By Elisabeth Gawthrop, Climate and Society ’13
Three of North America’s major rivers run through the Midwestern U.S. In the spring of 2011, major flooding in region caused an estimated $3 billion in damages and killed seven people. Although scientists cannot predict exact precipitation amounts for a given season, they can attempt to predict the odds that a given season will have below average, average, and above average precipitation. If forecasts show an increased likelihood for above average precipitation, the odds of flooding usually increase, too. The International Research Institute for Climate and Society’s Andrew Robertson studies how climate variability across multiple timescales, from daily to decades, affects these forecasts. Using the American Midwest as a case study, Robertson and his colleagues at the Lamont-Doherty Earth Observatory and Columbia Water Center analyzed the relationships between flooding events and weather and climate patterns on multiple timescales over the 20th century. Find out more about how Robertson and his colleagues are trying to improve flood prediction in the Q&A below and stop by his talk at AGU.
How do El Niño–Southern Oscillation (ENSO), Madden Julian Oscillation (MJO) and Pacific Decadal Oscillation (PDO) interact to make climate patterns more or less favorable for precipitation in the region?
We have analyzed recurrent daily atmospheric circulation patterns, attempting to link these daily patterns to patterns of longer time scales and covering wider regions and, separately, to extreme floods. We found some weak but statistically significant linkages between weather patterns associated with both floods and ENSO/MJO patterns. La Niña, the cool phase of ENSO, tends to cause a large-scale pattern that’s more conducive to creating the conditions that lead to floods the Midwest. We also found that an active MJO event tends to lead to cause an atmospheric “wave” that passes over the Midwest two weeks following the event. This wave is also conducive to floods. Even though we have a century of records, it’s too short to say much about relationships with the PDO, a phenomenon that shifts over a period of decades as opposed to the monthly and seasonal fluctuations of ENSO and MJO.
What is the skill of your results? Will forecasters be able to incorporate more of these longer time scale variabilities into seasonal forecasts?
The prospects for improved seasonal forecasts are limited because the ENSO linkage is weak. There are better prospects for eventually developing “seamless” forecasts in which forecast information is combined together and capitalizes on the MJO relationships. Particular combinations of ENSO and MJO could lead to better “forecasts of opportunity” in situations where both ENSO and MJO impacts are reinforcing each other.
Is there a human-induced climate change signal that could change these relationships in the future?
The century-long record of floods does not reveal an increasing trend toward more frequent extreme floods over the Midwest signature. Many of the flood events occurred in the early and midcentury, with fewer at the end of the twentieth century.
Why did you focus on precipitation from March through May in the Midwest in particular?
The spring season over the Midwest is a time of heightened flood risk, due to potential confluence of factors conducive to floods. Combinations of snow melt, high ground saturation, and strong interactions between Gulf of Mexico moisture and slow moving cyclones that can occur in the spring lead to increased likelihood of flooding events.
I came across this doing some research after the recent (Spring 2013) floods in the Midwest. It’s really informative and puts things into perspective, especially when you say “The century-long record of floods does not reveal an increasing trend toward more frequent extreme floods over the Midwest signature”.