1 / 49

Centre report: Recent changes to and plans for the NWP suites of Environment Canada

Centre report: Recent changes to and plans for the NWP suites of Environment Canada. WGNE-29 – Melbourne, Australia Ayrton Zadra RPN – Environment Canada 10-13 March 2014. Acknowledgements.

blaze
Télécharger la présentation

Centre report: Recent changes to and plans for the NWP suites of Environment Canada

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Centre report:Recent changes to and plans for the NWP suites of Environment Canada WGNE-29 – Melbourne, Australia Ayrton Zadra RPN – Environment Canada 10-13 March 2014

  2. Acknowledgements Weather Prediction: Martin Charron, Ron Mctaggart-Cowan, Jason Milbrandt, Abdessamad Qaddouri, Claude Girard Environmental Prediction: Greg Smith, Pierre Pellerin, Vincent Fortin, Stephane Belair Data Assimilation: Mark Buehner, Jean-Francois Caron, Luc Fillion, Stephane Laroche, Peter Houtekamer CMC-Development: Normand Gagnon

  3. -- Part 1 --Recent changes to operational suites

  4. Summary of recent changes Major upgrade to Global Prediction Systems (Deterministic & Ensemble)… Major upgrade to Ensemble Prediction Systems (Global & Regional) Adjustments to High Resolution Deterministic Prediction System Update to Regional Air Quality Deterministic Prediction System Additional satellite (CSR, ATOVS, polar winds) added to deterministic systems New Operational Hydro-dynamic Simulation System Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Fev Mar 2013 2014 2012 2013 METOP-1 added to GPS-RO … accompanied by upgrades in Regional Deterministic Prediction System Adjustments to ocean analysis in seasonal prediction system (CanSIPS) Experimental Pan-Canadian High Resolution Deterministic Prediction System Satwinds (METEOSAT10) + ASCAT winds (METOP-1) added to DA system Experimental Global Ice-Ocean Prediction System (GIOPS) New Regional Deterministic Air Quality Analysis Upgrades to Nowcasting system (INCS) Experimental Regional Ice Prediction System (RIPS)

  5. Major upgrade of the Global Deterministic Prediction System (GDPS):summary of changes forecast model DA (4Dvar)

  6. GDPS upgrade parallel suite: Oct-2012 to Jan-2013,geopotential height RMSE at day-5 (verification against analyses) N.Hemisp. OLD – NEW @ 500 hPa OLD NEW @ day 5 pressure level (hPa) RMSE difference (dam) Day RMSE (m) S.Hemisp. pressure level (hPa) RMSE difference (dam) Day RMSE (m)

  7. Major upgrade of the Global Deterministic Prediction System (GDPS):impact of main upgrades since 2001 annual running mean of day-5 GZ-500hPa RSME against radiosondes over N. Hemisphere latest upgrade

  8. Two upgrades of the Global Ensemble Prediction System (GEPS) (1) Feb-2013 upgrade • multi-scalealgorithm • time-step: from30 to 20min • horiz. resol.: from100 to 66km • vertical levels: from58 to 74 • topographyfilter • reducedthinning of observations (2.7 X radiances) • improveddynamics and physics OLD NEW CRPS(OLD) - CRPS(NEW) Fig.: Global verification (CRPS* error) of temperature at 500hPa against radiosondes, of OLD versus NEW GEPS, showing a gain in predictability of 12h and plus. [*CRPS = Continuous Rank Probability Score]

  9. Two upgrades of the Global Ensemble Prediction System (GEPS) (2) Oct-2013 upgrade • evolving SST (based on anomalypersistencemethod) • extension to monthlyforecasts (32 days) once a week • operationalhistoricalforecasts (72 hindcasts per week, over the 1995-2012 period) OLD NEW CRPS(OLD) - CRPS(NEW) Fig.: Verification (CRPS error) of 2-m temperature against SYNOP data over N. America, of OLD versus NEW GEPS, showing improvements due to the use of an evolving SST.

  10. Major upgrade Regional of theRegional Ensemble Prediction System (REPS) Changes model component only: - horiz. resolution: from 33 to 15km - vertical levels from 28 to 40 - improved treatment of stochastic physical tendency perturbations to avoid unrealistic precipitation rates - improved boundary layer parameterization Verification: Significant improvements of the scores for all upper-air variables at all levels, as well as screen-level temperature and dew-point depression. OLD NEW

  11. -- Part 2 --Ongoing and future projects

  12. Upcoming Global Deterministic Prediction System (GDPS 4.0): assimilation related elements • EnVarreplaces 4D-Var • Horizontal grids: • Analysis increment: 50km instead of 100km • Satellite radiance observations: • Additional AIRS/IASI channels assimilated • UpgradeRTTOV8 to RTTOV10 • Modifiedobs error stddevfor all radianceobservations • Improved satellite radiance bias correction scheme • Improved treatment of radiosonde (4D) and aircraft observations • Assimilation of ground-based GPS data • Use of new global sea ice concentration analysis (based on 3D-Var) • 4D Incremental Analysis Update (IAU) replaces digital filter • Use of sequencer Maestro for R/D/O • * NOTE: Most elements also apply to new regional system (RDPS)

  13. Ensemble-Variational assimilation: EnVar • EnVar uses a variational assimilation approach in combination with the already available 4D ensemble covariances from the EnKF • By making use of the 4D ensembles, EnVar performs a 4D analysis without the need of the tangent-linear and adjoint of forecast model • Consequently, it is more computationally efficient and easier to maintain/adapt than 4D-Var • Hybrid covariances used in EnVar by averaging the ensemble covariances with the static NMC-method covariances • Future improvements to EnKF should benefit both GEPS and GDPS  incentive to increase overall effort on EnKF development

  14. EnVar: a replacement of 4D-Var • Overall, EnVar (~10 min) analysis ~6X faster than 4D-Var (>1 hr) on half as many cpus, even though much higher resolution increments • Nearly identical configuration of EnVar used for both global and regional systems (unified deterministic analysis) • Large portions of fortran code already being shared between EnVar and EnKF, unification effort continuing • Results from both global and regional EnVar are mostly comparable or better than 4D-Var, especially at shorter lead times • Decision made to replace 4D-Var with more efficient EnVar in GDPS 4.0, if EnVar is at least as good as current 4D-Var

  15. 2013-2017: Toward a Reorganization of the NWP Suites at Environment Canada Current systems Perturbed members of the regional ensemble prediction system (REPS) Perturbed members of the global ensemble prediction system (GEPS) Global EnKF Global deterministic prediction system (GDPS) Regional deterministic prediction system (RDPS) Global 4D-Var Regional 4D-Var regional system global system

  16. 2013-2017: Toward a Reorganization of the NWP Suites at Environment Canada Increasing role of global ensembles… GDPS4.0 Regional Ensemble forecasts (REPS) Global ensemble forecasts (GEPS) Global EnKF Background error covariances Global deterministic forecast (GDPS) Regional Deterministic forecast (RDPS) Global EnVar Regional EnVar regional system global system

  17. 2013-2017: Toward a Reorganization of the NWP Suites at Environment Canada Global and regional ensembles… Regional ensemble forecasts (REPS) Regional EnKF Global ensemble forecasts (GEPS) Background error covariances Global EnKF Regional deterministic forecast (RDPS) Regional EnVar Background error covariances High-resolution deterministic prediction system (HRDPS) Global deterministic forecast (GDPS) High-res EnVar Global EnVar regional system global system

  18. Dependencies between global systems • Current system (1-way dependence): • GEPS relies on GDPS to perform quality control (background check) for all observations and bias correction for satellite radiance observations xa xb xb xb, obs GDPS: Bgcheck+BC 4D-Var GEM (9h fcst) obs xb xa xb GEPS: EnKF GEM (9h fcst)

  19. Dependencies between global systems • Current system (1-way dependence): • With EnVar (2-way dependence): • 2-way dependence (EnVar uses EnKF ensemble of background states) increases complexity of overall system  2 systems have to be run simultaneously xa xb xb xb, obs GDPS: Bgcheck+BC 4D-Var GEM (9h fcst) obs xb xa xb GEPS: EnKF GEM (9h fcst) xa xb xb xb, obs GDPS: Bgcheck+BC EnVar GEM (9h fcst) xb obs xb xb xa GEPS: EnKF GEM (9h fcst)

  20. Upcoming Regional Deterministic Prediction System (RDPS) RDPS v.3.1.0: Intermittent cycling using 4D/3D-Var G2 (current operational version) Global 25km 4DVar Xa=100km D2 D1 Global 33km Global 33 km interpolation 3DVar Xa=100km R2 R1 LAM-Reg 10km LAM-Reg 10 km 4DVar Xa=100km T-6h T+48h T

  21. RDPS v.4.0.0: Intermittent cycling using 4D-EnVar based on global EnKF* Upcoming Regional Deterministic Prediction System (RDPS) G2 (to be operational late 2014) Global 25km EnVar Xa=50 km * EnVar setup in D1 and R1 identical to the GDPS D2 D1 Global 33km Global 33 km • D1 and R1 upgrade also includes (as in the GDPS) • New Bias Correction • Radiosondes drift • Added IR channels • Ground-based GPS interpolation EnVar Xa=50km R2 R1 LAM-Reg 10km LAM-Reg 10 km EnVar Xa=50km T-6h T+48h T

  22. Continuous Cycling Regional EnKF Global xa xa xb xa xa xb ... ... GEM (66km) GEM (66km) • Regional EnKF starts from the global analysis ensemble. • 192 ensemble members (same as the global). • Lateral boundary conditions from the global EnKF. • Model top around 14 hPa. • No model parameter perturbations. • Prepare 21 initial conditions for REPS at 00 and 12 UTC. EnKF EnKF EnKF Driver Driver xb xa xb xa ... GEM-LAM (15km) GEM-LAM (15km) Regional EnKF EnKF

  23. High Resolution Deterministic Prediction System (HRDPS) HRDPS (multi-grid) HRDPS (pan-Canadian) RDPS (10 km)

  24. Main objective for the pan-Canadian HRDPS  To become the primary source of NWP guidance for day 1 and 2 To be accomplished in 2 major steps: 1. Phase 1 (2014) Implementation of an experimental pan-Canadian sub-component • add new domain • surface ICs supplied by coupled 2.5-km CaLDAS • hydrometeor fields are “recycled” from the previous 2.5-km run • modifications to GEM configuration 2. Phase 2 (2015) • upper-air data assimilation cycle • model/configuration upgrades (physics, vertical resolution, …) • expansion of coverage • removal of (remaining) local domains

  25. DA for a convective-scale model • HRDPS: A pan-Canadian 2.5-km forecasting system • Grid points: 2584 x 1334 • Forecasts up to +48-h • Should eventually replace the RDPS as the main guidance for short-term forecasts in Canada Phase 1 (2014) : No atmospheric DA; Downscaling of the 10-km RDPS analysis; hydrometeors are ‘recycled’ from the previous 2.5-km run (i.e. every 6-h) Phase 2 (2015) : Continuous cycling using 4D-EnVar (+IAU) based on a 10-km limited-area EnKF

  26. ICs and BCs: 2.5-km CaLDAS CaLDAS-screen (2.5 km) Valid on 25 June 25 2011, 1200 UTC Near-Surface Soil Moisture (0-10 cm)

  27. Yin-Yang grid for global forecasting A two-way coupling method between two limited-area models Qaddouri & Lee, 2011: The Canadian Global Environmental Multiscale model on the Yin-Yang grid system, QJRMS 137, 1913-1926) Yin Yang • No poles + global quasi-uniform grid => simplification of numerical schemes: • semi-Lagragian scheme without considering fluid parcel trajectory as great circle • explicit numerical diffusion solver • More balanced computational load for scalability purposes when compared to lat-lon grids

  28. Options for Semi-Lagrangian Trajectory Calculations - Averaging rule : Mid-point / Trapezoidal - Interpolation : Linear / Cubic Here we compare mid-point rule and trapezoidalrule for the calculation of displacements Dr in the semi-Lagrangian scheme. The mid-point rule (a time mean followed by a space interpolation) can be described as follows: where i is for iterations being made due to the non-linear nature of the process, while the trapezoidal rule (a space interpolation followed by a space-time mean) can be written: Changing rule is fairly straightforward except for the ‘horizontal’ on the sphere. Information: Girard et al., MWR 2014, Appendix 14, Trapezoidal rule for trajectory calculations

  29. Mid-point rule/linear interp Trapezoial rule/linear interp Mid-point rule/cubic interp Trapezoidal rule/cubic interp Idealized Flow past Topography (Schär’s case): Trajectory calculations using …

  30. Trapezoidal rule Global Averaged Scores 44 Winter Cases 6-Day Forecasts Gem Yin-Yang 15km Resolution Semi-Lagrangian Trajectory Calculations Blue: Mid-point rule/linear interpolation Red: Various modifications Cubic interpolation Trapezoidal rule/cubic interpolation

  31. -- Appendices --

  32. A) 4D-IAU + selective physics recycling δ Incremental Period Trial Period Forecast Period (G1) • Analysis increment (δ) is applied as δ/N , where N is # of timesteps in 6h assimilation window (T-3h to T+3h). • Increments are allowed to evolve following the 4D B matrix available in EnVar. • Some physical quantities (cloud condensate and PBL quantities) from the previous integration (background) are recycled into the next integration. • When IAU and physics recycling are combined, model spin-up is virtually eliminated. • The replacement of DF by IAU also appears to have a strong positive impact on the semidiurnal tide, apparent in tropical scores 00Z Run UTC 21 00 03 06 09 δ “Analysis” 06Z Run 03 06 09 12 15

  33. B) Upgrades and Improvements to the MSC Data Processing for Radiosonde and Aircraft Data • Increased volume of data: selection of observations according to model levels • Revised observation error statistics • Revised rejection criteria for radiosonde data based on those used at ECMWF • Horizontal drift of radiosonde balloon taken into account in both data assimilation and verification systems • Bias correction scheme for aircraft temperature reports wind speed temperature proposed for both radiosonde & aircraft operational Impact of proposed changes • General short-range forecast improvements above 500 hPa in both wind and temperature fields • The temperature forecast biases are significantly improved due to the bias correction scheme for aircraft below 200 hPa and to the new rejection criteria for radiosonde humidity data above • See Laroche & Sarrazin 2013, Weather and Forecasting, 28, pp 772-782 12h 48h Fig.: Verification scores against radiosondes over the N. Hemisphere, Jan-Feb 2009 (dash = bias; solid = stde)

  34. BIAS STDE C) The new Canadian Land Data Assimilation System (CaLDAS)(in 2013) Fig.: Impact of CaLDAS on screen lecel air dew-point temperature forecasts over Canada, over the summer 2008: operational system versus CaLDAS. IN CaLDAS OUT • Land surface initial conditions for NWP and hydro systems • Ancillary land surface data ISBA LAND-SURFACE MODEL Orography, vegetation, soils, water fraction, ... • Land surface conditions for atmospheric assimilation systems • Atmospheric forcing ASSIMILATION xb T, q, U, V, Pr, SW, LW y • Current state of land surface conditions for other applications (agriculture, drought, ...) • Observations (EnKF approach) Screen-level (T, Td) Stations snow depth L-band passive (SMOS,SMAP) MW passive (AMSR-E) Multispectral (MODIS) Combined products (GlobSnow) OBS xa = xb+ K { y – H(xb) } with K = BHT ( HBHT+R)-1

  35. D) Water cycle prediction system based on coupled numerical models • Focus on Great Lakes and St. Lawrence watershed: • Great Lakes: 2-way coupled atmos.-ocean model (GEM+NEMO) • Watershed: 1D model of land-surface + routing (MESH) • St. Lawrence: 2D hydrodynamic model (H2D2) • Includes pollutant transport model and habitat models Impact of lakes on weather needs to be captured correctly: DJF 05-09 daily precip. shown Tributary flow predicted, (with data assimilation of streamflow obs.) @ 500m Connected to water quality and ecosystem models: e.g. predicted wastewater plume for Montreal

  36. E) EnVar Pre-Final Cycles* vs. 4D-Var *Using 66-km Ensemble and 25-km 4DVar-based Global Analysis Radiosonde verification scores – 120 cases, Winter 2011 U |Vh| U |Vh| Z T Z T T+48h North America T+24h North America T-Td T-Td

  37. Satellite data assimilation at EC F) Satellite data assimilation: R&D 2014-2015 • To be assimilated within EnVar late 2014 or 2015 • Upgrade of AIRS & IASI, add Cris (~140 channels each) • Add ATMS (~16 channels) • Inter-channel observation error (IR & MW sounders) • Higher density of radiances (from 150 km to much lower) • GPS-RO extended to surface • Currently the object of research • Assimilation of surface-sensitive channels over land • Higher temporal assimilation based on simulations (OSSE) in view of upcoming hyperspectral IR sounders on GEO • Ozone assimilation from various sensors • Remote sensing of CO2

  38. F) Satellite data assimilation: R&D 2014-2015 Canadian satellite missions with link to operational meteorology • Radarsat constellation (3 satellites, funded, 2018 launch) • - Main applications: sea ice mapping and ocean surface wind • Polar Communications and Weather (PCW, 2 satellites in HEO, under review, Planned for 2021) • - Same applications as MTG-FCI, GOES-R-ABI, but filling high latitude gap • (15 min imagery, multispectral, 100% coverage 60-90oN)

  39. G2) Global Ice-Ocean Prediction System (GIOPS) G1) Regional Ice Prediction System (RIPS) • Mercator Ocean Assimilation System (SAM2-SEEK): • Sea surface temperature • Temperature and salinity profiles • Sea level anomaly from satellite altimeters • 3DVar Ice analysis • Daily blended ice-ocean analysis and 10day forecast • Model configuration: • ORCA025 (~1/4°), <15km in Arctic • NEMOv3.1, LIM2-EVP • Experimental implementation: Jul 2013 • 5km N.American grid • 3DVar Ice analysis • SSMI, AMSR-E, CIS daily charts • CICE4.1 Ice model • Forced by CMC RDPS • 48hr forecasts at 0, 6, 18, 24Z • Experimental implementation: March 2013

  40. H) Future plans for Canadian EPS • In 2014: • Ensemble layer in NinJo • Horizontal resolution of 50 km for the GEPS • Provide trial fields error statistics for EnVAR • NAEFS-LAM (exchange of REPS and SREF data) • In 2015-2016: • Better soil properties via assimilation with CALDAS and stochastic perturbations • New Yin-Yang model grid • Model top at 0.1 hPa (80 km) • Regional EnKF • Increase horizontal resolution for both systems in function of the available computer power. • Stochastic convection

  41. Within the nextfive years... • The ensemble approach will become mainstream • Next-Gen SCRIBE will incorporate the ensemble paradigm • Model resolution will become very attractive to forecasters • Regional EPS at 10 km grid spacing with dedicated data assimilation • Global EPS at ~20-25 km grid spacing • Research on ensemble forecasting will be performed at 1-3 km grid spacing, but no operational kilometer-scale EPS within 3-5 years

  42. Improvements to the GEPS in 2013 • Februaryimplementation : • Better analyses (higherresolution, more observations) • Only one surface scheme (ISBA, Noilhan and Planton, ) • Limitation of the stochasticPhysics Tendency Perturbations when convection occurs • See the technical note of Gagnon et al. 2013 : • http://collaboration.cmc.ec.gc.ca/cmc/cmoi/product_guide/docs/lib/op_systems/doc_opchanges/technote_geps300_20130213_e.pdf • Decemberimplementation: • Evolutive SST (monthlyforecasting on thursdays) • Operationalreforecasting over the last 18 years (with 4 members) • See the technical note of Gagnon et al. 2013: • http://collaboration.cmc.ec.gc.ca/cmc/cmoi/product_guide/docs/lib/technote_geps310_20131204_e.pdf

  43. I) Preliminary results for the new reginal EnKF:6h forecasts verification against radiosondes (20 days) Reduced horizontal localization distance. Variable horizontal localization distance: Near surface: 1600kmNear top: 2800km Same vertical localization as the global. Reduced isotropic model error perturbation. REnKF features UU VV GZ TT ES

  44. J) Canadian AQ Forecasting System • Primary messaging tool is the Air Quality Health Index (AQHI) • Main target is urban areas > 100,000 population • On-line forecast model GEM-MACH provides guidance on AQHI component values (NO2, O3, PM2.5) and meteorological fields out to 48 hours

  45. Canada’s National Air Quality Health Index (AQHI) • Follows example of Canadian national UV index • Year-round, health-based, additive, no-threshold, hourly AQ index • Developed from daily time-series analysis of air pollutant concentrations and mortality data (Stieb et al., 2008) • Weighted sum of NO2, O3, & PM2.5 concentrations • 0 to 10+ range

  46. Elements of Canada’s AQ Forecasting System Schematic diagram of an AQHI forecast Numerical forecast - Next 48 hr - GEM-MACH UMOS-AQ Past and present situation - Last 48 hr - Real-time observations of O3, PM2.5, NO2 Forecasted future situation - Next 48hr - Modelled forecast values of O3, PM2.5, NO2 Forecaster (1 desk/forecast region) AQHI = 10/10.4*100*[(exp(0.000871*NO2)-1) +(exp(0.000537*O3) -1)+(exp(0.000487*PM2.5) -1)] AQHI = 10/10.4*100*[(exp(0.000871*NO2)-1) +(exp(0.000537*O3) -1)+(exp(0.000487*PM2.5) -1)]

  47. GEM-MACH • GEM-MACH is a multi-scale chemical weather forecast model composed of dynamics and physics (GEM) and on-line chemistry modules • Operational configuration of GEM-MACH includes • limited-area-model (LAM) grid configuration for North America • 10-km horizontal grid spacing, 80 vertical levels to 0.1 hPa • 2-bin sectional representation of PM size distribution (i.e., 0-2.5 and 2.5-10 μm) with 9 chemical components • forecast species include O3, NO2, and PM2.5 needed for AQHI

  48. RDPS and Operational GEM-MACH Grids • EC’s limited-area regional deterministic prediction system (RDPS) provides required initial and boundary conditions for GEM-MACH • GEM-MACH’s grid points are co-located with RDPS grid points RDPS grid (blue); GEM-MACH grid (red)

  49. Ongoing developments for GEM-MACH • Operational configuration: • Lengthen forecast from 48 to 72 hours • Include wildfire emissions • Global configuration for assimilation/piloting purposes • 12-bin version for AOD assimilation • Simplified stratospheric chemistry for the assimilation of ozone and GHGs.

More Related