Presentation is loading. Please wait.

Presentation is loading. Please wait.

1 WISE TG Meeting, Ispra, September 2011 Direct and Inverse Chemical Fate Modeling based on pan-European Datasets Dimitar MARINOV, Alberto PISTOCCHI, Robert.

Similar presentations


Presentation on theme: "1 WISE TG Meeting, Ispra, September 2011 Direct and Inverse Chemical Fate Modeling based on pan-European Datasets Dimitar MARINOV, Alberto PISTOCCHI, Robert."— Presentation transcript:

1 1 WISE TG Meeting, Ispra, September 2011 Direct and Inverse Chemical Fate Modeling based on pan-European Datasets Dimitar MARINOV, Alberto PISTOCCHI, Robert LOOS, Bernd GAWLIK European Commission, Joint Research Centre, Institute for Environment and Sustainability

2 2 WISE TG Meeting, Ispra, September 2011 Concept for large scale chemical modeling:  Now-a-days the availability of pan-continental datasets allows the development of spatially explicit GIS models for assessment of fate and distribution of pollutants into different environmental media (atmosphere, soil, surface water and sea)  In principle, spatial models predict chemical concentrations when the emissions to the environmental compartments at continental scale are known. This is the "direct" formulation of fate problem aiming to answer the question “where do pollutants go?“  Vice versa, when the emissions of chemicals are unknown, but their concentrations are widely monitored, the "inverse" modeling approach answers the question "where do pollutants come from?" The inverse models tracing back chemical emissions from measured concentrations and identifying possible sources of discharges.

3 3 WISE TG Meeting, Ispra, September 2011 Emissions, E Environmental removal rates, K Concentration = E/K

4 4 WISE TG Meeting, Ispra, September 2011 Datasets used in contaminant modeling : 1.Atmosphere: temperature, OH concentration, Aerosol concentration, Organic carbon content in aerosol, 10 m height wind velocity, Atmospheric mixing height, Precipitation, Duration of the wet period, Atmospheric Source-receptor relations, and Atmospheric Source-receptor time of travel 2.Soil: top soil organic carbon content, soil texture, diffusion runoff, evapotranspiration, infiltration, erosion rate, and Leaf Area Index (LAI) 3.Surface water: river discharge, river slop, river water velocity, water depth, suspended sediment concentration, and surface water residence time 4.Sea: mixing depth, seawater velocity, seawater temperature, total suspended matter, wind speed at 10m height over ocean surface, and chlorophyll

5 5 Characteristics of MAPPE direct model Main topic(s) Maps of chemical concentrations and pollutant fluxes Useful tool for screening hot spots or hazard zones Potential key client: DG ENV, DG SANCO, EEA, JRC Policy(-ies) Requirements & Application Water Framework Directive Marine Strategy Framework Directive Framework Directive on Sustainable Use of Pesticides Scenarios for horizon 2020 Scale (Time&Space) Temporal: one year Spatial: Europe Output Maps; Indicators; Model integration Other impact assessment models Challenges & Next Steps To became a transient model (currently steady state) To develop global version of the model Further integration with socio-economic models Multimedia Assessment of Pollutant Pathways in the Environment (MAPPE)

6 6 WISE TG Meeting, Ispra, September 2011

7 7

8 8

9 9

10 10 WISE TG Meeting, Ispra, September 2011

11 11 Inverse modeling approach C [ML -3 ] concentration Q [L 3 T -1 ] river water discharge E i [MT-1] chemical mass emitted at the i-th location t i [T] time spent in water from the i-th emission location to the river cross section of measurement DT50 [T] total removal half life in the stream network At steady state conditions the chemical mass (equivalently called “load”) at a given cross section is assumed to be equal to the sum of all emissions upstream each reduced by the removal occurring along the stream network

12 12 Two-parameter inverse model DT50 = 1, 3, 5, 7, 10, 50, 100 and 1000 days; DT50 = ∞ (no decay) DT50 = 1 dayDT50 = 100 days

13 13 Use of chemical monitoring data 27 countries; 122 sampling locations 16Polar organic persistent pollutants NAPROXEN KETOPROFEN BEZAFIBRATE IBUPROFEN DICLOFENAC GEMFIBROZIL BENZOTRIAZOLE CAFFEINE CARBAMAZEPINE SULFAMETHOXAZOLE METHYLBENZOTRIAZOLE NPE1C NONYLPHENOL BISPHENOL A ESTRONE OP

14 14 Inverse model verification

15 15 Mapping chemical loads to river network

16 16 Disaggregated loads to European seas 1. The average percentages of disaggregated load to European regional seas: Atlantic 18.9%; Baltic 15%; Black Sea 8.6%; Mediterranean 26.3% ; North Sea 31.2%. 2. The European discharges more heavily affect North Sea and the system of Mediterranean and Black seas

17 17 PFOS modeling PFOS annual load to sea BAU scenario (baseline 2007) [t/y]% of total Atlantic ocean Baltic sea Black sea Mediterranean sea North sea Total5.840N.A.

18 18 WISE TG Meeting, Ispra, September 2011

19 19 Aims  to test capabilities of the inverse modelling to perform back analysis of chemical emissions and half lives for pharmaceuticals, personal care products and substances widely used in households  to extend and improve the modelling approach by adding the further option for accounting the chemical decay in stream network  to map continuous continental spatial distribution of loads and concentrations in river network by generalisation of pan-European screening data from discrete points sampling campaign  to estimate the range of discharges of pharmaceuticals and personal care products to European seas

20 20 Estimation of model parameters Scatter diagrams were interpreted through best-fit linear trend line found by OLS method with zero intercept Linear regression model Measure of match: determination coefficient R 2 Non-linear regression model

21 21 Identification of half lives and emission factors The average R² correlation coefficients for all chemicals and half lives including ‘no decay’ case is 0.61 for the non-linear model versus 0.2 for the linear one linear model log-linear model

22 22 Estimates of total load to European seas 1.On average, the min extremes of total load to European seas count 40.1% less (range %) of the median estimates while the max values are 61.7% higher (range %) 2.Based on the median assessment for load coming from EU-27 plus Switzerland, Norway, Moldova, Ukraine, Belarus and part of Turkey, it was found that the total amount of all 16 polar compounds exported to European seas is about tonnes per year

23 23 Conclusions 1.The study showed that for screening assessment at continental level the concentrations of polar chemicals in river waters can be interpreted through a simple two-parameter model that assumes emissions proportional to catchment population and constant chemical half life 2.Besides, it is not possible for a chemical to select univocally a couple of emission factor and half life values. Instead that, a set of non-dominated combinations could be identified. In order to obtain more accurate estimates prior expert knowledge about emission factors or half life values is needed 3.Chemical loads are relatively insensitive to combinations of emission factors and half lives. Therefore, they can be estimated with a reasonable range of uncertainty, typically a factor 2. In this way, it is possible to produce continuous maps of loads and concentrations at continental scale 4.More generally, the inverse modeling from monitoring data supports cost- effective emission inventory of river chemical pollutants at European scale. The method could serve in chemical risk assessment or decision making in river basin management, as required by the European Water Framework Directive or Marine Strategy.


Download ppt "1 WISE TG Meeting, Ispra, September 2011 Direct and Inverse Chemical Fate Modeling based on pan-European Datasets Dimitar MARINOV, Alberto PISTOCCHI, Robert."

Similar presentations


Ads by Google