© Crown copyright Met Office Estimating UK and NW European CH 4 emissions using inversion modelling Alistair Manning with thanks to Simon O’Doherty & Dickon Young (University of Bristol)

© Crown copyright Met Office Overview Estimate UK and NWEU (north west European) emissions of greenhouse gases (GHG) totally independent of UNFCCC inventory process. Use in-situ high-frequency atmospheric observations from the remote station on the west coast of Ireland (Mace Head). Employ an atmospheric dispersion model (NAME) coupled with 3-D meteorology to understand the recent history of the air arriving at Mace Head. Two stage process: Estimate long-term Northern Hemisphere baseline concentration. Estimate regional emissions through inversion modelling (NAME-Inversion). Compare NAME-inversion estimates to UNFCCC/EDGAR inventory estimates.

© Crown copyright Met Office Estimating Baseline Concentrations at remote measurement site (Mace Head) NAME model (Lagrangian particle dispersion model) run ‘backwards’. Uses 3-D meteorological data from UK Met Office NWP model (25-60 km resolution) and ECMWF ERA Interim (re-analysis) (~80 km). Derive air history map for each site for a 2-hour period: Combination of tens of thousands of trajectories. Darker shade means greater contribution from that area. All surface sources within previous 30 days of travel (domain limited) that contribute to an observation during a 2-hour period are recorded. Mace Head air history maps generated each 2-hour period: UK Met Office UM ECMWF ERA Interim Mace Head

© Crown copyright Met Office Where has the air come from? Examples: th March 1996 Baseline st March 1996 Polluted nd March 1996 Complicated

© Crown copyright Met Office Baseline Concentration Methodology (1): Classifying Mace Head Observations Methane: Raw observations (ppb)

© Crown copyright Met Office Baseline Concentration Methodology (2): Classifying Mace Head Observations Methane (ppb): Observations colour-coded based on recent history of air arriving at the station (Mace Head)

© Crown copyright Met Office Baseline Concentration Methodology (3): Classifying Mace Head Observations Methane (ppb): Select only those observations that are considered to be from a ‘clean’ non-regionally polluted sector.

© Crown copyright Met Office Baseline Concentration Methodology (4): Classifying Mace Head Observations Methane (ppb): Baseline observations – Statistically identify outliers (black) and remove

© Crown copyright Met Office Baseline Concentration Methodology (5): Classifying Mace Head Observations Methane (ppb): Remaining observations classed as ‘baseline’

© Crown copyright Met Office Baseline Concentration Methodology (6): Classifying Mace Head Observations Methane (ppb): Best-fit (moving window) line through baseline observations

© Crown copyright Met Office Baseline Concentration Methodology (7): Classifying Mace Head Observations Methane (ppb): Zoom-in on baseline estimate

© Crown copyright Met Office Baseline Concentration Methodology (8): Classifying Mace Head Observations Methane (ppb): Estimate baseline uncertainty (‘noise’ in baseline)

© Crown copyright Met Office Mid-latitude baseline concentration of: Methane (ppb) Baseline Growth Rate Seasonal Cycle

© Crown copyright Met Office Mid-latitude baseline concentration of: Nitrous Oxide (ppb) Baseline Growth Rate Seasonal Cycle

© Crown copyright Met Office Estimating regional emissions from above baseline concentrations

© Crown copyright Met Office Estimating Regional Emissions: Inverse Modelling Aim: Generate emission estimates from ‘polluted’ (above baseline) observations. Subtract the baseline concentration from each observation.

© Crown copyright Met Office The NAME-Inversion Method (1) M [t x m] ¤ e [mx1] = O’ [tx1] = O - b Dilution matrix Baseline Time series of observations Air history maps from Mace Head are generated using NAME model Record time-series of relative contributions of each surface source to observation station Emission map: the Solution t2t2 t3t3 t1t1

© Crown copyright Met Office Dilution matrix = M Measurement - Baseline = O’ Emission Map = e (the solution) Relationship: M e = O’ Problem: Minimise O’ - M e Remove observations that have a strong local influence. Through many iterations find emission distribution that is best-fit statistical match between model time-series and observations. No prior information – Random initial guess – Not steered by a priori. Solve for each 3-yr period stepping monthly e.g. Feb’89 – Jan’92, Mar’89 – Feb’92, … Solve multiple times, each time start from different random initial guess. Apply random ‘baseline noise’ to observations (log-normal distribution). The NAME-Inversion Method (2) Air History Map = Matrix M (N o times x N o grids)

© Crown copyright Met Office The NAME-Inversion Method (3) : Grid used to Estimate Emissions Balance equation so that each grid contributes equally (approximately). More distant grids grouped together so have same impact at the observation station. Example: Grouping of grids based on observations at Mace Head (MH) on west coast of Ireland. Grids = 25 km, 50 km, 100 km, 200 km, 400 km, 800 km

© Crown copyright Met Office The NAME-Inversion Method (4) : Resolved areas UK NWEU Mace Head

© Crown copyright Met Office Model Solutions Vs Observation ‘Best-Fit’ Model (Grey/Black) and Observation (Red) Time-series for methane (ppb) Jan-Mar 2008

© Crown copyright Met Office Nitrous Oxide

© Crown copyright Met Office Methane

© Crown copyright Met Office Conclusions Estimate Northern Hemisphere baseline trends from observation station on west coast of Ireland. N 2 O shows a steady rise, CH 4 rising but more irregular. NAME-Inversion method used to estimate UK and NW European emissions of N 2 O and CH 4 from 1990 until N 2 O estimates broadly agree (lower by 10-20% but trend agrees) with UNFCCC inventory. CH 4 estimates for UK show slow decline (10-20%) markedly different to UNFCCC trend which shows a 50% reduction over the 20 years but last 5 years agree well. CH 4 estimate for NW Europe broadly agrees with UNFCCC but with a slightly slower decline. Both NAME-Inversion and UNFCCC estimates have significant uncertainties.

© Crown copyright Met Office Future Increasing the number of representative observation sites will increase resolution of the inversion emission estimates Improved meteorological models will improve the ability of models to calculate the recent history and dilution of the air Satellites – Need to be able to ‘see’ the boundary layer better before they are used more widely