Outdoor Scale Model Experiments for the Evaluation of Urban Modeling Studies

For understanding the unique features of urban climates, there have been many studies in real cities in which data was acquired using towers, aircraft, and satellites. However, such full-scale studies have not yet provided a comprehensive understanding of the complicated physical processes that contribute to urban climate. As a complementary method, we have started outdoor model experiments at a scale of 1/5 and 1/50 (each compared with typical residential area of Tokyo) at the campus of Nippon Institute of Technology from September 2004. Such a model has the advantages of allowing us to make complete measurements and to obtain data on a relatively uniform area. For example, we can probe the physical processes within and above the roughness sub-layer more completely than those in a real urban area. Also, by working with a uniform area, the results are easier to interpret and more suitable for urban modeling than data from real cities. The scale models consist of cubic concrete blocks 1.5-m on a side with 0.1-m thick walls (1/5 model), and of cubic concrete blocks 0.15-m on a side (1/50 model). The blocks were distributed in an array on concrete pavements having total area of 100m x 50m (1/5 model) and 12m x 12m (1/50 model), respectively. In both models, blocks were arranged regularly with plane area density of 0.25. In addition to the conventional turbulent flux estimate using the eddy correlation method, in order to precisely close the energy balance, the conductive heat fluxes of all facets in a unit area were measured using very thin (0.4mm thickness) heat plates which were carefully coated with the same material that the obstacles are made of. In this study, (a) physical scale similarities including radiation, flow and thermal inertia, (b) unknown parameters for the simple modeling of turbulent transfers and (c) basic surface layer climate and energy balance are examined. The followings are the major results obtained.

(1) The albedos of 1/5 and 1/50 scale models correlate very well, which is the evident of the radiation similarity.

(2) Local drag coefficients (CD) in the roughness sublayer at the height of 2H between 1/5 and 1/50 models correlate well (U>1m/s). It is the indirect evident of the flow similarity since neutrally-stratified fully developed turbulences have a constant CD independent of the Reynolds numbers.

(3) Surface temperature is used for the index of thermal inertia. The maximum radiative temperature of 1/5 model is lower and later than that of 1/50 model. This suggests that the thermal inertia mismatch actually exist.

(4) The direct measure of conductive heat flux allows us to estimate local transfer coefficients (CH(i)) among local facet and reference height(2H). Relative values of CH(i) normalized by roof under nearly neutral stratification showed a good agreement with the wind channel experiment.

(5) Surface energy imbalance (SEI) was observed and the magnitude of SEI decreased as the wind velocity increased. The value of SEI is about -0.3, which is close to the value observed in a forest region.

(6) The directly measured heat storage relative to the net radiation decreased as the wind velocity increased.

(7) Nocturnal temperature profiles at inflow and outflow towers revealed the so-called cross over phenomena, that is, below the internal boundary layer produced by the scale model, the temperature is higher than that in the surroundings, while above the IBL temperature is lower. This can be explained by the strong mixing due to rough surfaces.