Presentation is loading. Please wait.

Presentation is loading. Please wait.

Title EMEP Unified model Importance of observations for model evaluation Svetlana Tsyro MSC-W / EMEP TFMM workshop, Lillestrøm, 19 October 2010.

Similar presentations


Presentation on theme: "Title EMEP Unified model Importance of observations for model evaluation Svetlana Tsyro MSC-W / EMEP TFMM workshop, Lillestrøm, 19 October 2010."— Presentation transcript:

1 Title EMEP Unified model Importance of observations for model evaluation Svetlana Tsyro MSC-W / EMEP TFMM workshop, Lillestrøm, 19 October 2010

2 Meteorologisk Institutt met.no Emission input ( anthropogenic): gaseous - SO x, NO x, NH 3, NMVOC, CO, particles - PM 2.5, PM 10  EMEP emission database (CEIP)‏  Yearly totals per country and per SNAP-1 sector (11)‏  Gridding is based on the reported data or by MSC-W using auxiliary data The Unified EMEP model – Eulerian 3D; describes the emissions, chemical transformations, transport and dry and wet removal of gaseous and particulate air pollutants ( 70 species, about 140 reactions)‏ Emissions natural: sea salt and wind blown dust - modelled

3 Meteorologisk Institutt met.no The Unified EMEP model Horizontal resolution: 50 x 50 km 2 (25x25, 10x10 km 2 ) ) ‏ Vertical resolution: 20 layers ( up to 100 hPa)‏ Off-line meteorology: 3-h HIRLAM (EC MWF)‏ Calculation domains Black – “old” 50x50 km2 Red – extended 50x50 km2 25x25 km2 Blue – 10x10 km2

4 Meteorologisk Institutt met.no Model structure Transport - Physics - Chemistry Initialisation, Boundary & Initial conditions Emissions – temporal variation Meteorology (3-hourly)‏ Time-step 20 min Daily/Monthly/Yearly output Advection + turbulent mixing Chemistry Dry Deposition‏ Wet Deposition Sea salt Windblown dust Hourly output

5 Meteorologisk Institutt met.no SIA SO4, NO3, NH4 Primary PM (EC, POC, dust) Mineral dust water Anthropogenic emissions Natural sources Sea salt SO2 NOx NH3 PM2.5 PM10 Atmospheric particle biogenic SOA Anthr. SOA bioaerosols

6 Meteorologisk Institutt met.no SO4 NO3NH4 EC POC Min. dust+ Sea salt PM10 PM2.5

7 Meteorologisk Institutt met.no SR calculations -15% 15% All countries 15% reduction in country A Pollution due to 15% emissions from country A x 100/15  Pollution due to A x Area_B  Pollution in B due to A Reference run RUNS: RUNS: for all countries,reduction of SOx, NH3, NOx+PM, VOC

8 Meteorologisk Institutt met.no PM2.5 in Russia due to (15%) Russian emissions TB contribution Sources of PM2.5 in RF:

9 Meteorologisk Institutt met.no TRENDS Consistent observation datasets are essential!!

10 Why should we trust the model? How can we know if calculations reproduce reality? We use observations (making an assumption they represent the reality) Meteorologisk Institutt met.no

11 Annual mean concentrations of PM in 2001 Meteorologisk Institutt met.no PM2.5 PM10 Largest in Spain Unified model EMEP obs General underestimation

12 ► EMEP: aerosol components in 2001 Meteorologisk Institutt met.no ATCHCZDEDKESFIFRHUIEISITLVNLNOPLRUSESISKTR total TSP1101416 PM10348101228 PM2.5123101219 SO4132531048131222743415172 NO312227335126 NH41222733121 Na3710 Cl77 Al11 Ca178 Mg77 ► AIRBASE (rural), PM10: - 49 sites in 2000, over 300 in 2001 (temporal coverage? chemical composition?) ► 4 Austrian stations : PM10, PM2.5, chemical composition, particle number (during June 99 - Oct 01) - urban / rural ► 3 Spanish sites: PM10, PM2.5, chemical composition (varying sampling periods and frequency, from 1999 to Aug. 2002) What is available of measurements

13 Annual mean PM10 in 2000 Meteorologisk Institutt met.no EMEP: 13 sitesAIRBASE: 49 sites Underestimation of PM10 by the model Smaller PM10 horizontal gradients (PPM? missing dust?) Bias= -31% Corr= 0.52 Bias= -44% Corr= 0.74 elevated PM10 – complex pollutant. To explain the discrepancies between calculated and measured PM10 verification of the individual components is needed.

14 Annual mean PM2.5 and PM10 (2001) Meteorologisk Institutt met.no Spain: bias= - 67%, corr= 0.44 N=17 Bias= -46% Corr= 0.61 N=25 Bias= -51% Corr= 0.15 Bias= -41% Corr= 0.59 Small modelled PM10 gradients Better prediction of PM2.5 regional gradients

15 Annual mean SIA (2001, EMEP) Meteorologisk Institutt met.no Bias= -19% Corr= 0.81 Bias= 15% Corr= 0.89 Sites without PM10 measurements These results alone cannot explain the model underestimation of PM10 and PM2.5 and too small PM10 gradients

16 Monthly series in 1999-2000 (all available EMEP) Meteorologisk Institutt met.no Better PM10 results for wrong reasons  N=78 N=27 N=20 N=13

17 What are measured PM10 and PM2.5 made of? Meteorologisk Institutt met.no

18 Averaged chemical composition of PM (UNI-AERO): Spain, rural background Meteorologisk Institutt met.no ???? Largest discrepancies: underestimation of (OC+EC) and mineral dust

19 Meteorologisk Institutt met.no

20 Summary:

21 Meteorologisk Institutt met.no 2008

22 Meteorologisk Institutt met.no Intensive measurement periods

23 Meteorologisk Institutt met.no Melpitz DE44: PM10 ECMWF – red HIRLAM - blue Model underestimation (exc. NO3, Na), more so using HIRLAM ECMWF – better correlations, slightly higher concentrations Most of episodes are reproduce with both met, models 0.7/0.570.71/0.49 0.68/0.52 0.51/0.34 0.42/0.36 0.71/0.71

24 Meteorologisk Institutt met.no EC (17 Sept – 17 Oct 2008) – Using obs + model to test for primary PM emissions

25 Meteorologisk Institutt met.no MINERAL DUST (Si, Mg, Al, Ca, K)

26 Meteorologisk Institutt met.no Still ….MINERAL DUST

27 Meteorologisk Institutt met.no In Summary…. It’s impossible to overestimate the importance on observation data of good quality for evaluation, improvement and development of models


Download ppt "Title EMEP Unified model Importance of observations for model evaluation Svetlana Tsyro MSC-W / EMEP TFMM workshop, Lillestrøm, 19 October 2010."

Similar presentations


Ads by Google