This study improves on previous studies of extreme weather developed for Seattle City Light.
In a previous study, City Light evaluated changes in temperature extremes using statistically downscaled climate projections (Multivariate Adaptive Constructed Analogs; Abatzoglou and Brown, 2012). This work improves on that study by using dynamical downscaling, which is expected to better capture local changes in climate than statistical downscaling by explicitly representing the physical processes governing the climate at each location.

Another study led by Salathé et al. (2015) used dynamical downscaling to evaluate changes in extreme winds, using the two Weather Research and Forecasting climate simulations that were available at the time (WRF, http://www.wrf-model.org; Skamarock et al., 2008). That study found no significant changes in winds, in part because with only three model projections it is not possible to distinguish a systematic change from one that is a result of sample bias. The same study also evaluated changes in lightning using the composite Thunderstorm Prediction Index (Knapp et al, 2006; Knapp and Brooks, 2000). The study concluded that a larger ensemble of WRF projections would help determine if a systematic trend is present, and recommended further analysis of both the lightning index and WRF representation of lightning conditions.

With support and collaboration from the Climate Impacts Group, Cliff Mass, professor of Atmospheric Sciences at the University of Washington, has recently produced 12 new regional climate model projections. Simulations were performed using WRF for the years 1970 through 2099, all driven by the high-end RCP 8.5 greenhouse gas scenario (Van Vuuren et al., 2011):

The work described on this page builds on the 2012 and 2015 efforts: evaluating the results from the new projections to better understand changes in extreme winds, lightning, and heat waves for locations and metrics of interest to City Light.

Working in close collaboration with Seattle City Light personnel, we identified 15 extreme weather metrics related to temperature, wind and lightning. Metrics were selected based on relevance to City Light operations, planning and engineering standards. With input from City Light staff, we identified observations that could be used as historical benchmarks for each metric.


To look at climate change impacts we used the new dynamically-downscaled regional climate model projections to estimate changes in each temperature, wind and lightning metric. These changes were then applied to the historical estimates, from the observations, to obtain projected future conditions for each metric. Results were combined into a set of fact sheets—one for each climate metric—for use by City Light staff.

Projected changes in temperature show the effects of warming across all metrics, with hotter and more frequent heat waves, fewer and less extreme cold snaps, more cooling degree days, fewer heating degree days and fewer freeze-thaw cycles in the future.

Wind extremes, in contrast, show little to no change. There is some tendency among models to show a decrease in the number and severity of peak wind events, but there is wide disagreement among models and in almost all cases there are some models showing increases while others project decreases in extreme winds.

For lightning, we recommend further research to better understand the strengths and weaknesses of different approaches. Nonetheless, we did evaluate projected changes based on the TPI index, finding the potential for increased lightning risk at all locations but no appreciable change in the seasonality of lightning strikes.

Heavy Precipitation Projections for Use in Stormwater Planning: This project summarizes new data for use in stormwater and CSO planning, and developed an online tool that allows users to evaluate projections as a function of precipitation intensity, duration, and frequency.

Regional Modeling for Windstorms and Lightning: This report presents results for the analysis of extreme wind events and lightning risk as simulated in a regional climate model for the historic period 1970-2000 and projected future period 2040-2070. These results are based on three regional climate simulations using the WRF (Weather Research and Forecasting) regional climate model, each driven by a different global climate model.