Atmospheric rivers (ARs) are long, narrow bands of intense moisture that originate in the tropics and travel long distances through the sky. Scientists have confirmed a link between ARs and extreme precipitation events on the U.S. West Coast. However, the likelihood and the characteristics of ARs to trigger extreme precipitation events have not been investigated. Researchers at the U.S. Department of Energy’s Pacific Northwest National Laboratory conducted regional modeling and analysis to examine the relationships between ARs and extreme precipitation. They found enough causality to provide predictability for the latter at watershed scale. They also found that two AR types—flash ARs (above-average intensity) and prolonged ARs (long duration)—accounted for less than 50 percent of the total AR events but dominated the relationships between ARs and extreme precipitation events.
Understanding the predictability of extreme precipitation events and improving their forecasts are important for emergency preparedness and building resilience. By relating extreme precipitation to large-scale ARs and identifying the AR types that contribute more to extreme precipitation events, this study provides advanced insights on forecasting of hydrologic extremes.
The research team analyzed precipitation data from a decades-long, high‐resolution regional climate simulation at 6-kilometer grid spacing and ARs identified by the Atmospheric River Tracking Method Intercomparison Project via different AR tracking methods. Regardless of the detection method, scientists found a correlation between the daily occurrence of ARs and extreme daily precipitation in the western United States. Researchers also found a single high-level index that combines AR intensity, duration, and landfalling area to be a good predictor of extreme precipitation amount on a monthly timescale. The findings suggest extended predictability of extreme precipitation—in both occurrence and magnitude—in the Pacific Northwest and California, as operational models have useful skill in forecasting ARs with a lead time of seven to 10 days.
Researchers also found that extreme precipitation in the western United States was closely related to two AR characteristics: their intensity and duration after making landfall. By using machine learning techniques, scientists proposed to classify ARs into three categories—weak ARs, flash ARs, and prolonged ARs. The team found that flash ARs and prolonged ARs—with above-average intensity and long duration, respectively—correlated more strongly with extreme precipitation. This work identifies the feasibility to improve short-term extreme precipitation forecasts along the West Coast and provides new insights on prioritizing the categories of ARs that require more attention in future research.