The effect of SRTM and Corine Land Cover data on calculated gas and PM10 concentrations in WRF-Chem

Abstract

The goal of this study is to investigate the impact of the high resolution Shuttle Radar Topography Mission (SRTM) 90 m × 90 m topography data, together with the 100 m × 100 m resolution Corine Land Cover 2006 on the simulated gas and particulate matter (PM10) concentrations by WRF-Chem. We focused our analysis on the well-known highly urbanized region of the Po Valley. Large differences are found in the geographical distribution of the land cover classes between Corine Land Cover and 30 arc seconds USGS. The simulation with the SRTM and Corine Land Cover increases modelled temperature at 2 m and reduces wind speeds due to more friction at the surface induced by the Corine Land Cover. Latent and sensible heat fluxes show large differences between the two simulations and the related boundary layer development and depth. The simulation with the SRTM and Corine Land Cover favours the precipitation amount over a large of part the Alps and follows the pattern of the difference in topography between the two topography data sets. In term of air quality indicators, impacts are also large and geographical dependent. Monthly average of CO, NO and SO2 concentrations over a large part of the Po Valley are higher when using Corine Land Cover, up to ∼20, ∼50 and ∼55%, respectively. With respect to PM10, the impacts are also geographical dependent. Over the Po valley area, calculated PM10 concentrations are in general higher using Corine Land Cover (up to 6.7 ug/m3 [∼26%] westerly of Milan) while differences are smaller over the Alps (∼0.25ug/m3 [∼20%]). Although the scope of this work is not to evaluate the model performance in calculated meteorological parameters and gas and PM10 concentrations, calculated values by the simulation with SRTM and Corine Land Cover show a better agreement with the observations than the simulation with the USGS topography and land cover data sets. A quantitative comparison between modelled and observed monthly average PM10 concentrations shows that both simulations underestimate the observed PM10 concentrations by a factor ∼4. The agreement is much better during episodes for the simulation with the SRTM and Corine Land Cover. For CO, SO2 and NOx, the modelled monthly mean concentrations are similar for the two simulations. Larger differences are found during some episodes and regions with the SRTM and Corine LC simulation being in better agreement with the observations.