Historical and Future Windstorms Affecting the Northeast US: Their Impacts and Origins

Title: Historical and Future Windstorms Affecting the Northeast US: Their Impacts and Origins

Authors: Fred Letson, Jacob Coburn, Xin Zhou, Melissa Bukovsky, Rebecca J. Barthelmie, Sara C. Pryor

Large-scale windstorms associated with extreme synoptic-scale cyclones are a major natural hazard in the Northeastern United States and much of northern Europe. This analysis focuses on the 13 states of the Northeast United States (US), as defined in the US National Climate Assessment. The Northeast is a region of high population density (and thus exposure to wind damage), and exhibits a high frequency of cyclone passages. Based on analyses of the ERA5 reanalysis product most are Alberta Clippers (AC) and Colorado Lows (CL), that are generated in the Lee of the Rocky Mountains. However, Tropical Cyclones (TC) and cyclones with their genesis along the east coast (EC), in the lee of the Appalachian Mountains, also have an impact (Letson et al. (2021): Intense windstorms in the northeastern United States. Natural Hazards and Earth System Sciences, 21, 2001-2020).

Modeling of future windstorms using storylines and/or pseudo-global warming frameworks can be used to characterize the likely impact of global non-stationarity on the dynamics and thermodynamics of historically important windstorms and the responsiveness to land use land cover change (e.g. winter de-icing of the Great Lakes) (Zhou et al. Future Extreme Winter Windstorms in the Northeastern US: a Storyline Based Pseudo-Global Warming Approach, this conference). However, such approaches are predicated on an assumption that the cyclone types associated with extreme wind events in the contemporary and historical climate will also dominate in the future. Equally, large ensembles of Earth System Models (ESMs) provide an excellent tool to quantify the role of internal climate variability in cyclone tracking/intensity/frequency and the resulting implications for windstorms (Coburn et al., Northeastern Windstorms and Midlatitude Cyclones in the MPI Large Ensemble, this conference). However, the resolution of the ESM large ensembles precludes capture of some details of the dynamics of the windstorms. Here we use simulations from a regional climate model nested within three ESM and run at moderate grid resolution to quantify possible macro-scale changes in windstorms and their associated cyclones.

Output from six ESM-nested NA-CORDEX Weather Research and Forecasting (WRF) simulations covering North America at a 25-km resolution and one nested in ERA-Interim are used in the current work (Mearns, L.O., et al., 2017: The NA-CORDEX dataset, version 1.0. NCAR Climate Data Gateway, Boulder CO, https://doi.org/10.5065/D6SJ1JCH). The WRF simulations are subject to nudging at the top of the domain. The ESMs used are from the Geophysical Fluid Dynamics Laboratory (GFDL), the Hadley Centre (Global Environment Model version 2; HadGEM2) and the Max Planck Institute (MPI). Two WRF simulations are nested in each ESM, one representing the past (1950 to 2005) and one, the future (2006 to 2099) using RCP 8.5 climate forcing scenario. The ERA-Interim-nested WRF simulation from 1980 to 2010, is used to represent the ‘ground truth’ contemporary climate. 10-m wind speed fields and sea-level pressure (SLP) with a three-hour time resolution are used to identify windstorms and track their associated cyclones. Gridded population data for the year 2000, and projections for the year 2090 are used to evaluate present and future exposure of humans and infrastructure to wind hazards.

Using a methodology similar to Letson et al. (2021), windstorm events are identified based on widespread, simultaneous exceedance of locally defined wind speed thresholds. The wind speed threshold is the long-term 99.9^th percentile 10-m wind speed (U₉₉₉) for each grid cell and represents low-probability, intense wind speeds likely to cause damage at that location. For each time series the top N events, where N is the number of years, are identified as the events with the highest spatial coverage of exceedance of U₉₉₉. Metrics of windstorm intensity include: storm extent (number of WRF grid cells with simultaneous exceedance of U₉₉₉, including cells outside the Northeast), maximum wind speed, storm duration, storm translational speed, and loss index (a measure of over-threshold winds, weighted by population, as a proxy for likely property damage). Each windstorm is matched with a parent cyclone, identified from local minima in the spatially-smoothed SLP fields in the 7 days leading up to the Northeast wind event. Cyclones are assigned to one of six types based on their region of cyclogenesis. In addition to AC, CL, TC, and EC cyclones, Mid-latitude West (MW) cyclones have their origins in the US and Canada (and offshore), west of the Rocky Mountains. Cyclones originating outside of all five regions are classified as ‘other’.

Long-term trends in U₉₉₉ values are not evident in the seven simulations among grid cells in northeast states, indicating that intense wind speed magnitudes in this region may remain relatively constant over time. The top N windstorms (simultaneous exceedances of these thresholds) in each simulation vary in extent and cover between 12 and 52 % of Northeast land area, with WRF nested in ERA-Interim exhibiting the most widespread exceedance of U₉₉₉. In all seven WRF simulations, windstorms are heavily concentrated in the cold season (October to April). All six (past and future) ESM-nested simulations show a larger proportion of Northeast windstorms associated with TCs than does the ERA-Interim-nested run. While decadal variations in cyclone types responsible for the windstorms are evident, the only statistically significant secular trend identified is an increase in the number of MW cyclones in the WRF simulations with the future HadGEM2 lateral boundary conditions. Loss indices show a strong upward trend over time, due primarily to the expected future increase in population, but also, in the case of the MPI-nested model, due to increases in intense wind speeds in highly-populated areas. Storm durations over 20 hours are more common in all three future simulations than in their corresponding historical simulations, indicating that the Northeast may be subject to an increased windstorm hazard from changes in storm duration in the coming decades.