CMAS Conference 2011 Comparative analysis of CMAQ simulations of a particulate matter episode over Germany Chapel Hill, October 26, 2011 V. Matthias, A. Aulinger, M. Quante, C. Chemel, J. L. Perez, R. San Jose, R. Sokhi
Volker Matthias, Oct 26, Case study on PM10 (Feb/March 2003) Stern et al., Atm. Env. 42, (2008) PM10 daily mean concentration (µg/m 3 ) over Germany on March 2, 2003
Volker Matthias, Oct 26, COST 728 study COST 728: European cooperation, participants from more than 20 nations „Enhancing Mesoscale Meteorological Modelling Capabilities for Air Pollution and Dispersion Applications“
Volker Matthias, Oct 26, GroupMet-ModelCTM IfK, HZGMM5CMAQ Uni Hertfordshire (UH)WRFCMAQ TU Madrid (UPM)MM5CMAQ CMAQ model intercomparison
Volker Matthias, Oct 26, What are the differences? Preparation of the emissions Meteorological fields Inital and boundary conditions Grids (horizontal and vertical structure) Computing platforms People who run the model
Volker Matthias, Oct 26, Concept Round 1: All groups provide input files (IC,BC,EMIS,METEO,GRIDDESC) All groups use common CMAQ version (4.7) and chemistry mechanism (cb05_ae_aq) All groups recalculate results of others Expected results: Determination of simple (model user) errors (switches …) Quantification of computing errors (compiler, platform, …)
Volker Matthias, Oct 26, Recalculations: Sulfate WesterlandMelpitz
Volker Matthias, Oct 26, WesterlandMelpitz Recalculations: Nitrate
Volker Matthias, Oct 26, Outcome: Recalculations CMAQ model results can be reproduced by other groups on different computing platforms Depending on species, some differences exist but they are much smaller than the differences in the “blind“ runs
Volker Matthias, Oct 26, Concept (2) Round 2: Agree on common grid Use same initinal and boundary conditions (IC & BC) Expected Results influence of emissions influence of meteorological fields
Volker Matthias, Oct 26, Impact of emissions Reconstruction of UH run Emission files from UPM Sulfate Nitrate
Volker Matthias, Oct 26, Emissions in Central Europe (spatial average) NOSO 2 NH 3 UH UPM HZG Time series from CMAQ input files
Volker Matthias, Oct 26, Outcome: Emissions Emissions may be prepared in different ways concerning their temporal and spatial variation. For short time series at certain grid points this may lead to significant differences in particle concentrations.
Volker Matthias, Oct 26, Impact of meteorology MM5 from UPM MM5 from HZG WRF from UHS Sulfate Nitrate
Volker Matthias, Oct 26, Outcome: Meteorology Numerous meteorological parameters may influence particle concentrations. The quality of the CTM results may not be judged from the quality of the meteorological fields.
Volker Matthias, Oct 26, Open questions Do we see „typical“ differences between „correct“ model runs or were there important errrors in the input data? Sensitivity study: Annual runs (year 2000) with CMAQ 4.6 with different Boundary conditions Emission files Meteorological data Goal: Quantify the variability of the hourly and daily concentrations at Melpitz
Volker Matthias, Oct 26, Boundary conditions BC from global models: Mozart and TM4 SO 2 SO 4 NO 3
Volker Matthias, Oct 26, Emissions SMOKE-EU emissions and EMEP emissions Additional comparisons to other emission data sets with similar results SO 2 SO 4 NO 3
Volker Matthias, Oct 26, Meteorology MM5 (FDDA with NCEP) and CCLM (Spectral nudging with NCEP) Hourly values SO 2 SO 4 NO 3
Volker Matthias, Oct 26, CMAQ intercomparison: different emissions Nitrate values with UH emissions lower than it could be expected. SO 4 NO 3
Volker Matthias, Oct 26, CMAQ intercomparison: different meteorological fields Sulfate values with UPM and HZG meteo within expected range Nitrate values low, but may be explained by variability due to meteo input SO 4 NO 3
Volker Matthias, Oct 26, Summary CMAQ intercomparison within COST 728 showed: Simulations are reproducable by other groups on other computing platforms Emission data may be prepared in very different ways Largest influence on simulation results comes from meteorology Unreliable results may be detected by comparisons to sensitivity runs Acknowledgements Emission data has been prepared by Johannes Bieser Most CMAQ sensitivity runs were set up by Johannes Bieser Total gridded emissions were provided by TNO, IER and EMEP Boundary conditions were provided by the RETRO project (TM4) and Ulrike Niemeier (Mozart)
Volker Matthias, Oct 26, Thank you