Are the and European Windstorm independent? Assessing teleconnections using a Seasonal Forecast Ensemble Approach Michael Angus1 ([email protected]), Ivan Kuhnel2 and Gregor C. Leckebusch1 1: University of Birmingham, School of Geography, Earth and Environmental Sciences, Birmingham, UK 2: CoreLogic, Catastrophe Risk Management, Paris, France Presentation number: EGU2019-14192

Motivation Event Climatology and verification • European winter windstorms and North Atlantic Hurricanes both cause European Windstorms Atlantic Hurricanes significant loss of life and property, particularly severe events ERAinterim SEAS5 Difference Monthly interannual variability (Leckebusch et al. 2007, Jonkman et al. 2018). Observations ERAinterim Observations - ERAinterim SEAS5 ERAinterim – SEAS5

All Events December • As the interannual variability of both seasons depends on large scale climate drivers (e.g. El Niño), it is reasonable to ask if the two seasons

are related through climate teleconnection patterns. A key goal of this January research is to establish the probability of occurrence of successive Event SSI > extremely severe seasons 75th percentile Track density of WiTRACK Tropical events occurring between August and October, 1981-2016. Observations from the International Best Track Archive for Climate Stewardship (IBTrACS). Ensemble mean Track Density of SEAS5 shown, for 51 members. February • It is difficult to assess the validity of this proposed climate relationship Floyd Sandy Discovery rate Interannual variability due to the relatively short timescale of best track data. Here we outline Event SSI > Total number of WiTRACK events per year, Matched an approach reutilising a seasonal forecast to address this problem by 90th percentile 1981-2016. For each box, the central mark Unmatched indicates the median and the bottom and top analysing the full ensemble of 1800 forecast years. edges the 25th and 75th percentiles, respectively Kyrill Klaus ERAinterim Experimental Design Event SSI > WiTRACK events 98th percentile in exceedance of 98th SSI percentile. Kyrill WiTRACK events matched to IBTrACS ERAinterim WiTRACK events matched to major hurricanes Light blue uncertainty range • Build event climatology from Ensemble Prediction System, examining synoptic formed January if overlapping by four 6 hourly th Floyd (formed September 7th 1999) and Sandy (formed represents 90 percentile range of Track density of WiTRACK windstorm events, each grid cell ranked as a percentage 15th 2007, Klaus timesteps, each within 400km. scale variability by increasing the sample size of “observed” events October 22nd 2012). Best Track from IBTrACS shown in SEAS5 ensemble. All variables from lowest to highest. For ERAinterim, the top 2% (bottom left) represents 12 formed January Hurricane categories defined following blue crosses. significantly correlated (r>0.5) (Osinski et al. 2016) events; in SEAS5 the top 2% (bottom middle) epresents 422 events. 23rd 2009. Saffir–Simpson scale • Necessary to build stable statistical analysis of severe storm events Climate Drivers

Extratropical Cyclone Climatology Climatology El Niño phase Intense El Niño phase Hypothesized pathways (n=576 years) (n=576 years) (n=157 years) (n=55 years) El Niño induced Rossby Wave Atlantic Tripole Arctic Oscillation

Wind field size for storm duration between 30 and 84 h; (a) for ERA-Interim and (b) for ECMWF EPS. From Osinski et al. (2016) 1. Event Climatology and Verification • 51 Ensemble member ECMWF seasonal forecast product (SEAS5), initialized in August of each year 1981 to 2016 Atlantic Tripole (n=58 years) NAO/AO (n=58 years) • Cyclone tracking of both Atlantic Tropical Cyclone (TC) and (ETC) seasons, using 98th percentile exceedance of 10m wind (WiTRACK; Positive Phase Negative Phase Positive Phase Negative Phase e.g. Scaife et al. 2017 e.g. Fan and Schneider 2012, e.g. Smith et al. 2010 Leckebusch et al. 2008, Kruschke 2015) Wild et al. 2015,Dunstone et al. 2016 • Compare SEAS5 tracks to observed best tracks and to tracking in reanalysis to verify SEAS5 climatology matches observations • Initial viability test of • Event intensity calculated using Storm Severity Index (SSI; Leckebusch et al. hypothesized pathways using 2007), which scales with the third power of surface wind to match observed 16 of 51 available members. insurance loss-wind relationship. • Composites constructed in 2. Identify Climate Drivers SEAS5 of three climate indices: El Niño (Nino3.4), • Propose pathways in the climate system between the Atlantic Hurricane and El Niño phase defined by ENSO3.4 index of more than 0.5, Intense El Niño • El Niño associated with decrease on order European windstorm seasons, based on current literature Atlantic Tripole (near right) phase by ENSO3.4 index of more than 1.5. Positive and negative phases of the of 0.5 Tropical per forecast year and North Atlantic Oscillation Tripole and NAO/AO defined as top 10% and bottom 10% from EOF • Extratropical cyclone storm track shifts • Examine interannual variability between large scale climate drivers, TC and ETC tiemseries respectively. Hatching indicates significance at 95% confidence level tracks (far right) Left: First rotated EOF mode of Atlantic SST anomaly for December – February. in a two-tailed t-test. associated with Atlantic Tripole and NAO Right: First rotated EOF mode of 700mb Geopotential Height anomaly for variability • Build composite climatology of positive/negative phases of climate drivers December – February. Both constructed over 576 SEAS5 ensemble years

References: Acknowledgements: Dunstone, N., Smith, D., Scaife, A., Hermanson, L., Eade, R., Robinson, N., ... & Knight, J. (2016). Skilful predictions of the winter North Atlantic Oscillation one year ahead. Nature Geoscience, We would like to acknowledge funding from CoreLogic and thank Summary 9(11), 809. Fan, M., & Schneider, E. K. (2012). Observed decadal North Atlantic tripole SST variability. Part I: noise forcing and coupled response. Journal of the Atmospheric Sciences, 69(1), 35-50. F. Eddounia and F. Chopin for valuable input and discussion. Jonkman, S. N., Godfroy, M., Sebastian, A., & Kolen, B. (2018). Brief communication: Loss of life due to Hurricane Harvey. Natural Hazards and Earth System Sciences, 18(4), 1073-1078. Classification of TC and ETC events developed in collaboration • ECMWF seasonal forecast product (SEAS5) successfully recreates the climatology of both the Atlantic Hurricane and European Winter Windstorm Kruschke, T. (2015). Winter wind : Identification, verification of decadal predictions, and regionalization (Doctoral dissertation, PhD thesis, Freie Universität Berlin, Dept. of Earth Sciences, Institute of , Berlin, Germany, with colleague at the University of Birmingham, K. Ng. System 5 seasons, increasing available storm “observations” by a factor of 50 Leckebusch, G.C., U. Ulbrich, L. Fröhlich, and J.G. Pinto, (2007): Property loss potentials for European mid-latitude storms in a changing climate. Geophys. Res. Letters, 34, L05703, Ensemble forecast data provided by European Centre for Medium- Leckebusch, G. C., Renggli, D., & Ulbrich, U. (2008). Development and application of an objective storm severity measure for the Northeast Atlantic region. Meteorologische Zeitschrift, 17(5), 575-587. Range Weather Forecasts (ECMWF). Osinski, R., Lorenz, P., Kruschke, T., Voigt, M., Ulbrich, U., Leckebusch, G. C., Faust, E., Hofherr T., & Majewski, D. (2016). An approach to build an event set of European wind storms based on • We hypothesize three pathways between the two seasons, consistent with the climate literature ECMWF EPS. Natural Hazards and Earth System Sciences, 16, 255-268 Scaife, A. A., Comer, R. E., Dunstone, N. J., Knight, J. R., Smith, D. M., MacLachlan, C., ... & Belcher, S. (2017). Tropical rainfall, Rossby waves and regional winter climate predictions. Quarterly Journal of the Royal Meteorological Society, 143(702), 1-11. Smith, D. M., Eade, R., Dunstone, N. J., Fereday, D., Murphy, J. M., Pohlmann, H., & Scaife, A. A. (2010). Skilful multi-year predictions of Atlantic hurricane frequency. Nature geoscience, 3(12), • Key Climate drivers are shown to significantly influence the location and frequency of TC and ETC events. Future work will focus on assessing the 846. Wild, S., D.J. Befort, and G. C. Leckebusch. (2015) Was the extreme storm in winter 2013/14 over the North Atlantic and the triggered by changes in the West Pacific independence of the two seasons, testing the proposed hypotheses and establishing the probability of occurrence of successive extreme seasons warm pool?. Bulletin of the American Meteorological Society 96.12 (2015): S29-S34.