ARkStorm Retrospective—Making Sense of the Extreme Events of Winter 2017 in California By Comparisons to a Well-Explored Extreme-Storms Scenario

Winter 2017 in California was a time of monstrous extremes. Following severe warm-drought conditions that spanned most of 2012-2016, autumn 2016 was notably wet in northern California (receiving almost twice normal precipitation) but, because the storms were early and warm, snowpacks were only about 50% of normal by December 31. Thus 2017 began with the State still flirting with long-term drought. However, unusually intense sequences of atmospheric river began making landfall in early January and, by January 10, the snowpack in parts of the Sierra Nevada was 250% of normal for January. Atmospheric rivers continued to arrive through much of the month, and by early February, the range’s reservoirs were also quite full. When another intense string of atmospheric rivers arrived in early February, between rapid inflows to an already full reservoir and damages to the primary spillway there, events at Lake Oroville got out of hand, and 180,000 people in communities downstream were evacuated in response to overflow-fueled erosion of the emergency spillway. Oroville wasn’t the only reservoir where operators were scrambling to avoid dam-safety and downstream-flooding problems. Soon, broad areas of the Central Valley were flooded and many communities were dealing with their own wetness-based hazards and disasters. The same storms continued to pile huge amounts of snow in the Sierra Nevada, causing many infrastructure, transportation, and economic disruptions, as well as damage to many structures there. Then when temperatures warmed and spring snowmelt commenced, the area—so recently in drought—was faced with far more snowmelt (3 and 4 times more) anticipated than space to store it in the region’s reservoirs, so that another scramble was on to figure out where and when to spread the overly large flood flows on both sides of the Sierra Nevada, for the second time that year.

Seven years earlier, the USGS Science Applications for Risk Reductions program developed a scientifically plausible, exquisitely detailed scenario of a severe-storm sequence (called the ARkStorm scenario), based on then-emerging atmospheric-river science and examples from several past storm sequences that have impacted California (e.g., storm sequences in 1997, 1986, 1969, and even 1862—the worst of them all). The program then worked with subject experts, agencies, and stakeholders across California and, especially in northwest Nevada, to identify likely impacts in all sectors, likely responses by communities and emergency responders, mitigation actions that could reduce the impacts, and scientific needs to reduce risks. In the Lake Tahoe-Reno-Carson City area of northwestern Nevada in 2013, over 100 agencies and organizations were engaged in meetings with over 300 stakeholders from a full range of sectors. When the full range of sectors that would be impacted were considered, economic analyses of the expected storm and flood damages and long-term economic disruptions in California as a whole estimated that the ARkStorm might cost about ¾ of a trillion dollars in the long run (roughly 50% of the State’s annual GDP).

As it happens, the storms of winter 2017 dropped very nearly the same amount of precipitation as was simulated for ARkStorm. However, the amount of precipitation that fell in 23 days in the ARkStorm scenario fell over the course of 80 days in winter 2017. Because the storms were more “spread out” (in time) in winter 2017 than in ARkStorm, flooding in 2017 was not as extensive as was expected from ARkStorm, and impacts were certainly far less, although still significant and taxing. Nonetheless, the events of winter 2017 offer a (thus far) unique opportunity to revisit the many stakeholder discussions and conclusions that were part of the ARkStorm project. A few of the lessons learned thus far are (1) even storms that do not approach precipitation (or runoff) amounts comparable to probable-maximum-precipitation (PMP) thresholds are quite capable of pushing reservoir operators and dam-safety concerns to their limits (even though, in ARkStorm, dam safety was generally dismissed as unlikely to become a major issue because the precipitation totals were nowhere near PMP levels), (2) snow-mass damages to structures in the Sierra Nevada were far greater and more widespread than recognized in discussions of ARkStorm, and (3) the hazards associated with drought-busting storm sequences were found to be greater and much more varied than previous discussions of “drought busting” had appreciated, including of course flooding but also subsequent outbreaks of invasive species and other pests, new damages to agricultural fields already damaged by drought, and enhanced wildfire risks after the drought was broken.

This presentation will compare and contrast the events of winter 2017 and ARkStorm, including what the ARkStorm scenario got right, what it got wrong or missed, and how the scenario process might be improved to better identify the most pressing challenges and opportunities inherent in major storm sequences on the West Coast.