Research

How will human activity, natural variability, and seasonal extremes shape the emerging climates of the 21st century? How can we use information about high magnitude, low frequency, climate variations in the past to constrain the future? What are the key dynamical processes in the Northern Hemisphere governing the seasonal cycle of atmospheric flow, and how are they likely to change in a warming climate?

At the Emergent Climate Risk Lab (ECRL), we address these questions through our research into the following interrelated areas of climate science: (1) prolonged drought and megadrought occurrence; (2) the timing, causes, and impacts of spring onset in the Northern Hemisphere; (3) tropical Pacific decadal variability; and, (4) near-term weather prediction using novel systems for collecting near real-time data. By integrating observational data, global climate model output, and paleoclimate insights, our team has published papers on climate variability on sub-seasonal to multi-century time horizons, and regional to global scales.

While ECRL has focused primarily on North America and the tropical Pacific, the core ideas underpinning our recent advances are broadly applicable to other parts of the world, and we are actively expanding the scope of our research program to include cross-campus collaborations in the area of Cornell’s Digital Agriculture initiative, and new basic research into the dynamics, thermodynamics, and predictability of Caribbean drought, and its effects on food security, nutrition, and migration.

Our work on drought has made national headlines, been covered by major news outlets in both English and Spanish, and been featured by scientific and public media outlets such as Physics Today and The Atlantic, New York Times, National Public Radio, and others. Below, we summarize a few key advances from our recent research efforts at Cornell that represent the nature of our work and its broader impacts.

Drought and Multi-decadal megadrought
Paleoclimate evidence suggests that multi-decadal periods of aridity—megadroughts—have occurred in North America and other parts of the world within the last millennium. These events were as severe as, but more prolonged than, even the worst multi-year droughts in the US (e.g., 1930s dust bowl). And, they may have been connected to the demise of several preindustrial civilizations. While it has long been recognized that such events occurred in past centuries, our research into megadrought has attempted to estimate how likely these events will be in the future (Ault et al. 2014; Cook et al. 2015; Ault et al. 2016), as well as defining a robust null hypothesis for their occurrence without climate change (Ault et al., 2018).

kkkk
From Ault et al. (2018). A time series of PDSI from one LIM realization (a) and the associated SST anomalies (b) and SW PDSI (c) at the red circle.

From 2013-2016 the Caribbean region endured the worst drought in over half a century (Herrera and Ault, 2017). Although it received little attention in English-language media, it was as bad or worse than the recent California drought and brought widespread crop losses, hampered energy production, and threatened food security for millions across the region. Our group developed a first-of-its-kind Caribbean Drought Atlas to characterize historical and ongoing changes in the region. Further, we found that higher temperatures from human-induced climate change made this “Pan-Caribbean” drought even worse than it would have otherwise been (Herrera et al., 2018).

pcd
From Herrera and Ault (2017) major droughts registered in the study area from the perspective of the Caribbean between 1950 and 2016 of at least one year in duration.

Spring onset & seasonal prediction
During springtime, winter snowpack melts and fills reservoirs, plants flower and develop leaves, invertebrates and vertebrates migrate and reproduce, and farmers plant crops and prepare for the coming growing season.  At the same time, the terrestrial biosphere modifies the flow of energy, water, carbon, and other gasses through the atmosphere. That is, spring is the phase of the annual cycle when––across much of the northern hemisphere––the atmosphere transitions from a hostile winter regime into one that allows plant and animal life to flourish. And, it encompasses a myriad of large-scale processes in the circulation of the atmosphere, as well as the coupling of the land-surface to the atmosphere. We are interested in understanding the trends, atmospheric dynamics, land surface physics, ecological impacts, and potential predictability of spring onset (Ault et al., 2015; Carrillo et al., 2018).

Tropical Pacific Decadal Variability
There is an unresolved and intriguing disagreement between the statistics of paleoclimate and climate model data. In the tropics, and especially in the tropical Pacific, the amplitude of decadal to centennial variations appears to be greater in paleoclimate archives and reconstructions than in Global Climate Models (e.g., Ault et al. 2013; Laepple and Huybers, 2014; Parsons et al., 2018). If internal climate variability is “strong,” then it will modulate the forced responses of the climate system, causing warming trends to look more episodic than linear. In contrast, if decadal climate variations are weak, then the forced responses will tend to be more monotonic, responding directly to changes in boundary conditions.

pdo
From Newman, Alexander, and Ault et al. (2016) Cold season relationship between climate indices discussed in this paper and U.S. precipitation and temperature anomalies determined from U.S. climate division data (Vose et al. 2014), for the years 1901–2014. NDJFM U.S. precipitation anomalies correlated with (a) the PDO index, (b) the ENSO index, and (c) the NPI.NDJFM U.S. temperature anomalies correlated with (d) the PDO index, (e) the ENSO index, and (f) the NPI.

The origins of this proxy-model mismatch are unclear, and remain an area of active research. On the one hand, the proxies themselves could be artificially attenuating high-frequency signals and amplifying low-frequency ones through biological or physical sources of autocorrelation (Ault, 2017). On the other hand, the models tend to have El Niño and La Niña fluctuations that are excessively energetic, and may interannual variations dominate model spectra (Parsons et al., 2018). Furthermore, Global Climate Models can correctly simulate the continuum of climate variations across a wider range of timescales (Zhu et al., 2019), yet it is unclear if they actually do so on decadal to multi-decadal ones.

Our current results suggest both factors are at play: paleoclimate archives support higher amplitude decadal variability than seen in models, but model spectra are biased, with respect to observations and proxies alike, in their interannual El Niño amplitudes. These findings hold serious implications for megadrought risk assessments, among other hazards, because the tropical Pacific strongly impacts North American climate. If it varies too strongly and too frequently on interannual timescales in models, then this effect will tend to mask the true risk of megadrought.