Outdoor scale model experiment to evaluate the spatio-temporal variability of urban surface temperature F. Meier* 1 , J. Richters 1 , D. Scherer 1 , A. Inagaki 2 , M. Kanda 2 & A. Hagishima 3 1 Chair of Climatology, Department of Ecology, Technische Universität Berlin, Germany 2 Department of International Development Engineering, Tokyo Institute of Technology, Japan 3 Interdisciplinary Graduate School of Engineering Sciences, Kyushu University, Japan fred.meier@tu-berlin.de http://www.klima.tu-berlin.de References Kanda, M. (2006): Progress in the scale modeling of urban climate. Theoretical and Applied Climatology, 84, 23-34. Meier, F., Scherer, D. & Richters, J. (2010): Determination of persistence effects in spatio-temporal patterns of upward long-wave radiation flux density from an urban courtyard by means of time-sequential thermography. Remote Sensing of Environment, 114, 21-34. Richters, J., Meier, F. & Scherer, D. (2009): Analysis of long-wave radiation from urban facets derived from time-sequential thermography (TST) and 3D city model. 7 th International Conference on Urban Climate, 29 June - 3 July 2009, Yokohama, Japan. Introduction Knowledge about the physical processes that contribute to urban micro-scale variability of surface temperature is derived from field studies conducted in real cities using ground-based, aircraft or satellite measurement techniques. These full-scale observations though require high experimental effort. An additional approach is the use of an outdoor scale model. It reduces the efforts of experimental control, but data incorporate natural atmospheric forcing. An advantage is the interpretability of results due to the use of uniform geometry and known thermo-physical properties of materials. Additionally, it provides an approach to evaluate numerical modelling results by using scale model properties as input parameters (cf. Kanda 2006). We present spatio-temporal patterns of surface temperatures derived from time- sequential thermography (TST) measurements at the COSMO (Comprehensive Outdoor Scale MOdel) experimental facility build up on the campus of the Nippon Institute of Technology, Saitama prefecture, Japan (39°01’N, 139°42’E). Experimental Setup The experimental setup consist of a thermal-infrared (TIR) camera mounted on a tower, various thermocouples directly attached at the walls and roofs of selected cubes inside the field of view (FOV) of the TIR camera and ultrasonic anemometers installed directly above the roof, above the ground and between the cubes. The model consists of cubic concrete blocks of 1.5 m edge length and the thickness of walls is 0.1 m. Results and Discussion At first, we present four spatio-temporal patterns of surface temperature derived from a 24-hourly TST measurement using a recording frequency of one image per minute. The Standard Deviation (σ) pattern, the Skewness Pattern (SP) and the Kurtosis Pattern (KP) show strong intra-facet variability, partly in form of regular patterns, which will be correlated to surface geometry parameters e.g. sky-view factor in future studies. Particularly noticeable in the Mean Pattern (MP) are the cube edges showing lower values and the south oriented upper wall parts showing higher values. These intra-facet variability was further analysed. Methods We developed a method to match 3D surface models and spatial distribution of surface temperatures derived from TIR images. In order to fit the central perspective projected TIR image on the 3D vector model, based on the regular geometry of COSMO, the model must be oriented into the viewing position of the TIR camera. By just click on a vector facet, the software extracts the corresponding pixels in the TIR image. For details on the method and pre-processing, radiometric and geometric correction steps of TIR imagery, please see Richters et al. (2009) and Meier et al. (2010). Here, an emissivity value of 0.9 and no atmospheric correction were applied. s n i s T T n MP - = ∑ - = 1 0 1 ( ) ∑ - = - - = 1 0 2 1 1 n i s s T T n σ ∑ - =         - = 1 0 3 1 n i s s Variance T T n SP ∑ - =         - = 1 0 4 3 - 1 n i s s Variance T T n KP Intra-facet variability of surface temperature In order to examine intra-facet variability a spatial σ of Ts for single facets (colored cube in 3D model, left) were computed for every measurement. The irradiated south oriented facet unexpectedly shows lower spatial variability than east oriented facet for the high frequency run (left graph). Also at night intra-facet variability of Ts is present. 04.07.2009 19:39:12 – 05.07.2009 19:38:12 2.21 K (max) 1.54 K (95%) -1.33 K (5%) -6.33 K (min) Atmospheric Effects The Mean Pattern (MD) shows a continuously decrease of surface temperature with increasing distance to TIR camera position. The atmospheric path length and analysed roof pixels are derived from the 3D model. Though overall short path lengths at COSMO, differences up to 5 K are present in the daytime example. High variability of Ts also results from pixels which not correspond to roof facets. Thus the extraction of TIR image pixels for specific facets of the 3D vector model has to be improved. 04.07.2009 19:39:12 – 05.07.2009 19:38:12 6.02 K (max) 5.11 K (95%) 2.49 K (5%) 1.64 K (min) N Street Axis 04.07.2009 19:39:12 – 05.07.2009 19:38:12 -0.38 K (max) -1.16 K (95%) -1.52 K (5%) -1.58 K (min) 04.07.2009 19:39:12 – 05.07.2009 19:38:12 0.73 K (max) 0.52 K (95%) -0.01 K (5%) -0.08 K (min) 07.07.2009 08:50:12 – 05.07.2009 19:38:12 8.33 K (max) 6.5 K (95%) -4.57 K (5%) -10.7 K (min) 51.2 m (max) 36.0 m (95%) 6.94 m (5%) 0 m (min) N MD (24h) MD (day) MD (night) Atmospheric path length COSMO photograph showing approx. FOV and TIR camera MD pattern 3D model and selected facets for intra-facet variability analysis 96 m North Street Axis 41.15° 48 m Tower TIR Camera FOV Thermocouple Type T Roof & floor Wall h=1,5m Ultrasonic Anemometer 1h 2h 61° 96 m North Street Axis 41.15° 48 m Tower TIR Camera FOV Thermocouple Type T Roof & floor Wall h=1,5m Ultrasonic Anemometer 1h 2h 61° blue = roof red = south oriented wall yellow = east oriented wall TST recording frequency = 1 Hz TST recording frequency = 1 / minute 15.7 m (Mean)