Modeling innovations at the London VAAC

Transcription

1 Modeling innovations at the London VAAC Matthew Hort, Claire Witham, Frances Beckett, Ben Devenish, Susan Leadbetter, Rachel Pelley, David Thomson, Helen Webster

2 Talk Outline Forecast process Inputs, processes and outputs Model developments NWP Impact of ash properties Plume rise model SO 2 modelling Inversion Improved deep convection scheme Uncertainty Summary

3 Forecast Process Interpretation kg Source Terms Modelling Interpretation Interpretation Observations Forecast

4 Meteorological Impact a.k.a. why getting the weather correct is important Gri msvo tn 2011 Ba rðarbunga 2014 (Schmidt et al., JGR 2015) UM NWP Boundary layer depth Plume should be here. Transport distance = 1800 km ECMWF NWP

5 Meteorology Evaluation of Skill RMSE of wind 300 hpa for 2015 Wind speed (m/s) for Z and 12Z sondes UM Global model configuration has been shown to be consistently more accurate than UM LAM output at forecasting upper air winds over the area of responsibility covered by the London VAAC. (Beckett et al 2015 MetO Futurevolc report)

6 SO 2 Modelling Why: Considered a UK National Risk (Laki 1783) Aviation interest Another source of verification/testing Ba rðarbunga UK government interest Great modelling/observation opportunity UK Ireland

7 OMI vs. NAME Column density of SO2 (Schmidt et al., JGR 2015)

8 Inversion Making model = Observations (sort of) Model predicts observations as function of e: o p (e) Observational Data (satellite column loads) o p (e) o a e e ap A priori view of emissions e ap (z,t) (from plume rise height) Best fit estimate of emissions e(z,t)

9 Inversion Model predictions of observations: o p (e) Emissions grid Particles are released from the volcano at a rate of 1g/s within each height block. Results M ij give column load prediction for observation i resulting from emission from source j Prediction for arbitrary source vector e is then Me

10 Grimsvotn Results Images for /05/ /05/11 Ash only inversion Ash+clear sky inversion Satellite ash and clear sky

11 HANDLING SOURCES OF UNCERTAINTY

12 Species Amount Inputs, Processes and Outputs a.k.a. sources of uncertainty Source Dilution Dispersion Particle Size Deposition Losses at source Reactions Transport Fluctuations Buoyancy Duration Coagulation Altitude Reactions Decay Meteorology Pathway Total Exposure Receptor Environment Time to act TIME Exposure Duration Location Human

13 Changing size distribution Uncertainty Explicit Representation Coarse Probabilistic forecast Default Fine 1.7% 5% 15% Changing mass fraction

14 Model Visualisation Supporting Interpretation Composite Plume age plots Max height plots Concentration Chart Lidar look-alike plots Modelled total column Simulated ash

15 Expert Exchange/Elicitation Close links across science and operations Cross team training Strong national and international links Focused science input e.g. Grimsvötn Advisory team Make up: MetO science and academia Focus: supporting operational changes to input parameters Output: Agreed, accepted common position

16 Summary Ongoing research vital for Understanding, constraining and reducing uncertainty Advancing capability Collaboration is key Subjects are wide ranging and need multi-disciplinary collaboration/understanding Delivery agencies/organisations must also maintain ongoing/growing collaboration and links Validation volcanic ash and wider Operational use presents additional challenges for science