A study of the Athens Urban Heat Island effect during the 2009 THERMOPOLIS campaign K. Kourtidis 1 , A. Georgoulias 1 , S. Rapsomanikis 1 , I. Keramitsoglou 2 , I.A. Daglis 2 , P. Manunta 3 1 Dept. of Environmental Engineering, Demokritus Univ. of Thrace, Xanthi, Greece, 2 Institute of Space and Remote Sensing, National Observatory of Athens, Athens, Greece, 3 Planetek Italia S.A., Bari, Italy Introduction Measurements of air temperature in the urban canopy layer were performed during July 2009 in 29 sites in Athens (Fig. 1). These allowed for the mapping of the daily spatiotempo- ral evolution of the Urban Heat Island in Athens. City dis- tricts to the east were the hottest during the afternoon, while being among the coolest during the early morning hours. While during the early morning some coastal sites were the hottest, the “heat plume” slowly moved to the center of the city until 14:00 – 15:00 moving then further east during the afternoon. Satellite-derived Land Surface Temperature (LST) data from AATSR, ASTER, AVHRR and MODIS for the pixels corresponding to the Tair ground stations showed that LST can be up to 5 K lower than the respective Tair (during late afternoon/nighttime), while it can be up to 15 K higher (during the rest of the day). For each station, timeseries of Tair and the corresponding LST timeseries (i.e. the LST for the satellite pixel where each station lies within) show that generally late afternoon AAT- SR LST acquisitions agree very well with Tair for all sta- tions and all days, i.e. for Athens the AATSR LST afternoon retrieval is a very good approximation of Tair. The same holds for the AVHRR LST late afternoon acquisitions. The MODIS late afternoon/early morning LST data agree also fairly well (although not as good as AATSR and AVHRR) with Tair for most stations and most days, in all cases the difference being < 4 K. Results Foolowing the classification of all sites (see example in the Appendix), to obtain T air UHI maps, a spatial interpolation of the 29 station data was performed using the Delaunay trian- gulation (Delaunay, 1934). To test the triangulation results, the mean spatial T air was computed using the Delauney tri- angulation for each hour of the day for three cases (Fig. 2): 1. using measurements at all 29 available stations. 2. Using only 10 stations in the center, and finally 3. using only 16 stations. Cases 1 and 3 gave almost identical results. Case 2 gives results only for the central part, since this is where these stations were located, and despite the fact that in this Case much more stations are located in the center, the results for this part of Athens are almost identical with Case 3 where very few stations are located in the center. This gives further credibility to the assumption of a very good interpolation. In fig. 2, results of this interpolation exercise for 00, 06, 12, and 18 hrs are presented. Fig. 3 shows the spatiotemporal evolution of T air UHI. Figure 1. Map of measuring sites used for urban canopy air tem- perature (Tair) analyses in the Greater Athens Area. Figure 2. Results of the interpolation for 00, 06, 12 and 18 hrs us- ing different station data as input. Mean diurnal variation of Tair Mean diurnal variation of spatial Tair features for the Athens area during the THERMOPOLIS 2009 campaign. All stations (29) with available data were used. Time is 00 hrs at the upper left panel. Time proceeds with 1-hr step from left to right and from top to bottom. T air (in situ data)-LST (satellite data) differences Satellite acquisition data for the pixels corresponding to the T air ground stations were retrieved from four sensors. For each station, timeseries of T air and the corresponding LST timeseries (i.e. the LST for the satellite pixel where each sta- tion lies within) were plotted (Fig. 4). Generally, it can be said that 1. Late afternoon AATSR LST acquisitions agree very well with Tair for all stations and all days, i.e. for Athens the AATSR LST afternoon retrieval is a very good approxima- tion of T air . 2. The same holds for the AVHRR LST late afternoon ac- quisitions. 3. The MODIS late afternoon/early morning data agree also fairly well (although not as good as AATSR ande AVHRR) with T air for most stations and most days, in all cases the discrepancy being < 4 K. It follows from the above that it might be possible to recon- struct the spatial evolution of the daily course of the T air in Athens from afternoon AVHRR observations (or AATSR, although at this case there is no daily coverage) IN THE CASE that a robust statistical relationship exists between af- ternoon Tair and Tair in other times of the day. Figure 3. Timeseries of Tair and concurrent satellite acqui- sition data for the pixels corresponding to the Tair ground stations.Figure 3. Timeseries of Tair and concurrent satellite ac- quisition data for the pixels corresponding to the Tair ground stations. References Davenport A.G., C.S.B. Grimmond, T.R. Okeand J. Wieringa (2000), Estimating the roughness of cities and sheltered country, Proc. 12th Conf. on Applied Climatology, Asheville, NC, American Meteorological Society, Boston, pp. 96-99. Delaunay B. (1934), Sur la sphère vide, Izvestia Akademii Nauk SSSR, Otdelenie Matematicheskikh i Estestvennykh Nauk, 7:793–800. Oke T.R. (2006), Initial Guidance to obtain representative meteorological observations at urban sites, Instruments and observing methods Report No. 81, WMO/TD-No. 1250, World Meteorological Organisation, Geneva. Wieringa J. (1992), Updating the Davenport roughness clas- sification, Journal of Wind Engineering and Industrial Aero- dynamics, 41-44, pp. 357-368. APPENDIX Station id: DUTH_001, Address: 12 Thaleias str., Date of update: 30.7.2009 Local Scale Microscale Example of station classification files.