Impact of WRF urban parameterizations in the Tar-industrial ------2 Tecnologías Ambientales y performance of CMAQ on 1 km resolution Recursos industriales http://tarindustrial.etsii.upm.es annual runs in () aDavid de la Paz, aRafael Borge, bAlberto Martilli

a Chemical & environmental engineering department. (Technical University of Madrid, Spain. ([email protected]) b CIEMAT (Research Centre for Energy, Environment and Technology), Madrid, Spain Abstract Eulerian 3D mesoscale models can consistently describe a wide range of spatial scales, from continental to urban scale. However, urban areas present features that are usually missed by land-surface and PBL modules commonly implemented in those models. Meteorological models such as the Weather Research and Forecasting model (WRF) incorporate urban parameterizations to take into account changes in albedo, roughness length and thermal properties imposed by buildings. In this contribution, two different urban canopy models implemented within WRF are tested in Madrid and the impact on high-resolution annual air quality simulations through the Community Air Quality Model (CMAQ) is assessed.

Lambert conformal projection (α=20°N, β=60°, ϒ=3°W) Methodology Lambert Conformal projection (α=20ºN, β=60ºN, γ=3ºW) Four nested domains were used for the WRF (Skamarock and Klemp, 2008) simulations, with a resolution up D4 WRF – 1 Km to 1 km2 over the (Figure 1) with the general model set-up used in Borge et al. (2008). The urban parameterizations tested are the default urban model within the Noah Land Surface Model (BULK)(Liu et al., 2006) (a bulk-transfer scheme with increased surface roughness length and Alcobendas Torrejon de Ardoz reduced evapotranspiration for urban surfaces) and a more complex multi-layer canopy parameterization Madrid that takes into account the urban drag force, heating and turbulent kinetic energy production and Leganes D3 WRF – 4 Km GetaGetafefe dissipation induced by the urban canopy, the Building Energy Parameterization (BEP) (Martilli et al., 2002). CMAQ – 1 KKmm D2 WRF – 1616 Km Models outputs were compared with observations from 6 meteorological and 36 air quality stations D1 WRF – 48 Km throughout the Madrid urban area, representative of different urban morphologies. Different land use covers were processed to produce the urban classes considered in the parameterization (Figure 1). Figure 1. WRF model domains and urban classes Meteorology Air quality A complete annual run was performed with both, BEP parameterization WRF outputs were used to feed the CMAQ chemical-transport model. To and BULK scheme. Some of the most influential variables from the air assess the implications of WRF configuration in air quality, the quality point of view (such as temperature and wind speed) were corresponding outputs were compared with observations of NO2 and PM2.5 compared with representative measurements (Figure 2). from 26 monitoring stations throughout the modelling domain (Figure 3). EvaluationEvaluation Evaluation URBAN [BEP] – BASE [BULK] URBAN [BEP] – BASE [BULK] a b a b 2 NO Temperature(°C) Temperature(°C)

1 2,5 2,5 PM Wind (m/s) speed

1 Figure 2. a) Differences T2, WSpeed; b) Evaluation with observations Figure 3. a) Differences NO2, PM2.5; b) Evaluation with observations

§ T2m. BEP yields higher surface § Wind Speed. Wind speed is § NO2 BEP brings about an increase § PM2.5 Improvements up to 2 temperature (0.5 to 1.2K), mostly reduced by 0.8 m/s as an average in of 10-18 µg/m3 in the annual mean µg/m3 in the annual mean are due to higher predictions in winter the city centre and up to 3 m/s in within the city. Average NO2 bias of observed in the high-density and nightime. BEP predictions are the city outskirts. BEP brings about the model is reduced to 1 μg/m3. innermost part of the city. Despite worse in all cases with an overal a clear improvement of wind fields Better IOA is achieved for most of improving biases and errors, also a bias of 1.7K (0.7K for the BULK (up to 1.6 m/s compared to those of the monitoring stations, mainly better index of agreement. reference model) the BULK parameterization). those in high density urban areas. Conclusions References § BEP application significantly improves wind speed predictions in the city. § Borge, R., Alexandrov, V., del Vas, J.J., Lumbreras, J., Rodríguez, M.E., § Wind speed can be pointed out as a key variable for AQ modelling in 2008. A comprehensive sensitivity analysis of the WRF model for air urban areas, since this improvement has a strong positive effect on CMAQ quality applications over the Iberian Peninsula. Atmospheric Environment predictions (NO , O and PM ) 2 3 2.5 42, 8560–8574. § A realistic representation of the interaction between the urban canopy and the atmosphere improves meteorological forecasting and substantially § Liu, Y., Chen, F., Warner, T., Basara, J., 2006. Verification of a mesoscale improves the skills of Eulerian 3D air quality models data assimilation and forecasting system for the Oklahoma city area during the joint urban 2003 field project. Journal of Applied Meteorology Acknowledgements 45, 912–929. § Madrid City Council for providing the meteorological § Martilli, A., Clappier, A., Rotach, M.W., 2002. An urban surface exchange and air quality data. parameterisation for mesoscale models. Boundary Layer Met. 104, 261-304. § The TECNAIRE-CM scientific programme is funded by § Skamarock, W.C. and Klemp, J.B., 2008. A time-split non-hydrostatic the Directorate General for Universities and Research of atmospheric model. Journal of Computational Physics 227, 3465–3485. the Greater Madrid Region (S2013/MAE-2972) 113