Northeastern Windstorms and Midlatitude Cyclones in the MPI Large Ensemble

Extreme windstorms resulting from midlatitude cyclones have significant impacts in highly populated regions such as the Northeastern United States (NEUSA). Quantifying the cyclone types responsible for these events along with their scale and intensity in current and possible future climate conditions is essential to building regional resilience. Past work has focused on the types of cyclones, their characteristics and the wind hazards they produced over the historical period (Letson F., Barthelmie R.J., Hodges K. and Pryor S.C. (2021): Windstorms in the Northeastern United States. Natural Hazards and Earth System Sciences 21 2001–2020), applying pseudo-global warming experiments to a series of three extreme cyclone storylines (see Xin et al. 2024, Future Extreme Winter Windstorms in the Northeastern US: a Storyline Based Pseudo-Global Warming Approach, this conference), and event identification and cyclone tracking within the regionally downscaled NA-CORDEX suite of regional climate models (RCM) (Letson et al. 2024, Historical and Future Windstorms Affecting the Northeast US: Their Impacts and Origins, this conference). Although use of 12 km grid spacing simulations with RCM allows clear identification and characterization of windstorms, the NA-CORDEX model ensemble under-samples the full range of possible climate futures. In this work, output from a moderately sized ensemble generated using an Earth System Model is used to identify windstorms and their parent cyclones so as to examine sources of variability in windstorm events and parent cyclone types under historical and future climate states, as well as to identify ensemble members that are outliers in terms of these properties.

Large ensembles are advantageous in that they produce a range of possible variations under the same boundary conditions, external forcings and model parameters. Three-hour mean sea level pressure (SLP) and 10 m near-surface wind speed output at 1.875° resolution from thirty members of the Max Planck Institute (MPI-ESM-LR) ensemble are used to identify cyclones associated with the most intense windstorms which pass over the NEUSA region, define their tracks and source regions and assess their characteristics over the historical (1850-2014) and projected (2015-2100) periods. Projections are generated for four Shared Socioeconomic Pathways (SSPs) simulations with low (2.6 Wm^-2), moderate (4.5 Wm^-2), high (7.0 Wm^-2) and extreme (8.5 Wm^-2) net forcing by 2100. Historical validation is conducted using ERA5 using once-hourly wind speed and SLP output at 0.25° grid resolution over 1981-2022 which has been re-gridded to match the ensemble resolution and temporally averaged to match the 3-hour time step in MPI.

A cyclone tracking algorithm is applied which works by first identifying the most extreme windstorms, defined here as the 3-hour periods in which wind speeds exceed the 99.9^th percentile (U₉₉₉) at each grid cell over at least 10 percent of the NEUSA region. Use of a local threshold (U₉₉₉) is designed to account for differences in surface roughness lengths that are critical to determining the absolute magnitude of the wind speed. For each time step where the local U₉₉₉ is exceeded in > 10% of NEUSA grid cells, the nearest low-pressure center in the smoothed SLP field within the NEUSA is found and a back-trajectory of the cyclone is found using the minimum SLP within a 375 km radius of the center at each previous timestep. This process is repeated until a low-pressure center can no longer be detected. Cyclone central pressures and locations, as well as peak wind speeds, are reported for each timestep. Cyclone types are assigned by the region in which their track begins, including Alberta Clippers (AC), Colorado Lows (CL), Pacific Midlatitude storms (PM), Tropical Cyclones (CL), Eastern Lows (EL) and Other (O). Bomb cyclones are identified using bergeron units, where the scaled 24-hour change in SLP surpass a latitudinally-dependent threshold. Time series of cyclone types and windstorm characteristics are calculated for each ensemble member. Specific focus is given to members of the ensemble that are statistical outliers, in which windstorm and cyclone occurrence and characteristics are outside of the 95^th percentile of the ensemble range.

Externally forced changes in windstorm and cyclone characteristics are assessed by comparing outcomes from members that sample across SSP projections. Internal forcing is assessed using the phases of internal climate modes. Stepwise regression is used to model the relationship between cyclone types and windstorm characteristics on five mode indices derived from each member which may impact circulation patterns over the NEUSA, including the Northern Annular Mode (NAM), the Pacific – North American pattern (PNA), El Niño – Southern Oscillation (ENSO), the Pacific Decadal Oscillation (PDO) and the Atlantic Multidecadal Oscillation (AMO).

Analysis of output from the historical and high radiative forcing (8.5 Wm^-2) model members shows that the ensemble mean (95^th percentile range) total number of windstorm events identified over 1850-2100 is 321 (302-348), with two outlier members in which 350 and 300 storms occurred, respectively. The majority of storms peak in the NEUSA, with SLP minima ranging from 930-1005 hPa which typically occurred within 6 hours of peak windstorm coverage. The bergeron index indicates that the number of bomb cyclones in each member ranges from 80-114 over the 251-year record, or about 25-33 percent of all cyclones driving the extreme windstorms in the NEUSA. This presentation will provide details of the windstorm and cyclone characteristics, quantify agreement between the ESM ensemble and ERA5 plus the importance of external and internal forcing of variability and explain causes of the outlier members that have windstorm characteristics that deviate substantially from the other ensemble members.