Droughts in North America are common, but spatially widespread droughts that span most of continent stand out as prominent and relatively infrequent events that pose significant challenges. Such pan-continental droughts claim higher costs when compared to localized droughts of similar severity because, by definition, they impact more areas while also reducing redundancies in associated water or agricultural resources that might act to mitigate more regionalized events. For instance, the 1988 US drought ranked among the most spatially extensive in recent history and cost the U.S. $40 billion, largely in agricultural losses. Similarly, the 2012 US drought, which covered over 60% of the contiguous U.S. and is the most recent pan-continental drought, cost an estimated $30 billion, again largely in agricultural losses. Understanding the causes, characteristics, and dynamics of these droughts is thus important for developing adaptation strategies and defining their potentially changing frequency and severity in a warming climate.

The primary motivation of this project was to determine how frequently pan-continental droughts occur, understand their causes, and determine how their risk changes in the future under increasing greenhouse gas concentrations. 

We used paleoclimatic evidence, observational records, and model simulations to estimate the historical incidence of pan-continental droughts in North America.  These droughts occur an average of about 20 times per century and have been a consistent feature of U.S. hydroclimate over the last thousand years or more.  We subsequently confirmed that variations in sea surface temperatures strongly influence the occurrence of pan-continental droughts.  There is a particularly strong association between these events and La Niñas in the tropical Pacific Ocean.  We nevertheless found little association between pan-continental droughts and sea surface temperatures in the Atlantic Ocean, which is a notable finding considering the influence of the Atlantic on other hydroclimate variability in North America.  We also demonstrated that internal atmospheric variability (weather) has a strong influence on the probability of pan-continental drought occurrence.  These findings are important with regard to efforts to predict the occurrence of pan-continental droughts before they happen.  While we can predict La Niñas with a reasonable degree of success on 6-12 month timescales, atmospheric variability is inherently unpredictable beyond several weeks.  Advance prediction of pan-continental droughts will therefore be very limited. 

The work on the project has also refined understanding of droughts that occur in specific areas of the continental U.S., with implications for how regional dynamics may combine to generate pan-continental droughts. We determined the causes of the 2016 U.S. southeastern drought, which was tied to moderate La Niña conditions and internal atmospheric variability. Our understanding of the causes of the drought supported our larger interpretations of the causes of more broadly distributed pan-continental droughts.  We similarly investigated the pattern of widespread coastal droughts that occur along the western coast of the U.S. Overturning contemporary thinking, we showed that the leading pattern of cold-season precipitation variability along the western U.S. coast favors spatially coherent precipitation anomalies instead of a north-south dipole pattern. Sixteen coastal-wide droughts occurred from 1895–2016, that clustered during the 1920s–1930s and post-2000 periods, all of which were most strongly linked with internal atmospheric variability.

Another important aspect of pan-continental droughts is the degree to which they can be impacted by multidecadal variability, of the type that is known to have caused megadrought events during the Medieval period.  Our work showed that the occurrences of pan-continental drought events were most frequent during megadrought periods. We also showed that megadrought clustering during the Medieval period is potentially the result of natural variability in ocean conditions. We found important connections to both the tropical Pacific and the Atlantic, suggesting that natural ocean variability in both basins may have been important for explaining past megadrought clustering.

Given the importance of the tropical Pacific Ocean in our pan-continental framework, we additionally clarified some specifics of tropical Pacific variability and how they translate into hydroclimatic connections on land.  These investigations led to improved understanding of how specific patterns in sea surface temperatures in the tropical Pacific Ocean are related to drought and its predictability in the U.S.

Finally, we worked to connect our findings to their agricultural implications.  The conditions that lead to pan-continental droughts in North America depress crop yields.  We specifically showed that tropical Pacific Ocean conditions account for significant variability in U.S. maize, soybean, and wheat production.

This award supported the development of multiple large ensembles of atmospheric model simulations in which different configurations of sea surface temperatures were used.  To investigate the influence of soil moisture feedbacks, a large ensemble was also derived for conditions in which soil moisture was constrained to vary only according to seasonal climatology.  All of these simulations have been publicly archived and are freely available without restriction.

This project partially supported the doctoral work of two graduate students.

Last Modified: 09/27/2018
Modified by: Jason E Smerdon