in CMIP5 GCM projections for the British Isles winters 20 09 2 099 Jan Stryhal amp Radan Huth Faculty of Science Charles University Prague Institute of Atmospheric Physics Academy of Sciences of the Czech Republic Prague ID: 1022526
Download Presentation The PPT/PDF document "Analysis of automated circulation classi..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
1. Analysis of automated circulation classifications in CMIP5 GCM projections for the British Isles-(winters 2009–2099)Jan Stryhal & Radan HuthFaculty of Science, Charles University, PragueInstitute of Atmospheric Physics, Academy of Sciences of the Czech Republic, Prague EMS Dublin 2017
2. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesCONTENTSINTRODUCTIONproject/concepts: winter Euro-Atlantic circulation: reanalyses – historical runs – projectionsautomated circulation classifications DATA & METHODSRESULTSreference datasetvalidation of GCMsprojections of CT frequency, persistence & airflow strengthSUMMARY & CONCLUSIONS
3. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesINTRODUCTIONLarge-scale atmospheric circulation has received considerable attention owing to its close link to meteorological & environmental elements at regional to local scales,especially in winter
4. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesINTRODUCTIONLarge-scale atmospheric circulation has received considerable attention owing to its close link to meteorological & environmental elements at regional to local scales,especially in winterCahynová & Huth (2016), IJC 36:2743–2760
5. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesINTRODUCTIONLarge-scale atmospheric circulation has received considerable attention owing to its close link to meteorological & environmental elements at regional to local scales,especially in winterThe ongoing development of climate models raises the question whether the models are able to simulate large-scale circulation & its links to local climate, and how circulation could change
6. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesINTRODUCTIONLarge-scale atmospheric circulation has received considerable attention owing to its close link to meteorological & environmental elements at regional to local scales,especially in winterThe ongoing development of climate models raises the question whether the models are able to simulate large-scale circulation & its links to local climate, and how circulation could change Several approaches have been used to assess the circulation in models, including analyses of mean fields of SLP or GPHs, blocking, cyclone tracks and intensity, position of jet streams, teleconnections, …Zappa et al. (2013), J Climate 26:5379–5396Wójcik (2015), IJC 35:714–732
7. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesINTRODUCTION…one of the most wide spread approaches having been classifications of atmospheric circulation patterns (circulation classifications)
8. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesINTRODUCTION…one of the most wide spread approaches having been classifications of atmospheric circulation patterns (circulation classifications)a tool that…1. describes the entire variety of the atmospheric circulation by defining a catalogue of a few circulation typesDEC 1 1960 DEC 2 1960 DEC 3 1960 DEC 4 1960 DEC 5 1960 DEC 6 1960 …….wetter3.de(20CR)
9. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesINTRODUCTION…one of the most wide spread approaches having been classifications of atmospheric circulation patterns (circulation classifications)a tool that…1. describes the entire variety of the atmospheric circulation by defining a catalogue of a few circulation types and2. subsequently classifies each circulation pattern to one of these CTsDEC 1 1960 DEC 2 1960 DEC 3 1960 DEC 4 1960 DEC 5 1960 DEC 6 1960 …….wetter3.de(20CR)
10. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesINTRODUCTION simplification requires many subjective choices (number of CTs, data, method, …) which affect the results
11. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesINTRODUCTION simplification requires many subjective choices (number of CTs, data, method, …) which affect the results → relying on a single classification means that one gets only an incomplete picture of reality and, consequently, is at risk of misinterpreting the resultstemperature, precipitation, trends in meteorological variables/circulation, droughts, air quality, surface ozone, wild fires, landslides, phenological phases
12. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesINTRODUCTION simplification requires many subjective choices (number of CTs, data, method, …) which affect the results → relying on a single classification means that one gets only an incomplete picture of reality and, consequently, is at risk of misinterpreting the resultstemperature, precipitation, trends in meteorological variables/circulation, droughts, air quality, surface ozone, wild fires, landslides, phenological phasesanalyses of CTs in models have typically relied on one single classification (Demuzere et al. 2009; Perez et al. 2014; Otero et al. 2017), two at the best (Pastor and Casado 2012; Rohrer et al. 2017)
13. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesINTRODUCTION simplification requires many subjective choices (number of CTs, data, method, …) which affect the results → relying on a single classification means that one gets only an incomplete picture of reality and, consequently, is at risk of misinterpreting the resultstemperature, precipitation, trends in meteorological variables/circulation, droughts, air quality, surface ozone, wild fires, landslides, phenological phasesanalyses of CTs in models have typically relied on one single classification (Demuzere et al. 2009; Perez et al. 2014; Otero et al. 2017), two at the best (Pastor and Casado 2012; Rohrer et al. 2017)GOAL: use multiple classification (methods) and produce robust estimates of biases/changes + quantify the effect of the choice of the method on results
14. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesDATA & METHODS 5 reanalyses × 40 winters × 90 days = 18,000 patternsNameInstituteReferenceERA-40European Centre for Medium-Range Weather ForecastsUppala et al. 2005NCEP-1National Centers for Environmental Prediction, National Center for Atmospheric ResearchKalnay et al. 1996JRA-55Japan Meteorological AgencyKobayashi et al. 201520CRv2NOAA Earth System Research Laboratory, University of Colorado CIRES Climate Diagnostics CenterCompo et al. 2011ERA-20CEuropean Centre for Medium-Range Weather ForecastsPoli et al. 2016Intercomparison of reanalyses:Stryhal & Huth (2017)J. Climate 30:7847–7861
15. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesDATA & METHODS 5 reanalyses × 40 winters × 90 days = 18,000 patternsNameInstituteReferenceERA-40European Centre for Medium-Range Weather ForecastsUppala et al. 2005NCEP-1National Centers for Environmental Prediction, National Center for Atmospheric ResearchKalnay et al. 1996JRA-55Japan Meteorological AgencyKobayashi et al. 201520CRv2NOAA Earth System Research Laboratory, University of Colorado CIRES Climate Diagnostics CenterCompo et al. 2011ERA-20CEuropean Centre for Medium-Range Weather ForecastsPoli et al. 2016Method abbreviationMethod nameNo of CTsGWTGrosswettertypes10JCT1Jenkinson–Collison10JCT2Jenkinson–Collison10LNDLund9PCTT-mode PCA obliquely rotated9CKMk-means (differing start partitions)9SANsimulated annealing (SANDRA)9KMDk-medoids9
16. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesDATA & METHODS 5 reanalyses × 40 winters × 90 days = 18,000 patternsNameInstituteReferenceERA-40European Centre for Medium-Range Weather ForecastsUppala et al. 2005NCEP-1National Centers for Environmental Prediction, National Center for Atmospheric ResearchKalnay et al. 1996JRA-55Japan Meteorological AgencyKobayashi et al. 201520CRv2NOAA Earth System Research Laboratory, University of Colorado CIRES Climate Diagnostics CenterCompo et al. 2011ERA-20CEuropean Centre for Medium-Range Weather ForecastsPoli et al. 2016Method abbreviationMethod nameNo of CTsGWTGrosswettertypes10JCT1Jenkinson–Collison10JCT2Jenkinson–Collison10LNDLund9PCTT-mode PCA obliquely rotated9CKMk-means (differing start partitions)9SANsimulated annealing (SANDRA)9KMDk-medoids9
17. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesDATA & METHODS 5 reanalyses × 40 winters × 90 days = 18,000 patternsNameInstituteReferenceERA-40European Centre for Medium-Range Weather ForecastsUppala et al. 2005NCEP-1National Centers for Environmental Prediction, National Center for Atmospheric ResearchKalnay et al. 1996JRA-55Japan Meteorological AgencyKobayashi et al. 201520CRv2NOAA Earth System Research Laboratory, University of Colorado CIRES Climate Diagnostics CenterCompo et al. 2011ERA-20CEuropean Centre for Medium-Range Weather ForecastsPoli et al. 2016Method abbreviationMethod nameNo of CTsGWTGrosswettertypes10JCT1Jenkinson–Collison10JCT2Jenkinson–Collison10LNDLund9PCTT-mode PCA obliquely rotated9CKMk-means (differing start partitions)9SANsimulated annealing (SANDRA)9KMDk-medoids9
18. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesDATA & METHODS 5 reanalyses × 40 winters × 90 days = 18,000 patternsNameInstituteReferenceERA-40European Centre for Medium-Range Weather ForecastsUppala et al. 2005NCEP-1National Centers for Environmental Prediction, National Center for Atmospheric ResearchKalnay et al. 1996JRA-55Japan Meteorological AgencyKobayashi et al. 201520CRv2NOAA Earth System Research Laboratory, University of Colorado CIRES Climate Diagnostics CenterCompo et al. 2011ERA-20CEuropean Centre for Medium-Range Weather ForecastsPoli et al. 2016Method abbreviationMethod nameNo of CTsGWTGrosswettertypes10JCT1Jenkinson–Collison10JCT2Jenkinson–Collison10LNDLund9PCTT-mode PCA obliquely rotated9CKMk-means (differing start partitions)9SANsimulated annealing (SANDRA)9KMDk-medoids9
19. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesDATA & METHODS 5 reanalyses × 40 winters × 90 days = 18,000 patternsNameInstituteReferenceERA-40European Centre for Medium-Range Weather ForecastsUppala et al. 2005NCEP-1National Centers for Environmental Prediction, National Center for Atmospheric ResearchKalnay et al. 1996JRA-55Japan Meteorological AgencyKobayashi et al. 201520CRv2NOAA Earth System Research Laboratory, University of Colorado CIRES Climate Diagnostics CenterCompo et al. 2011ERA-20CEuropean Centre for Medium-Range Weather ForecastsPoli et al. 2016Method abbreviationMethod nameNo of CTsGWTGrosswettertypes10JCT1Jenkinson–Collison10JCT2Jenkinson–Collison10LNDLund9PCTT-mode PCA obliquely rotated9CKMk-means (differing start partitions)9SANsimulated annealing (SANDRA)9KMDk-medoids9
20. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesDATA & METHODSASSIGNMENT of simulated patterns for each model and classificationVALIDATION: winter 1960–2000PROJECTION: winter 2009 – 2099 (consecutive 30-year windows)Method abbreviationMethod nameNo of CTsGWTGrosswettertypes10JCT1Jenkinson–Collison10JCT2Jenkinson–Collison10LNDLund9PCTT-mode PCA obliquely rotated9CKMk-means (differing start partitions)9SANsimulated annealing (SANDRA)9KMDk-medoids9NoModel name1CanESM22CCSM43CMCC-CESM4CMCC-CM5CMCC-CMS6CNRM-CM57EC-EARTH8GFDL-CM39GFDL-ESM2G10GFDL-ESM2M11HadGEM2-AO12HadGEM2-CC13HadGEM2-ES14INM-CM415IPSL-CM5A-LR16IPSL-CM5A-MR17IPSL-CM5B-LR18MIROC519MIROC-ESM20MIROC-ESM-CHEM21MPI-ESM-LR22MPI-ESM-MR23MRI-CGCM324MRI-ESM125NorESM1-M
21. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesDATA & METHODSASSIGNMENT of simulated patterns for each model and classificationVALIDATION: winter 1960–2000PROJECTION: winter 2009 – 2099 (consecutive 30-year windows)Method abbreviationMethod nameNo of CTsGWTGrosswettertypes10JCT1Jenkinson–Collison10JCT2Jenkinson–Collison10LNDLund9PCTT-mode PCA obliquely rotated9CKMk-means (differing start partitions)9SANsimulated annealing (SANDRA)9KMDk-medoids9NoModel name1CanESM22CCSM43CMCC-CESM4CMCC-CM5CMCC-CMS6CNRM-CM57EC-EARTH8GFDL-CM39GFDL-ESM2G10GFDL-ESM2M11HadGEM2-AO12HadGEM2-CC13HadGEM2-ES14INM-CM415IPSL-CM5A-LR16IPSL-CM5A-MR17IPSL-CM5B-LR18MIROC519MIROC-ESM20MIROC-ESM-CHEM21MPI-ESM-LR22MPI-ESM-MR23MRI-CGCM324MRI-ESM125NorESM1-MSeparately for reanalyses and models:relative frequency + persistence (each CT)Separately for the two ensebles:airflow strength + direction and shear vorticity (each CT mean map)
22. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesDATA & METHODS
23. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesRESULTS1. Frequency of occurrence
24. Analysis of automated circulation classifications in CMIP5 GCM projections for the British Isles1. Frequency of occurrenceRESULTS
25. Analysis of automated circulation classifications in CMIP5 GCM projections for the British Isles1. Frequency of occurrenceRESULTS
26. Analysis of automated circulation classifications in CMIP5 GCM projections for the British Isles1. Frequency of occurrenceRESULTS
27. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesRESULTS
28. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesRESULTS2. Persistence
29. Analysis of automated circulation classifications in CMIP5 GCM projections for the British Isles2. PersistenceRESULTS
30. Analysis of automated circulation classifications in CMIP5 GCM projections for the British Isles2. PersistenceRESULTS
31. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesRESULTS3. Flow strength
32. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesRESULTS3. Flow strength
33. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesRESULTS3. Flow strength
34. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesThe most apparent trends in CTs over the 21st century:increase in the frequency of CTs with advection from W, towards in the of the century from the whole western quadrant (10–20% relative to control)the remaining directional CTs + AC CTs have a similar trend of the opposite signproblem – simulation of AC CTs (-35% relative to reanalyses) – projections?persistence is slightly underestimated and is projected to further slightly decreasestrength of airflow is simulated well and is not projected to changevorticity is overestimated (more cyclonic) in mean maps of most CTs and is projected to increasethere is no statistically significant difference in trends projected by models that perform well and those performing poorly over the control periodrobust estimates of biases/trends require a comparison of multiple classifications, interpretations of single classifications often lead to contradictory resultsSUMMARY
35. Analysis of automated circulation classifications in CMIP5 GCM projections for the British IslesThank you for your attention!ACKNOWLEDGEMENTSThe work was funded by the Grant Agency of Charles University, project 188214. We acknowledge the following organizations for providing their reanalysis datasets: NOAA/OAR/ESRL PSD, Boulder, Colorado, USA for NCEP/NCAR, and 20th Century Reanalysis V2, ECMWF for ERA-40, and ERA-20C, and JMA for JRA-55. We thank all climate modelling groups for their GCM simulations, and the PCMDI for enabling access to the data. Thanks are also due to all developers of the COST733 software.