Past Events  
   

IN THE FRAME OF COST 715, WORKING GROUP 2 EXPERT MEETING ON SURFACE ENERGY BALANCE IN URBAN AREAS

12 APRIL 2000

By D R Middleton1), A. Martilli2), and M. Piringer3)

1)Meteorological Office, Bracknell, Berkshire, UK.
2)Swiss Federal Institute of Technology, Lausanne
3)Central Institute for Meteorology and Geodynamics, Vienna, Austria

Introduction

The meeting was arranged under COST Action-715 ("Meteorology applied to urban air pollution problems", chaired by B. Fisher, University of Greenwich, UK, and M. Schatzmann, University of Hamburg, D) to bring together experts to present reports on current understanding on the surface energy balance in urban areas and discuss future research needs. COST is an acronym of "Co-operation in the fields of Science and Technology", financially supported by the European Commission to encourage and facilitate mutual scientific exchange among Member States. Each action is devoted to a specific topic whose main scientific goals are fixed in a so-called "Memorandum of Understanding", to be signed by COST Member States. An Action lasts in general five years. COST Action-715 has been signed by 18 countries and will end in September 2003.

COST Action-715 is organised in four working groups, dealing with urban wind fields, surface energy budget and mixing height, air pollution episodes in cities, and meteorological input data for urban site studies. The members of working group 2, having organised the expert meeting in Antwerp, are in alphabetic order: A. Baklanov, Danish Met. Institute; K. De Ridder, VITO, Belgium; J. Ferreira, Met. Institute, Portugal; S. Joffre (vice-chairman) and A. Karppinen, Finnish Met. Institute; P. Mestayer, EC Nantes, France; D. R. Middleton; M. Piringer (chairman); M. Tombrou-Tzella, University of Athens; and R. Vogt, University of Basle.

Program

The expert meeting in Antwerp consisted of eight presentations, followed by a discussion. In the following, the presentations are listed in their temporal order:

In a very 'rough' way the talks can be split in two groups: a first group more 'experimental', focusing on how to carry out measurements in urban areas and on their interpretation (Rotach, Grimmond, De Ridder, Oke, Middleton); a second group more 'numerical', presenting some parameterisations for mesoscale models (Masson, Guilloteau, Martilli).

Summary of presentations

Experimental presentations

Mathias Rotach considered proper siting of instrumentation in urban areas to be difficult and controversial: the purpose of the measurements must be considered. Research measurements have requirements different from routine observing. The urban area exhibits huge complexity and measurements must match their intended purpose. For routine observation, the station needs to be representative. He noted the following layers from ground up: canopy (between elements) as a part of the roughness sub-layer (individual elements have local effects, up to 2-5 times the height of the buildings), inertial sub-layer (flow is representative of a larger area of elements), with an urban mixed layer above. Their heights are of order h, z* ~ 2-5 h, 0.1 x zi, zi, respectively. Deciding on a representative site and erecting a mast to reach into the inertial sub-layer is not trivial. The layer may vary in height; it may differ for heat/momentum. Rotach mentioned the problem that in particular conditions (high buildings and low mixing height) the inertial sub-layer might not exist. Comparisons of urban winds and airport winds (as presented to COST 715 WG1 in Roskilde last year) from different cities are useful. At WMO stations, 10m for wind or 1.5-2.0m screen height for temperatures are normal, but there are yet no recommendations for routine urban measurements. Spatial inhomogeneity must be taken into account.

In the discussion following Rotach's presentation, Oke, as a rapporteur of WMO on routine urban measurement specifications, explained that he is making recommendations regarding the siting of urban stations and would like to be sent any national guidelines for the siting of urban stations. Referring to the choice of instrumentation, it is important to keep in mind that the typical urban meteorological conditions are: low winds, small vertical gradients and vertical winds not negligible close to the surface. Due to the strong variability of the urban surface it is important to have instruments with a good time resolution and they should be well distributed in space (in vertical and horizontal directions).

Sue Grimmond and Tim Oke described their measurement campaigns in North American cities (Mexico City, two sites from Vancouver, Chicago, Tucson, Miami, three sites from Los Angeles, Sacramento). Energy fluxes are important model inputs. They use tall towers, >2 x h, for measurements of total radiation, sensible and latent heat fluxes, with careful attention to fetch. Sites are characterised by GIS using maps/photos/aerial photos/surveys of the area after Grimmond and Souch (1994). The source area flux model of Schmid (1994) serves to identify the footprint of influence on a sensor on a mast. Their studies include dry and moist areas, and irrigated cities; fraction of buildings cover 0.2-0.5, fraction of hard surface up to 0.4. Vegetation cover varies from 0.01 (Mexico City site) up to 0.56 (Arcadia at Los Angeles). Results are plotted versus local solar time; water flux (latent heat) proves very important for the heat island. Irrigated cities in dry areas (Arizona, California) can release more latent heat than their surroundings due to irrigation and extra vegetation. Measured Bowen ratio (ratio of latent and sensible fluxes) is used to compare cities. Variables include Q* (net radiometer), QH (sonic), QE (krypton hygrometer), heat storage term dQS from Q*-QH-QE., and a parametrization scheme. Roughness terms zd and z0 are determined via morphometric (geometric) methods, via plan area or frontal area indices; Raupach's method is preferred, see Grimmond and Oke (1999). Using these with kinematic heat flux they calculate L, and plot z/L using measurement height z for different times of day. The analysis of the latent heat fluxes shows big variations between the cities mainly as a function of the fraction of vegetation cover and the fraction of irrigated area. The sensible heat fluxes show a strong variation from one day to another (even of 200Wm-2). Nevertheless, on average, the ratio between the sensible heat flux and total radiation increases during the day for all the cities (even if the slope of the curve can change from one site to another even by a factor of six). As explained before, the heat storage was computed from the values of heat fluxes and total radiation. The diurnal curve for the ratio between the heat storage and the total radiation shows a decrease during the day for all the sites. This means that the urban surface has a tendency to store more energy during the morning hours and to release more sensible heat in the evening. The heat storage term Qs shows hysteresis in morning and evening; it may reach 100-250W m-2. The objective hysteresis model of Grimmond and Oke (1999) describes this using independently derived parameters. Incoming solar radiation Kv (or net all wave radiation Q*) is a very important parameter and should be routinely measured, especially in cities.

Tim Oke and Sue Grimmond evaluated typical values for radiation budgets for rural areas and cities in summer at temperate latitudes and he concluded that differences in net radiation are small. On the other hand, the analysis of typical values for the energy budget shows that the urban areas favour sensible over latent heat releases and an increase in the heat storage. Nevertheless, there can be a big variability from one site to another, especially for the rural values. As an example, Oke analysed the differences between urban and rural surface energy balances for three sites (Tucson, Sacramento, Vancouver). The rural areas surrounding the three cities are very different: desert bush (Tucson), both semi-arid and irrigated (Sacramento) and moist farmland (Vancouver). In general, the variability in the surface energy budgets is much bigger between the different rural sites than between the urban sites. In particular, there is the case of Tucson and Sacramento semi-arid, where the sensible heat flux at the rural site is bigger than at the urban site with an inversion of the traditional heat island effect. Also, in both sites, the latent 'urban' heat flux is bigger than the corresponding rural flux. Heat storage is always greater in the cities. Errors in modelling fluxes may arise unless account is made for the fact that the eddy diffusivities (averaged over an urban area) are not equal, i.e. KE<>KH. Removal of urban irrigation can change the Bowen ratio from about 1.1 to 2.7, as in Vancouver when there was an irrigation ban imposed.

Measurements shown by Grimmond and Oke are representative of the surface energy balance in a specific point of the city and were obtained by means of instruments located on towers. Koen De Ridder presented two methods for deriving a map of the surface energy balance from satellite images and he discussed the applicability of these methods to the urban case. In the first method, called (Ts-Ta)-based, the sensible heat flux is derived using the surface radiation temperature Ts (measured by the satellite) and local air temperature (Ta), surface wind speed and information on the roughness parameters. The net radiation RN can be estimated and ground flux approximates 0.1 to 0.4 times RN for urban surfaces. The latent heat flux is the residual term of the budget. The advantage of this method is a straightforward application of the aerodynamic formula. The disadvantages are that the radiation temperature is generally different from the aerodynamic temperature, the method is highly sensitive to the errors in the difference between surface and air temperature (especially for very rough surfaces, like a city), and that correct specifications of the roughness length are difficult to obtain. The satellite measurement of surface temperature can be prone to errors due to the atmospheric absorption and the uncertainty in the specification of the surface emissivity. Moreover, for urban surfaces, the ratio between the roughness length for heat and momentum is not well known and the presence of the anthropogenic heat fluxes makes the estimation of the surface energy balance more difficult.

The second method consists in estimating the surface parameters from satellite remote sensing (albedo, emissivity, fractional vegetation cover, etc.) and land-use maps (roughness length, etc.) and then use a SVAT (Soil Vegetation Atmosphere Transfer) module to compute the surface energy balance. The SVAT module can be used either off-line (forced with surface meteorological observations) or on-line (implemented in a mesoscale meteorological model). The advantages of the method are the direct applicability in an atmospheric model and the possibility to run even under cloudy conditions. The disadvantages are a less straightforward approach with respect to the (Ts-Ta) -based method and the need of more ancillary data. In particular, for urban areas, the SVAT modules must be 'urbanised' by modifying the albedo, emissivity, surface permeability, roughness length for heat, thermal properties (conductivity and heat capacity) and by introducing the anthropogenic heat fluxes in the surface energy budget.

Doug Middleton in collaboration with Nicola Ellis presented results from field measurements and modelling of surface fluxes in Birmingham, UK. The reasons behind interest in diagnosing the correct timing and sign of stability in urban areas was mentioned with regard to notable air pollution episodes, and the needs for air pollution modelling. Early work to parameterise Tim Oke's urban-rural differences for QH via the integral of Q* was described. This is used routinely in the model Boxurb for air quality forecasts. The urban measurements were then described, showing the Birmingham site. Synoptic data show the January/February 1999 data were largely neutral, with regular rain observed in the first two weeks. The SEB model by Best was used to analyse the results. The radiation model works well, given a good cloud observation. The surface temperature has significant impact on the heat flux calculated near the ground, and some difficulty arises in this part of the modelling. In this type of comparison it is important to have the sensors and model calculations at comparable heights. The problem of what to do when 1/L takes a default value was also discussed. Actions arising include: a sensitivity study of how the model behaves under different forcing, a test of model soil initialisation using an input or first measurement of deep soil temperature, role of leaf area index when modelling a 'concrete canopy', and exploration of coupling between soil surface and the canopy using other terms in addition to long wave radiation (Best having simplified Deardorff's work). There is also a need to try turbulence processing in line with the ideas of Mathias Rotach. He is attempting to match data from wind tunnels and urban studies.

Numerical presentations

Valery Masson presented his surface scheme TEB (Town Energy Budget), implemented in the atmospheric model MESO-NH (Masson, 2000). The TEB scheme computes three energy budgets for roof, road and walls by taking into account the long and short wave radiation trapped in the urban canyon. An average over all street orientations is done. All walls (roads) have the same solar input and are exposed to the same wind in the canyon and, consequently, all walls (roads) have the same temperature. The anthropogenic fluxes from traffic and factories are also considered in the scheme. In order to estimate the latent heat fluxes, water and snow interceptions are considered. The momentum fluxes are deduced with Monin-Obukhov similarity theory. Canyons serve as solar radiation traps absorbing up to ~70 W m-2 from Q* with multiple reflections. The wind profile is logarithmic to roof level; exponential below. The model also simulates heat transfer through the walls and roofs of buildings. The comparison with the field measurements (temperature, humidity, wind) shows that the model MESO-NH with the TEB scheme was able to reproduce the main features of the anticyclonic case above the Paris urban area. The study of the impact of the urban area on the circulation shows that during the night an Urban Heat Island of 10K was present which induced a neutral or slightly unstable layer 100m deep over the city, without any development of urban breeze. On the contrary, during the day an urban heat island of 5K developed a strong urban breeze with convergence near the ground over Paris, vertical winds and divergence at the top of the Boundary Layer.

Emmanuel Guilloteau extended the Force Restore soil model (Noilhan and Planton, 1989) used in the SUBMESO atmospheric model to model urban soil-atmosphere interactions. The model is composed of a soil layer (11 different soil types), thick enough that day averaged fluxes are negligible at this depth, and a canopy layer without thickness. In every grid cell the percentage of five different soil covers is defined: bare soil (the interface with the atmosphere is the ground surface), natural soil with vegetation (11 types of vegetation), water (3 types), artificial soil (2 types, asphalt and concrete), buildings (4 types of roofs). Each soil cover is submitted to the atmospheric forcing on one hand, and, to a return to the equilibrium state on the other hand. Phenomena taken into account are evaporation from bare and artificial soils, transpiration from vegetation, runoff towards the soils and evaporation of the water intercepted by vegetation and buildings, infiltration through artificial soil towards the underlying soil layer, evacuation of the exceeding water intercepted by buildings and artificial soils towards the sewers, evacuation of the water from the underlying soil layer towards a deeper soil layer and sensible heat exchange between the interface and the atmosphere. For the artificial soil the contribution of the anthropogenic heat sources (modelling the moving vehicles) is also taken into account. The model was tested in one dimension during one year with imposed meteorological data (no feedback from the soil model) for different typical urban quarters (change in the percentage of the five soil covers). Results show that the model is able to predict qualitatively well the surface temperature and humidity, while it has some problems in the estimation of the heat storage. According to Guilloteau, this might be due to a poor representation of the building effects.

Alberto Martilli in collaboration with Clappier and Rotach presented a parameterisation to compute heat and momentum fluxes in urban areas for mesoscale models. The method involves modifications to the momentum, energy and TKE equations below roof level (the model surface being street level) to take into account the contribution of the three active surfaces of the roughness element (wall, street and roof). For the exchange of momentum, two different roughness lengths are defined for roof and canyon floors, respectively, while the contribution of the walls is parameterised with a drag force approach. The sensible heat fluxes are determined as a function of the difference between the air temperature and the corresponding surface temperatures. A complete energy budget equation is solved for each of the three surfaces. The short and long wave radiative fluxes are computed by taking into account the shadows and multiple reflection effects of the street canyon element. For every grid cell it is possible to define several street directions. The impact of the modifications on the results for a simple 2D case (rural-urban-rural) were analysed and compared with available field measurements. The comparison shows that the model results are in good agreement (at least qualitatively) with the available measurements. An analysis of the impact of the different terms in the equations shows, furthermore, that walls are the most active surfaces for momentum and TKE during day and night, while for temperature the impact of wall's fluxes is more important during night, but street and roof are more active during day.

Conclusions

  1. A number of European groups run mesoscale models with sub-models of fluxes for urban areas at higher spatial resolution than some schemes routinely applied by some European weather services.

  2. Building energy flows and thermal properties should be modelled, along with multiple reflection absorption between canyon walls.

  3. Water flux is a very important determinant of city heat island effects; surrounding countryside must also be considered as 'rural' areas vary significantly.

  4. The behaviour of turbulent flux profiles as the roughness elements are approached from above requires study.

  5. Masts should go above 2 x h, into the inertial sub-layer and above the roughness sub-layer. The heights of these layers vary with conditions and fetch.

  6. Horizontal inhomogeneity means diffusivities differ, KE<>KH.

  7. Sites should be characterised by aerial photos/local surveys/maps/building measures/GIS.

  8. Satellites measure some aspects of the urban environment, but are incomplete on their own and require skilled interpretation.

  9. As mesoscale models become more sophisticated, efforts have to be undertaken to increase the capabilities of routine urban measurements, i.e. to install masts and to modernise the equipment, e.g. by routine use of sonics.

References

Grimmond, C.S.B. and T.R. Oke (1999) Aerodynamic properties of urban areas derived from analysis of surface form. Journal of Applied Meteorology, 38, 1262-1292.

Grimmond, C.S.B. and C. Souch (1994) Surface description for urban climate studies: a GIS based methodology. Geocarto International, 9, 47-59.

Masson, V. (2000) A physically-based scheme for the urban energy budget in atmospheric models. Boundary-Layer Meteorology , 94, 357-397.

Noilhan, J. and S. Planton (1989) A simple parameterisation of the land surface processes for meteorological models. Mon. Wea. Rev., 117, 536 - 549.

Schmid, H. P. (1994) Source areas for scalars and scalar fluxes. Boundary-Layer Meteorology, 67, 293 - 318.
       
  Past Events  
   

[To Contents]    [To Next Topic]