing physical parameterisations of the surface layer in NWP models, but with higher resolutions and improved land-use classification. This is applied in the city-scale version of the DMI-HIRLAM model.

  2. 2.

    EPFL module of the Building Effect Parameterisation (BEP): Based on the second approach and an improved urban surface exchange parameterisation submodel (Martilli et al., 2002; Hamdi, 2005). First this was tested in the research models FVM and TVM and then considered for inclusion in the DMI-HIRLAM and LM NWP models.

  3. 3.

    ECN module: Based on the detailed urban area soil and sublayer SM2-U model (Dupont and Mestayer, 2004; Dupont et al., 2004). First it was first tested with the large eddy simulation SUBMESO research model and then considered for incorporation into the DMI-HIRLAM NWP model.

  4. 4.

    Combined module: This includes all non-overlapping mechanisms from the SM2-U and BEP models. It was used in MM5 (Dupont et al., 2004) and applied to Paris by CORIA.

Fig. 8.1
figure 8_1_185808_1_En

General scheme of the FUMAPEX urban module for NWP models

Full size image

3 Results and Recommendations

3.1 Experience of Model Urbanisation

The range of improvement made by the FUMAPEX participants is summarized in Table 8.1. Many of the parameterisations were evaluated using data sets collected as part of the BUBBLE and ESCOMPTE projects. Implementation of the urban modules significantly improved the forecasted meteorological fields for urban areas. The first module, the cheapest way to “urbanise”, can be easily implemented into operational NWP models as well as into Regional Climate Models. The second module, although more expensive (≈5–10% computational time increase), provides the possibility to consider the energy budget fluxes inside the urban canopy. However, this approach is sensitive to the vertical resolution of NWP models and is not very effective if the first model level is greater than 30 m. Therefore, an increase in the vertical resolution of current NWP models is required. The third module is considerably more expensive computationally than the first two. It provides the possibility to accurately study the urban soil and canopy energy exchange including the water budget. Consequently, the second and third modules are recommended for use in advanced urban-scale NWP and meso-meteorological research models. This will be demonstrated for NWP models in a forthcoming paper. The third module may be very useful for implementation into research submeso-scale or micro-meteorological models (e.g., SUBMESO) for large eddy simulation or assessment (non-prognostic) studies. The first and second modules can be also used as urban interfaces or post-processors of NWP data for UAQ models.

Table 8.1 Overview of the improvements made by the FUMAPEX partners (FUMAPEX, 2005)
Full size table

Simulation results with these urban modules show that the radiation budget does not differ significantly for urban compared to rural surfaces, as the increased loss of a net longwave radiation is partly compensated by a gain in net shortwave radiation due to a lower albedo. The turbulent fluxes of sensible and latent heat, as well as their ratio, are variable and dependent in particular, on the amount of rainfall during the preceding period. The storage heat flux is usually significantly higher in urban areas compared to densely vegetated surfaces. This cannot be explained entirely by a higher thermal inertia, as this quantity is only slightly higher for urban than rural environments. Other factors of importance are the reduced moisture availability and the extremely low roughness length for heat fluxes. The anthropogenic heat flux, a typical urban energy flux, is absent in rural or natural areas.

A sophisticated way to simulate the storage heat flux is using the BEP or SM2-U modules. One goal is to simplify the parameterisation of the storage heat flux in NWP model simulations for the main types of urban area and concentrations of urban elements. Use of these modules may provide the possibility to develop a simplified method for urban classes used in NWP models.

3.2 Further Improvements in NWP and UAQ Forecasting Systems

The next step should be the comparison of urban modules within the operational NWP models and their verification with respect to urban meteorological forecasts. It is also suggested to develop stand alone urban canopy models for uses as an interface or post processor module. These also need to be tested (module 2 has been successfully tested, FUMAPEX, 2005). These will be run with readily available NWP data as a first approximation and will improve meteorological fields of higher resolution that are near or within the urban canopy. This type of approach does not improve the meteorological forecast for the urban area and does not allow feedbacks. It does have the advantage that it does not require any modifications to an operational NWP model (which is usually very difficult and time consuming). This approach thus considers the urban sublayer models (together with several upper layers and surrounded areas) as interface modules between the NWP and UAQ models.

The current versions of the considered urban modules have several shortcomings which need to be further improved. For the first approach (module 1), the analytical model for wind velocity and diffusivity profiles inside the urban canopy (Zilitinkevich and Baklanov, 2004; Baklanov, 2008) has to be tested with different NWP models and meteorological pre-processors, and carefully evaluated using experimental data for different regimes. Additionally, it would be advisable to extend this for temperature and humidity profiles. The current version of the second module (BEP) does not consider the moisture and latent heat fluxes and does not completely incorporate the anthropogenic heat flux. Therefore, these should be included into a new version of the BEP module. Recalculation of accessible meteorological fields (e.g. wind prifiles) in the lowest sub-layers (not only on the NWP main vertical levels) is necessary. The third module (SM2-U) needs to consider the building drag effect (it will be realised in module 4), and snow and ice need to be included for NWP during winter periods, especially for high latitude areas. The existing version is computationally too expensive for operational NWP models, so it needs to be optimised to make calculations only for the urban cells.

Obviously these developments require more evaluation with appropriate data. Data availability, would also lead to addition development and initialisation improvements for NWP or meso-meteorological models simulations. This includes a need to conduct future urban field campaigns to provide data from which insights may be gained to help devise simpler models/parameterizations for complex models. The existing measurements have limitations which arise due to inescapable constraints on field programmes in cities.