17th Task Force on Measurement and Modelling Meeting

Slides:



Advertisements
Similar presentations
Norwegian Meteorological Institute met.no EMEP data on the web SR Tables National reports Michael Gauss.
Advertisements

© Crown copyright Met Office Regional/local climate projections: present ability and future plans Research funded by Richard Jones: WCRP workshop on regional.
Title EMEP Unified model Importance of observations for model evaluation Svetlana Tsyro MSC-W / EMEP TFMM workshop, Lillestrøm, 19 October 2010.
 Similar picture from MODIS and MISR aerosol optical depth (AOD)  Both biomass and dust emissions in the Sahel during the winter season  Emissions.
The robustness of the source receptor relationships used in GAINS Hilde Fagerli, EMEP/MSC-W EMEP/MSC-W.
Investigating the Accuracy and Robustness of the Icelandic Cod Assessment and Catch Control Rule A. Rosenberg, G. Kirkwood, M. Mangel, S. Hill and G. Parkes.
On the Model’s Ability to Capture Key Measures Relevant to Air Quality Policies through Analysis of Multi-Year O 3 Observations and CMAQ Simulations Daiwen.
Evaluation of sulfate simulations using CMAQ version 4.6: The role of cloud Chao Luo 1, Yuhang Wang 1, Stephen Mueller 2, and Eladio Knipping 3 1 Georgia.
Air Quality trend analyses under EMEP/TFMM and link to EEA work Augustin COLETTE (INERIS), Chair of the TFMM/CLRTAP TFMM National Experts, CCC, MSC-E,
AEROCOM AODs are systematically smaller than MODIS, with slightly larger/smaller differences in winter/summer. Aerosol optical properties are difficult.
MBF1413 | Quantitative Methods Prepared by Dr Khairul Anuar 8: Time Series Analysis & Forecasting – Part 1
17 th TFMM Meeting, May, 2016 EMEP Case study: Assessment of HM pollution levels with fine spatial resolution in Belarus, Poland and UK Ilia Ilyin,
Effects of Word Concreteness and Spacing on EFL Vocabulary Acquisition 吴翼飞 (南京工业大学,外国语言文学学院,江苏 南京211816) Introduction Vocabulary acquisition is of great.
Yuqiang Zhang1, Owen R, Cooper2,3, J. Jason West1
Clinical practice involves measuring quantities for a variety of purposes, such as: aiding diagnosis, predicting future patient outcomes, serving as endpoints.
Describing Scatterplots
Meteorological drivers of surface ozone biases in the Southeast US
Global Impacts and Consequences of Climate Change
Joint thematic session on B(a)P pollution: main activities and results
Adverse Effects of Drought on Air Quality in the US
CHAPTER 7 Sampling Distributions
Statistics for the Social Sciences
Overview of Downscaling
PCB 3043L - General Ecology Data Analysis.
Analysis of Covariance (ANCOVA)
Simulating daily precipitation variability.
Starter: complete the research methods paper
(Residuals and
Monika Kopacz, Daniel Jacob, Jenny Fisher, Meghan Purdy
AQMEII3: the EU and NA regional scale program of the Hemispheric Transport of Air Pollution Task Force The AQMEII 3 modelling team S. Galmarini, C. Hogrefe,
Svetlana Tsyro, David Simpson, Leonor Tarrason
MBF1413 | Quantitative Methods Prepared by Dr Khairul Anuar
Fig. 1 Area over which VIC simulated soil moisture has been spatially averaged. Blue shadded area represents contributing area above Sacramento at Bend.
Validation of mesoscale generalisation procedure for the WRF model
INSTYTUT METEOROLOGII I GOSPODARKI WODNEJ
Fig. 1 Area over which VIC simulated soil moisture has been spatially averaged. Blue shadded area represents contributing area above Sacramento.
POPs and HMs Summary , EMEP TFMM.
A. Aulinger, V. Matthias, M. Quante, Institute for Coastal Research
Alexey Gusev, Victor Shatalov, Olga Rozovskaya, Nadejda Vulyh
16th Task Force on Measurement and Modelling Meeting:
MSC-E: Alexey Gusev, Victor Shatalov, Olga Rozovskaya, Nadejda Vulykh
Multi-model and Observed PM Trends
TFMM trend analysis: use of AirBase. Preliminary results
EMEP Case study: Assessment of HM pollution levels with fine spatial resolution in Belarus, Poland and UK Ilia Ilyin, Olga Rozovskaya, Oleg Travnikov.
WRAP Modeling Forum, San Diego
EURODELTA 3 – Trend Analysis
REGIONAL AND LOCAL-SCALE EVALUATION OF 2002 MM5 METEOROLOGICAL FIELDS FOR VARIOUS AIR QUALITY MODELING APPLICATIONS Pat Dolwick*, U.S. EPA, RTP, NC, USA.
Trend analysis of sulphur and nitrogen species in air and precipitation in Sweden Martin Ferm
Visualization of first (Model - to – Observations/Emissions)
Title Inorganic PM at selected sites during intensive period 2008:
EURODELTA Preliminary results
30/06/2019 Trends in measured and modelled precipitation quality parameters in Croatia Sonja Vidič Meteorological and Hydrological Service.
Uncertainties in atmospheric observations
Trends in sulphur and nitrogen components
Wenche Aas, Hilde Fagerli, Svetlana Tsyro, Sverre Solberg
The EuroDelta inter-comparison, Phase I Variability of model responses
PM Trends PM10, PM2.5, PM10-PM2.5 (not possible in 90s)
Title Why do we underestimate Elemental Carbon in PM?
19th TFMM Meeting, Geneva May 3rd 2018
PM10 trends in Switzerland using random forest models
Trend analysis for HMs and POPs
Maarten van Loon and Leonor Tarrasón (met.no/EMEP)
Update on activities of Bulgaria within/related to EMEP
A Visualization/Analysis Tool for Model - to – Observations/Emissions
based on EMEP/MSC-W model and EMEP monitoring data
MGS 3100 Business Analysis Regression Feb 18, 2016
Svetlana Tsyro, David Simpson, Leonor Tarrason
Modelling of BaP concentrations over France.
Summary of discussion (1)
Discussant Suresh Chand Aggarwal University of Delhi, India
Presentation transcript:

17th Task Force on Measurement and Modelling Meeting Deposition trends in Europe for the period 1990-2010, in the framework of the EURODELTA III project M. Theobald, M. Garcia Vivanco, A. Colette, F. Couvidat, K. Cuvelier, B. Bessagnet, A. Manders, M. Mircea, K. Mar, M.T. Pay, V. Raffort, L. Rouïl, Y. Roustan, S. Tsyro, Please revise author list Thanks to Nilu for observations (W. Aas) 17th Task Force on Measurement and Modelling Meeting Utrecht, 18-20 May , 2016

Observed/modelled trends in wet deposition for the period 1990-2010 Outline Aims of the study Observed/modelled trends in wet deposition for the period 1990-2010 Model performance assessment Relationship with atmospheric concentrations and precipitation Influence of meteorological variability Conclusions

Can the models reproduce the observed trends in wet deposition? Questions to answer Can the models estimate wet deposition for the period 1990-2010 with sufficient accuracy? Can the models reproduce the observed trends in wet deposition? What are the reasons for the differences between the models and between modelled and observed trends?

Datasets Used + Simulations with constant emissions (2010) Variables (annual, seasonal) Wet deposition of oxidised N (WNOx) Wet deposition of reduced N (WNHx) Wet deposition of S (WSOx)* Total oxidised N = HNO3 + NO3-10 (TNO3) Total reduced N = NH3 + NH4-10 (TNH4) Total sulphur = SO2 + SO4-10 (TSO4) * Precipitation * Not including sea-salt sulphate Simulations (1990-2010) Chimere (CHIM) EMEP MSC-W (EMEP) Lotos-Euros (LOTO) Simulations (1990, 2000, 2010) Chimere (CHIM) CMAQ (CMAQB) EMEP MSC-W (EMEP) Lotos-Euros (LOTO) MINNI Polair3D (POLR) WRF-Chem (WRFC) Observations (EMEP sites: 1990-2010) Wet deposition (41 sites) Precipitation (41 sites) Concentrations (29 sites) Sulphur measurements are sea-salt sulphate corrected Observations provided by Wenche (NILU) WRF-Chem is participating in the exercise but the data have not been included in this analysis + Simulations with constant emissions (2010)

21 year time series (1990-2010) – Mean of all sites By season WNOx (mg N m-2 yr-1) Observed (green) and modelled wet deposition averaged over all measurement sites Line: Mean, Shaded: +/- one standard error EMEP > CHIM > LOTO EMEP similar to observed (especially in Spring, not so in Winter). LOTO significantly lower (especially in Winter)

21 year time series (1990-2010) – Mean of all sites By season WNHx (mg N m-2 yr-1) EMEP similar to observed except for period 1990-1995 EMEP > LOTO > CHIM CHIM similar to LOTO in Winter

21 year time series (1990-2010) – Mean of all sites By season WSOx (mg S m-2 yr-1) EMEP > LOTO > CHIM EMEP starts higher than OBS but the difference gets smaller with time (2010 almost equal) Largest obs – model difference in Winter

21 year time series (1990-2010) – Trends Conditions: Observations for > 75% of period Trend calculation: Sen’s slope method Relative trend = Trend/Estimated value at start of period Trend significance: Mann Kendall method (p < 0.05) e.g. Randomly chosen site (DE002R) Linear trends have been assumed (might not be valid for WSOx) Need to point out that there are two trends: the absolute trend (in mg N (or S) /m2/yr) and relative trend (in %/yr) We have focussed on the relative trends although a complete analysis should consider both Relative trends (yr-1) -2.7% (***) -2.1% (***) -2.5% (***) -0.9% (ns)

Spatial Distributions of Measured and Modelled Trends - WNOx Chimere EMEP MSC-W Lotos-Euros Observed: About a third of trends are significant (one with increasing trend, FROO9R, Revin) Modelled: Significant decreasing trends (except in the Mediterranean area) All three models similar (although have individual features, e.g. larger CHIM trends in Germany)

Spatial Distributions of Measured and Modelled Trends - WNHx Chimere EMEP MSC-W Lotos-Euros Observed: about a quarter of trends are significant although about half of of those are very small (+/- 1%/yr) Modelled: Mostly insignificant. All models predict a decrease in the East. CHIM predicts more significant increasing trends than the other models

Spatial Distributions of Measured and Modelled Trends - WSOx Chimere EMEP MSC-W Lotos-Euros Observed: Nearly all sites have significantly decreasing trends Modelled: CHIM – smaller trends than EMEP/LOTO and more insignificant trends in Mediterranean. EMEP – larger trends than LOTO

21 year time series (1990-2010) – Relative Trends Box plots of all relaltive trends (significant and insignificant) WNOx: Most modelled trends larger than observed WNHx: EMEP predicts similar distribution of trends to observed, LOTO predicts small trends and CHIM predicts increasing trends! WSOx: Models and observations have decreasing trends although EMEP/LOTO overestimate most trends

Model Performance (All 6 models) – WNOx (1990, 2000, 2010) Moving onto the analysis with the six nodels (1990, 2000, 2010 data only). For WNOx: Normalised Mean Bias: All models underestimate deposition rates, on average (EMEP by only a few %) Normalised Mean Gross Error: Most of the model error is due to the negative bias (except for CMAQB and EMEP)

Model Performance (All 6 models) – WNHx (1990, 2000, 2010) Normalised Mean Bias: All models underestimate deposition rates, on average Normalised Mean Gross Error: Most of the model error is due to the negative bias (except for EMEP) POLR underestimates deposition rates for two of the years (1990 and 2000)

Model Performance (All 6 models) – WSOx (1990, 2000, 2010) Normalised Mean Bias: All models underestimate deposition rates, on average, except for EMEP (30% overestimate) Normalised Mean Gross Error: Model error is mostly due to the bias (except for CMAQB)

Change (1990-2010) With only 3 years of data we cannot calculated the trends and so we have calculated the change in deposition between 1990 and 2010 WNOx: Both observations and models predict a general decrease in deposition rates although the models tend to overestimate trends (except for EMEP)

Change (1990-2010) Without POLR WNHx: POLR predicts large increasing trends as a result of the underestimation for 1990 Observed trends are mostly decreasing. CMAQ predicts a similar trend distribution to that observed, EMEP, LOTO and MINNI underestimate trends and CHIM predicts mostly positive trends

Change (1990-2010) All trends (modelled and observed) are negative Some models predict larger trends than observed (EMEP, LOTO and MINNI), some similar (CHIM and CMAQB) and POLR predicts smaller trends than those observed

Wet deposition is influenced by (among other variables): Why do the modelled deposition rates differ between the models and between the models and the observations? Wet deposition is influenced by (among other variables): Atmospheric concentrations and precipitation rates We have carried out a similar trend analysis for the atmospheric concentrations of total oxidised and reduced nitrogen, and sulphur in order to look at the relationship between the model estimates of wet deposition and the model estimates of the corresponding atmospheric concentrations TNO3 TNH4 TSO4

Wet deposition & Concentrations – Bias Compared Oxidised N Reduced N Sulphur Wet deposition Concentrations Although we’re not going into detail of the model performance for concentrations (no time), we show here the model biases for both the wet deposition (top row) and atmospheric concentrations (middle row). Generally, models underestimate wet deposition (except EMEP for WSOx) and overestimate concentrations Therefore the overestimate of concentrations could be due to an underestimate of wet deposition (although other processes, such as dry deposition could also be important) However, it is not the case that the models that underestimate deposition by the largest amount overestimate the concentrations by the largest amount, although this does seem to happen for WNHx (see scatter plots in bottom row) Note CMAQ may include sea-salt sulphate, which may explain the high bias in TSO4 (but won’t explain the high bias in the other components)

Precipitation (mm yr-1) e.g. CHIMERE Another explanation for an underestimate of wet deposition could an underestimate of precipitation. However, as this plot shows, although there is a slight (7%) underestimate in precipitation in CHIMERE, this is not sufficient to explain a 20-70% underestimate in deposition (although it will contribute)

Influence of Meteorological Variability Simulations with constant emissions (2010) Mean of all sites Finally, in order to discount the effect of meteorological varibility on the long-term trends, we have also done the trend analyses for the simulations with constant (2010) emissions. As these plots show, the mean modelled deposition rate (averaged over all measurement sites) does not change significantly during this period Note these are the old CHIM data. These simulations have not been done with the new version yet 25

Influence of Meteorological Variability Simulations with constant emissions (2010) Relative trends (1990-2010) ..and as the box plots show: WNOx: Models estimate mostly small increasing trends (significantly different to observed trends) WNHx: Models estimate mostly small increasing trends (significantly different to observed trends) WSOx: Models estimate mostly small increasing and decreasing trends (significantly different to observed trends)

Conclusions The models tend to underestimate rates of wet deposition (EMEP by only a few % for WNOx, and overestimates WSOx) ; and overestimate atmospheric concentrations WNOx Trends: Models tend to overestimate the observed relative decreases in deposition (modelled trends are more significant) WNHx Trends: Observed and modelled relative trends smaller than those for WNOx (trends are also less significant) WSOx Trends: Similar ranges for modelled and observed relative trends (the majority of which are significant) Meteorological variability appears to have only a small influence on modelled trends

Can the models reproduce the observed trends in wet deposition? So: Questions to answer Can the models estimate wet deposition for the period 1990-2010 with sufficient accuracy? Can the models reproduce the observed trends in wet deposition? What are the reasons for the differences between the models and between modelled and observed trends?

Thank you

I suggest we end here!