CL 2 16 Urban climate urban heat island

CL 2. 16 Urban climate, urban heat island urban biometeorology How to obtain atmospheric forcing fields for Surface Energy Balance models in climatic studies Julia HIDALGO 1, Bruno BUENO 1, 2 and Valery MASSON 1 (1) CNRM/Meteo-France; (2) MIT, USA This 1 project is funded by the French Reserach Agency ANR (ANR-09 -VILL-0003)

Objective: • To provide representative atmospheric fields for a project related with climate change impact in future energy demand in Paris • to be used in long-term off-line simulations (2000 -2100) • performed with the SEB model SURFEX (ISBA+TEB) • atmospheric fields are based on existing future projections of climate from Regional Climate Models: ENSEMBLES; MPI; CNRM • for a variety of IPCC emissions scenarios: Nakicenovic et al. (2000) 2

Key aspects of the study: 1. Related to the temporal frequency of RCM outputs 2. Related to the urban representation in RCMs 3

Key aspects: 1. Related to the temporal frequency SEB models need: • T; q; U, V; P; PP; +3 radiatif components (Lw, dir_sw, scat_sw) • z = canopy level. Hourly frenquency Future climate projections currently available for Europe: • T mean, max, min; Hu mean, max, min; U, Vmean; Pmean ; PPmean ; RGmean RAmean • z = 2 m; time-period 1961 - 2100 Methodology: - To classify hourly observational diurnal cycles to obtain a collection of clusters that represents the diversity of weather types affecting the site. - To use it to reconstruct the future projections at hourly frequency. 4

1. Clustering past observations 10 years; 1 h; ff, dd, T, HU, P, PP 30 years; 3 h frequency; ff, dd, T, HU 1961 1990 1998 2008 • Statistic method: K-means • variables included as inputs: Tmax -Tmin, q, ff, dd, pp • Objective: (ΔT(h)mean_season)k 28070001. CHARTRES PARIS Shape: deviation to the mean value 5

1. Clustering past observations 10 years; 1 h; ff, dd, T, HU, P, PP 30 years; 3 h frequency; ff, dd, T, HU 1961 Wind rose observed for PARIS 6 1990 1998 2008

1. Clustering past observations: Validation & Reconstruction T(C) time Temperature Specific Humidity Observations Reconstruction 7

1. Clustering observations: Validation & Reconstruction T(C) time Wind force 8 Wind direction Observations Reconstruction

1. Clustering observations: Validation & Reconstruction 10 years; 1 h; 30 years; 3 h frequency; ff, dd, T, HU, P, PP 1961 1990 1998 2008 Standard error r 2 T(C) q (kg/kg) ff (m/s) All points 0, 975 0, 911 0, 701 Mean 1 0, 999 0, 844 Max 0, 973 0, 978 0, 961 Min 0, 94 0, 917 0, 806 r^2 A C 9 B D 1998 -2008 Scatter plot: Tall points (A), Tmean (B); Tmax (C) and Tmin (D)

1. Clustering observations: Validation & Reconstruction 10 years; 1 h; 30 years; 3 h frequency; ff, dd, T, HU, P, PP 1961 1990 - Cluster attribution: AT, AH, (T, Hu)mean, max and min (u-v)mean wind components and PPmean - Reconstruction - Validation 1961 -1990: 2008 Standard error r 2 T(C) q (kg/kg) ff (m/s) All points 0, 975 0, 911 0, 701 Mean 1 0, 999 0, 844 r^2 T(C) q (kg/kg) wind All points 0, 975 0, 911 u Max 0, 973 0, 978 0, 961 Mean 1 1 0, 847 Min 0, 94 0, 917 0, 806 Max 0, 996 0, 873 v Min 0, 993 0, 987 0, 8518 r^2 10 1998

1. Clustering: Future projections 10 years; 1 h; ff, dd, T, HU, P, PP Obs. 30 years; 3 h frequency; ff, dd, T, HU 1961 It is the degree of fit between the model and the observations. It should be evaluated before future projections analysis. 1990 1998 2008 RCM 1961 1. 1990 Bias correction: similar method than in Déqué, M. (2007) 2. Cluster attribution: AT, AH, (T, Hu)mean, max and min (u-v)mean wind components and PPmean 3. Reconstruction 4. 11 Validation 1961 -1990 2100

1. Clustering future projections: exemple ECHAM 5 -r 3_RCA (SMHI center) 10 th, Tmin, winter Standard error r 2 • 1961 -1990 Control period + Model serie after bias correction * Hourly recontructed model serie 12

1. Related to the temporal frequency of RCM outputs 2. Related to the urban representation in RCMs 13

Key aspects: 2. Related to the urban representation in RCMs do not use urban parameterizations, so urban climate features are not included in these scenarios. 14

Key aspects: 2. Related to the urban representation in RCMs do not use urban parameterizations, so urban climate features are not included in these scenarios. Methodology: To combine UHI scaling laws with the previous hourly regional atmospheric fields to reconstruct themal spatial structure and day by day evolution. 15

2. UHI scalings: Night-time: Lu et al. 1997 Day-time: Hidalgo et al. 2008 16

Thanks for your attention! julia. hidalgo@ymail. com 17