Toward a 35-Year North American Precipitation and Ground Surface - PowerPoint PPT Presentation

Toward a 35-Year North American Precipitation and Ground Surface Reanalysis International Symposium on Regional Reanalysis, Bonn, Germany Nicolas Gasset 1 , Vincent Fortin 2 , Étienne Gaborit 1 , Marco Carrera 3 , Milena Dimitrijevic 3 , Guy Roy 1 and Normand Gagnon 1 1 Meteorological Service of Canada, ECCC 2 Institut de recherche d'Hydro-Québec, HQ 3 Meteorological Research Division, ECCC July 17 th 2018

Motivations • Needs for long term precipitation and ground surface analysis datasets at higher resolution: – Better understand processes, regional climatic studies, optimum resource management, model improvement, ... • However – Dedicated products: not ‘yet’ available for North America – Current long term reanalysis: too coarse space/time resolution, limited available variables, … – CMC analysis: too short period of time, non-uniform in time/ space (changes in the operational models) • But, models and data available to produce such a precipitation and surface analysis at an accessible cost Page 2 – 27 August 2018

Methodology GEM GDRS (50 km) GEM RDRS (15 km) ERA-Interim: atmospheric model atmospheric model 1979 to today (~80km/6h) Precipitation and surface data assimilation: CaPA and CaLDAS GEM-Surf (50 km) surface model Surface observations Initialisation + forcing Initialisation Page 3 – 27 August 2018 Assimilation

GEM Reforecast Configuration • Global: GDRS – GEPS reforecast control member config. RDRS GEM domain – GEM 4.8-LTS, GU grid, 50-km – 12-h cycle/24-h reforecast – Possible to couple with CaLDAS • Regional: RDRS – Modified REPS control member config. (cover Arctic Ocean) – GEM 4.8-LTS, LAM, 15-km – 12-h cycle/24-h reforecast – Coupled with CaLDAS/CaPA Page 4 – 27 August 2018

CaLDAS/CaPA Configuration • Online precipitation and surface analysis - CaLDAS/CaPA North-America CaLDAS domain – North-America, 15-km, 3-h (part of Europe possible) – CaLDAS-ISBA-Screen/OI – CaPA 6-h pcp analysis – 06-18h GEM forcings • A posteriori 24-h pcp analysis • Observations: surface obs. – ECCC Operational/Climat archive – ISD (NOAA DS463.3) – TDL (NOAA DS472.0) – Mesonet Québec, OPG Ontario – AHCCD AdjDlyRS Page 5 – 27 August 2018

Impact of surface assimilation on GEM Summer Results, 24h, North America • Simple reforecast and GEM GEM+CaLDAS/CaPA dynamic downscaling (no surface analysis) T – Develop important errors • Surface analysis coupling – Improve surface and atmospheric results high above the ground (600 mb) T - T d Final results inline with operational model (but much less Operational model (10-km) expensive to obtain) Page 6 – 27 August 2018 Reforecast (15-km) without (left) and with (right) CaLDAS coupling

RDRS vs Operational: T/T d (2 m, 06-18h) North America, SYNOP Dew point Temperature Page 7 – 27 August 2018

RDRS vs Operational: T/T d (2 m, 06-18h) DJF MAM JJA SON Avg. RMSE change (%) 10.90 12.37 8.28 8.97 T d North America plus 8.61 5.39 3.80 8.91 T 11.58 14.68 22.21 14.30 T d Alaska plus 11.32 12.72 5.14 13.83 Canadian Arctic S T Y 8.89 11.85 9.57 7.30 T d N Canada 2.81 2.12 1.59 5.12 T O 1.81 6.55 8.52 2.83 T d P Canada East -2.42 -5.74 -1.64 1.75 T 13.71 14.45 6.29 7.89 T d Canada West 5.55 6.50 2.88 5.35 T 9.99 4.23 0.83 5.59 T d United States of 13.20 6.81 7.11 10.23 America M T E 6.63 6.17 3.43 4.33 T d United States of T 8.04 3.83 4.86 4.81 America East T A 15.05 1.13 -2.30 7.59 T d R United States of 20.57 11.70 9.96 16.95 America West T Page 8 – 27 August 2018

RDRS (06-18h) vs. ERA5 vs ERA-Int. Dew point Canada, SYNOP Temperature Bias Bias 2011 2012 2013 2014 2011 2012 2013 2014 STD STD Page 9 – 27 August 2018 2011 2012 2013 2014 2011 2012 2013 2014

CaLDAS vs. ERA5 (T 2m Summer) Page 10 – 27 August 2018

CaLDAS vs. ERA5 (space+time) Temperature Dew point Spring BIAS STD RMSD Spring BIAS STD RMSD Alaska_plus_Canadian_Arctic 0.15 2.68 2.69 Alaska_plus_Canadian_Arctic -0.32 2.82 2.84 Canada_East -0.07 1.85 1.86 Canada_East -0.55 1.99 2.06 Canada_West -0.24 1.91 1.92 Canada_West -0.42 2.08 2.12 United_States_of_America_East -0.03 1.73 1.73 United_States_of_America_East -0.28 1.42 1.45 United_States_of_America_West -0.05 3.17 3.17 United_States_of_America_West -0.73 2.16 2.27 Summer BIAS STD RMSD Summer BIAS STD RMSD Alaska_plus_Canadian_Arctic -0.12 1.96 1.96 Alaska_plus_Canadian_Arctic 0.05 2.05 2.05 Canada_East -0.02 1.60 1.60 Canada_East 0.06 1.63 1.63 Canada_West -0.35 1.81 1.84 Canada_West -0.23 1.96 1.97 United_States_of_America_East 0.10 1.63 1.64 United_States_of_America_East -0.40 1.43 1.49 United_States_of_America_West 0.04 3.09 3.09 United_States_of_America_West -0.85 2.20 2.36 Autumn BIAS STD RMSD Autumn BIAS STD RMSD Alaska_plus_Canadian_Arctic -0.28 2.52 2.53 Alaska_plus_Canadian_Arctic -0.58 2.77 2.83 Canada_East -0.18 1.46 1.48 Canada_East -0.37 1.50 1.55 Canada_West -0.43 1.82 1.87 Canada_West -0.46 1.98 2.04 United_States_of_America_East 0.14 1.51 1.51 United_States_of_America_East -0.37 1.33 1.38 United_States_of_America_West -0.12 2.72 2.72 United_States_of_America_West -0.84 2.25 2.40 Winter BIAS STD RMSD Winter BIAS STD RMSD Alaska_plus_Canadian_Arctic -0.09 3.33 3.33 Alaska_plus_Canadian_Arctic -0.55 3.78 3.82 Canada_East -0.10 1.98 1.98 Canada_East -0.15 2.06 2.06 Page 11 – 27 August 2018 Canada_West -0.29 2.11 2.13 Canada_West -0.14 2.26 2.26 United_States_of_America_East 0.07 1.53 1.53 United_States_of_America_East -0.21 1.43 1.45 United_States_of_America_West -0.15 2.76 2.77 United_States_of_America_West -0.67 2.54 2.62

CaPA with AdjDlyRS (2010-2014) Compared with Canadian manual SYNOP Winter Summer Page 12 – 27 August 2018 Without AdjDlyRS With AdjDlyRS

CaPA24h vs Stage IV/GPCP (2010-2015) RDRS-CaPA24h MPE STAGE IV GPCP Page 13 – 27 August 2018

CaPA24h vs Stage IV/GPCP (2010-2015) RDRS-CaPA24h CaPA retrospective analysis available publicly : (North America, 24-h, 15-km, 5-years) http://collaboration.cmc.ec.gc.ca/science/outgoing/capa.grib/ MPE STAGE IV hindcast/capa_hindcast_rdrs_v1/ GPCP Page 14 – 27 August 2018

CaLDAS: Soil water content evaluation CaLDAS vs SMOS Satellite retrievals Bias: 0.029 m 3 /m 3 RMSD: 0.104 m 3 /m 3 In the literature (vs. surface obs.): Surface observations • Bias: Obs. nsta nrec Bias RMSD network – SMOS < abs(0.025) m 3 /m 3 – MERRA-Land ~ -0.06 m 3 /m 3 SCAN 450 3544212 -0.0054 0.1087 – ERA-Land ~ -0.08 m 3 /m 3 SNOTEL 499 2812903 -0.0128 0.1239 • RMSD: PBO_H2O 137 128791 0.0048 0.0837 – SMOS ~ 0.05-0.1 m 3 /m 3 Page 15 – 27 August 2018 – MERRA-Land ~ 0.1 m 3 /m 3 USCRN 228 2262235 -0.0018 0.1018 – ERA-Land ~ 0.13 m 3 /m 3

Hydrology applications: streamflows Page 16 – 27 August 2018

Production and computational cost Steps to come (FY 2018-2019) • Publication to peer-reviewed journal • Proposition to hydrometeorological prediction forum • Tech. transfer from R&D to operations • Production Computational cost for 35 years = 7 x (5 years + 1 year init.) ~ 110 days on 7 x (128+128+20) = 1900 cores ( 7 track in parallel ) ~ 630 days on (128+128+20) = 280 cores (sequential) ~ 200 000 days on 1 core Initialisation Reanalysis Page 17 – 27 August 2018 2015 1979 1984 1989 1994 1999 2004 2009

Public access to Reanalysis (future) (and all CMC surface related products) http://www.caspar-data.ca Juliane Mai, Kurt C. Kornelsen , Bryan Tolson, Paulin Coulibaly, François Anctil, Vincent Fortin, Michael Leahy, Brent Hall Page 18 – 27 August 2018

Thanks for your attention Questions and/or comments? This project is funded by the International Joint Commission (IJC) and Environment and Climate Change Canada (ECCC) Page 19 – 27 August 2018 Contact: nicolas.gasset@canada.ca