Research Statement, June 2006

Gidon Eshel

401 Hinds
Dept. of the Geophysical Sciences
5734 S. Ellis Ave.
The Univ. of Chicago
Chicago, IL 60637
Office: (773) 702-0440
Home: (773) 955-4058
Email: geshel@uchicago.edu


My research seeks to understand geophysical phenomena with great human relevance, mostly employing methods and ideas of classical physics and statistics. My thematic focus is dynamics-based ocean-atmosphere diagnostics, with a geographic focus on the arid subtropics, especially the Red Sea-Mediterranean region. Compared to the deep tropics or midlatitudes, the subtropics are not dynamically and climatically well characterized, nor are they so intensely studied. Yet, they are home to a large portion of the global human population, and are the locus of intense water scarcity-related strife. The physical and dynamical mechanisms that govern subtropical precipitation processes vary on a broad spectrum of spatial and temporal scales, and touch on most aspects of atmospheric science: dynamics; dry and moist thermodynamics; turbulent surface fluxes of momentum, water vapor and heat; cloud physics. My research straddles many of those topics as I strive to better understand the mechanisms of modern and paleo subtropical climate variability, especially subtropical precipitation processes, with the goal of improved prediction. Most of my research projects involve interactions of the atmosphere with the underlying ocean and land. Methodologically, all combine modeling - most often using simple, home-brewed models - with advanced and novel statistical data analysis methods designed to take advantage of large modern geophysical datasets. A subset of my recent, current, and planned research endeavors is described below.


Outer-Tropical Climate Mechanisms and Predictability. Climate research is currently in a position similar to that of weather research just after World War II. While society urgently needs reliable climate predictions to mitigate some of the human costs of extreme events such as droughts or heat waves, the scientific basis for such predictions is nascent. Given that climate is the long-term average of weather, it is natural that the principal tool used for climate prediction has been that used for weather forecasting: numerical models expressing the physical laws that govern climate. While a powerful tool, numerical climate models have obvious limitations. To be tractable, models neglect or crudely parameterize significant aspects of the relevant physics. In addition, climate can sometimes exhibit chaotic behavior, and estimates of its state are under-resolved, resulting in contamination of long model runs by exponentially growing error. While modern ensemble methods ameliorate some of those difficulties, their full, statistically robust, implementation is still prohibitively expensive in most cases. Because numerical long-lead forecasts are currently uncertain, the spatio-temporal statistical approach to interannual climate forecasting provides a useful interim approach until numerical models, and the hardware on which they run, mature further. Following several papers which collectively demonstrated the relationship of Middle-Eastern droughts to earlier North Atlantic (NA) climate anomalies (Eshel 2002; Eshel and Farrell 2001; Eshel, Cane and Farrell 2000; Eshel and Farrell 2000), I devised a statistical interannual climate prediction model of those droughts. The model is data-driven, choosing predictors from several three- and four-dimensional geophysical datasets based on optimized space reduction. Later adopted by the Israeli Meteorological Service, the model has led to substantial and verifiable improvement of their seasonal forecasts. In Eshel (2003a,b), I applied similar ideas to the principal mode of climate variability outside the tropics, the so-called NA Oscillation (NAO). My paper on NAO interannual prediction is the first NAO forecasting attempt, and its skill is still comparable to that of numerical model-based forecasts. My interest in atmospheric predictability has also led to investigating more fundamental questions of rates of information loss (in the Information Theoretic sense) in the ocean and atmosphere, as reported in Eshel (2003; JGR. 108(C2)) and Eshel (2006; JAS 63(2)), respectively. The former of these papers identified a location in the NA that is singularly revealing of the state of the NA ocean as a whole. The detailed examination of deep-sea cores from this location is the centerpiece of a collaborative paleoclimate project (with K. Billups, Delaware, and P. Martin, Chicago) currently underway. In the near future I plan to extend my statistical NAO predictability study to the underlying physics. I will be guided by the mechanism I view as most viable, in which snow cover variability over mountainous western North America plays the pivotal role in the Pacific-NA teleconnection. The basic idea is to present a model atmosphere with patterns of Pacific surface anomalies that were shown to be powerful predictors of future states of the NA, and to diagnose the dynamical response. These spatial patterns are not derived in the usual way, which is to maximize spanned variance. Rather, the choice of these patterns is based on their being linear optimal excitations, affecting future phase space trajectories most strongly for a given anomaly magnitude.


Coastal Upwelling and the African Humid Period. North Africa, currently extremely arid, is believed to have been much more hospitable ~8-6 kyBP (thousand years before present). The mild, moist climate of the region during this African Humid Period has been invoked as a central element in the ecological success of humans (e.g., deMenocal et al., 2000: Quat. Sci. Rev., 19, 347-361). Much of what is considered known about the climate of that period has been derived from subtropical Atlantic deep-sea cores. Complicating the interpretation of these cores is the fact that they record both continental airborne material of African origin and biogenic material produced in the intensely productive upwelling zone off northwestern Africa. Better understanding of the interplay between North African meteorology and coastal upwelling off northwestern Africa is needed to improve interpretations of the evidence for the African Humid Period climate. Jess Adkins (CalTech), Peter deMenocal (Columbia) and I have recently launched a collaborative effort striving to understand this interplay. In the first paper resulting from our work (Adkins, Eshel and deMenocal 2006), we analyze paleo information along with some modern ocean-atmosphere data. The followup study (Eshel, Adkins and deMenocal, in preparation for Paleoceanography), presents a thorough analysis of the various components of the modern state: vegetation cover, precipitation, coastal upwelling and the low-level atmospheric state. This multi-dataset analysis employs a generalization of two-field joint analysis using singular value decomposition, heavily used in climate research, to N>2 three- or four-dimensional datasets. This generalization is described in my forthcoming book (Eshel 2007, solicited for publication, based on my web-available notes, by Princeton University Press). A central element of our results is the importance of NAO variability to modulations of productivity off northwestern Africa. In the near future, we will use this dynamical framework to re-interpret the existing records of North African climate during the period 8-6 kyBP.


Reconstruction of Red Sea-Mediterranean Climate in the Last 10,000 Years. Climate fluctuations on decadal-to-millennial timescales have been suggested as underlying such diverse recorded near-eastern archaeological and human-development events as the Akkadian collapse and the development of agriculture during the Younger Dryas, among others. Some (e.g., Jared Diamond in Guns, Germs and Steel) suggest that detailed knowledge of climatic conditions in the Mediterranean basin and the Middle East is central to further progress in understanding early human history. Recent theoretical, observational and methodological developments make the present time ripe for attempting to create a unified, rigorously computed reconstruction of Mediterranean paleoclimate. I plan to construct such a dataset. The work will benefit from tools developed in the course of my research on the paleoclimate information recorded by Red Sea corals (Eshel, Schrag and Farrell 2000), my most recent Red Sea work [Eshel and Heavens (in preparation for Paleoceanography), Eshel 2006 (JAS 63(2))], and also from the strong ties between the Mediterranean and the North Atlantic (Eshel and Farrell 2000, Eshel Cane and Farrell 2000, Eshel 2002). Approximating the state of the Mediterranean at any time benefits discernibly from North Atlantic information (Eshel and Farrell (2000) and Eshel et al. (2000)), but is especially useful over the Holocene because paleo-climate proxy data in the North Atlantic are high-quality and available at relatively high spatial and temporal resolutions. The resultant Mediterranean reconstruction will enhance our understanding of past and present climates of both the Near East and North Atlantic, shedding light on the role of Middle-Eastern climate in archaeology and human development, and also contributing significantly toward preliminary understanding of the expected response of the Mediterranean region to anthropogenic changes. In addition, the results of this linear optimization study will identify geographical regions where data collection efforts will yield the most additional information for a given effort.


Human Behavior and Greenhouse Gas Emissions. While rigorously quantitative, this line of work is motivated by subjective personal views. First, I view global climate change research as clearly demonstrating the need for immediate, substantive action. Though significant remaining uncertainties will continue to cloud climate predictions in the foreseeable future, the physical sciences have provided answers to policy-related questions. The most pressing future issues will be in the realm of human behavior and societal structure, and therefore a major fraction of the burden of climate change research will fall to the social sciences. Second, the political structure of western democracies is less and less amenable to steering societal actions. In examples ranging from, e.g., Italy, Israel and the US, electorates are deeply and bitterly divided, and aspiring elected officials must appeal to the ever more elusive and rare `undecided voter'. As a consequence, little has been achieved politically on the climate change front, and I gravely doubt this situation will change for the better. This state of affairs renders voluntary individual daily-life decisions a central element of any realistic campaign that strives to achieve discernible reductions in greenhouse gas emissions. The line of work described in this section attempts to contribute to bridging the wide gap between the physical and social sciences in pursuit of climate change mitigation. Eshel and Martin (2006) quantified the effect of various personal dietary choices on greenhouse gas footprint. This is clearly just the beginning. There are a number of followup avenues I am now pursuing, or plan to in the near future. With Esther Bowen, a gifted Environmental Studies undergraduate, I am examining the roles corn subsidies play in the US greenhouse gas footprint, and the relationship these roles may have with rising obesity incidence. With Martin, I am making a more complete effort to establish the total contribution of diet to the national greenhouse gas footprint, as opposed to the difference between one set of choices and another, as done in Eshel and Martin (2006). Martin and I have also made considerable progress toward securing the necessary support and partnerships for setting up a watershed-scale, geophysics-based, sustainable development research and teaching Center in a 500-acre organic farm in Millerton, NY. A related followup to Eshel and Martin (2006) that will be facilitated by the Center will be to quantify the energetic and greenhouse gas consequences of organic versus conventional, as well as small- versus industrial-scale, farming. The field course I co-led with Martin in December 2005 went a long way toward realizing this goal, but the research is by no means finished.


I view my most meaningful contributions as: (1) developing a coherent, self-consistent dynamical framework to explain ocean-atmosphere motions and variability in the Red Sea-Eastern Mediterranean region; (2) developing simple methods for geophysical statistical forecasting and applying them to two climate phenomena with great human impact, Middle-Eastern droughts and the North Atlantic Oscillation; (3) writing a geophysical spatio-temporal statistics book which the Reviewers described as ``sorely needed'' and ``technically solid, lucidly written and engaging''; (4) contributing to enhanced dynamical and statistical rigor in paleoclimate work; (5) advancing a rigorous argument quantifying the effects of human dietary choices on greenhouse house emissions; and (6) maintaining extensive civic engagement that is firmly rooted in physical sciences, but which also recognizes the enormous importance of national and international politics to such societal challenges as global climate change.


older research statement