Uncertainties of heavy metal pollution assessment Oleg Travnikov EMEP/MSC-E
Outline Sensitivity and uncertainty analysis Models intercomparison Model results vs. measurements Back trajectory analysis Emission reporting for model application
Model sensitivity analysis Pb and Cd total deposition Hg total deposition
Model intrinsic uncertainty without effect of emissions Model uncertainty Model intrinsic uncertainty without effect of emissions Lead and cadmium Mercury Uncertainty 30-40% Uncertainty 20-50%
Workshop on review of MSC-E models on HMs and POPs (Moscow, 2005) Purpose to establish whether MSC-E models on HMs and POPs are state of the art and fit for the purpose of evaluating long-range transport of HMs and POPs. Conclusions [ ECE/EB.AIR/GE.1/2006/4 ] The model parameterization is appropriate for operational modelling of heavy metal concentration and deposition in Europe HM depositions, concentrations and transboundary fluxes of HMs calculated by MSC-E model corresponded well with other transport models Other models, such as the MSC-E model, underestimated air and precipitation concentrations of Pb and Cd when using official emission data
Model intercomparison Cd air concentration (2000) MSCE-HM CMAQ Comparison of MSCE-HM and CMAQ models for Pb and Cd CMAQ (Community Multi-scale Air Quality model) – 3D chemical transport model developed in US EPA www.cmaq-model.org Conditions of comparison: Anthropogenic emissions based on official and ESPREME data Identical meteorological data for 2000 Similar initial and boundary conditions
Model intercomparison Annual mean Cd concentration in precipitation (2000) based on official data MSCE-HM = 0.32 Obs CMAQ = 0.32 Obs 70% underestimation based on ESPREME data MSCE-HM = 0.62 Obs CMAQ = 0.68 Obs 30-40% underestimation
Wind re-suspension of HMs HM re-suspension scheme: Parameterization of mineral dust suspension [Marticorena and Bergametti, 1995; Alfaro and Gomes, 2001; Gomes et al., 2003] Parameterization of sea salt aerosol production [Monahan et al., 1986; Gong, 2003] Detailed soil properties data [ISLSCP (Initiative II), http://islscp2.sesda.com] Measured HM content in soil [FOREGS, Salminen et al., 2005] Total emission and re-suspension of Pb in Europe (1990-2005) Contribution of Pb re-suspension: 20% in 1990, 60% in 2005
Evaluation vs. observations Annual mean concentration in precipitation based on official emissions data (2005) Lead Cadmium Mod = 0.70 Obs Corr = 0.57 Mod = 0.45 Obs Corr = 0.51 Pb Cd 20-30% underestimation 30-50% underestimation
Analysis of discrepancies Daily mean Cd concentration in air (2005) Svratouch, Czech Republic (CZ1) 06.11.2005 01.09.2005
Analysis of discrepancies Monthly mean Cd concentration in precipitation (2005) Density of back trajectories February 2005 FI93 Kotinen, Finland (FI93) Neuglobsow, Germany (DE7) Cd emissions in 2005 FI93
Analysis of discrepancies Monthly mean Cd concentration in precipitation (2005) FI93 March 2005 Density of back trajectories Kotinen, Finland (FI93) Cd emissions in 2005 FI93
Evaluation vs. observations Hg concentration in air (2005) Mod = 0.94 Obs Hg Hg concentration in precipitation (2005) Mod = 0.78 Obs Hg
HM emissions reporting Coverage of EMEP region with emission data for Pb (2005) Reported Pb emission data for 2005: Gridded data: 23 countries Gridded sector data: 15 countries National totals: 30 countries According to submission 2007
HM emissions by sectors Cd emissions in large European countries (2005) According to submission 2007
Non-Party emission estimates Comparison of official data with non-Party estimates of Pb emissions in 2000 (TNO, ESPREME) Germany Germany According to submission 2007
Emissions data uncertainties reported by countries (Pb) Emission uncertainty Emissions data uncertainties reported by countries (Pb) Country Emission [t/y] Uncertainty Austria 14 -52% +110% Finland 24 -25% +26% United Kingdom 118 -30% +40% France 134 53% Denmark 6 266% Germany 107 ? Russia 355 Ukraine 195 …
Gaps of HM officially reported emissions data Incomplete data on emission totals Limited data on spatial distribution No data on temporal variation Scarce data on emission uncertainty … What data should be used to fill the gaps?
Summary (1) Estimated intrinsic model uncertainty is 30-40% for Pb and Cd and 20-50% for Hg Modelling results are highly sensitive to emissions data for Pb and Cd and to boundary conditions for Hg MSCE-HM and CMAQ models agree in underestimation of measurements (up to 70%) when officially reported emissions data is used Assessment of wind re-suspension allows improve agreement between modelling results and measurements Current model-to-measurement comparison demonstrates 20- 30% underestimation for Pb and 30-50% underestimation for Cd. Modelling results for Hg well agree with observations
Summary (2) Reported emission inventories for heavy metals are incomplete and of limited value in terms of model applications Procedure of the reported emission gaps filling is to be elaborated